ABSTRACT

Non-communicable diseases (NCDs) are a group of diseases that is the leading cause of mortality worldwide, accounting for 74% of all deaths. Three of the most common NCDs are being the focus of this study Cardiovascular diseases (CVD), diabetes mellitus and chronic respiratory diseases (CRD). These diseases are the leading causes of mortality and morbidity in the world. The burden of these diseases is growing rapidly due to fast urbanization, unhealthy lifestyles, environmental exposures and socio-demographic inequalities, all of which have a major impact on the most precious asset of our society: life regarding premature death, Disability-Adjusted Life Years (DALYs) and health system burden.  (GBD) data was analysed, and the future disease burden was estimated from 2022 to 2050 using Auto Regressive Integrated Moving Average (ARIMA) model. The data-driven clustering, temporal trend analysis, geospatial mapping and socio-demographic (SDI) thresholds analysis following Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER) evaluation perspectives, respectively, using the data of selected NCDs in this study. Mortality and DALYs due to diabetes, Cardiovascular diseases and chronic respiratory diseases were analysed at global, regional and country levels between 1990 and 2021. The findings revealed significant regional and country-specific differences in the burden of disease. Age-standardised mortality and DALYs rates tended to decrease for CVDs and CRDs, while age-standardised DALYs rates for diabetes were rising and are expected to be significant public health issues well into the future. Data-driven clustering identified clusters of countries with similar disease burden profiles which were not consistent with the traditional GBD country regions and geospatial analysis identified distinctive hotspot and cold spot regions. Furthermore, the Decision Tree model was applied to find the empirical SDI threshold for increased disease burden, thus establishing an intuitive framework to identify high-risk populations. Overall, this study provides a mathematically sound analytical framework for appreciating the patterns of global NCDs burden and valuable insights for policymakers and healthcare planners to design targeted interventions and optimize resource allocation to reduce premature deaths and disabilities due to NCDs.

Keywords: Non-communicable diseases, , DALYs, Mortality, Data-Driven Clustering, Socio-demographic Index.

                                                            CHAPTER-01

INTRODUCTION

1.1 Introduction

A communicable disease is any illness transmitted from one person to another. People sometimes refer to communicable diseases as “infectious” or “transmissible” diseases((WHO) W. H., 2023; Bloom & Cadarette, 2019).While on the other hand noncommunicable disease (NCDs) develop slowly over years and are not caused by an infection that cannot be transmitted from person to person. Lifestyle, environmental and genetic factors have a significant impact on the NCDs. (GBD Risk Factors Collaborators 2024; World Health Organization [WHO], 2023). Although communicable diseases have been the primary cause of death in the world for centuries, the number of NCDs has increased and now claims 74% of deaths around the world. (WHO, 2021; GBD, 2020). The main three NCDs are Cardiovascular disease, chronic respiratory disease and diabetes mellitus, accounting for the vast majority of deaths globally annually. (WHO, 2022; Vos et al., 2020).

The NCDs are the major causes of premature mortality and morbidity globally, and they are generally prevalent in low- and middle-income countries (LMICs) where the health facility is unable to manage the dual burden of infectious diseases and NCDs disorders (WHO 2021; Nugent 2018). Current lifestyle factors are highly correlated with NCD prevalence, such as low-quality diet, low physical activity, smoking and excessive alcohol consumption. In addition to strenuous physical activity, harmful alcohol use, tobacco use, and high blood pressure, unhealthy diet, physical inactivity, high body mass index and high fasting plasma glucose are all major risk factors, and they contribute significantly to the global burden of NCDs (GBD Risk Factors Collaborators, 2024; World Health Organization [WHO], 2023). This includes disease preventative measures, early diagnosis and treatment measures. These conditions, also known as “lifestyle diseases”, are mostly preventable but continue to rise due to poor public measures and a lack of access to medical services. The level of these diseases is greatly different between countries and regions, reflecting differing levels of socioeconomic development, health services, demographics, and exposure to risks. To gain a deeper understanding of patterns and drivers of disease and to inform policy decisions in the public health arena, estimates need to be disaggregated to the global, regional, and country level (GBD 2021 Disease and Injury Collaborators, 2024; GBD Risk Factors Collaborators, 2024).

Table 1.1 Deaths from Top 3 NCDs (2018-2022)

Data Source: GBD (2018-2022) Deaths,  (GBD).

1.1.1 CardiovascularDisease (CVDs)

Cardiovascular disorders are a group of diseases affecting the cardiovascular system, which consists of the heart and blood vessels (Roth et al., 2020). These diseases include coronary artery disease, heart attacks, stroke, hypertension and heart failure. The primary factors associated with CVDs are high blood pressure, smoking, overweight/obesity, diabetes and an unhealthy diet, in which narrowed or blocked blood vessels can result in severe complications such as heart attack or stroke. (Yusuf et al., 2020).

1.1.2 Diabetes Mellitus (DM)

Diabetes is a long-term metabolic condition that occurs when the blood sugar (glucose) level is high because the body has little or no insulin (American Diabetes Association, 2022). It is broken down into two main types: Type 1 diabetes, in which the immune system attacks insulin-producing cells, and Type 2 diabetes, caused by insulin resistance and lifestyle, including diet and physical inactivity. (WHO, 2022; Zheng et al., 2018). Diabetes can cause complications such as kidney disease, nerve damage and Cardiovascular disease if it is not controlled (Forouhi & Wareham, 2019).

1.1.3 Chronic Respiratory Disease (CRDs)

Chronic respiratory diseases involve a problem with the lungs and airways that can make it harder to breathe (Soriano et al., 2020). Common CRDs include chronic obstructive pulmonary disease (COPD), asthma, pulmonary fibrosis and occupational lung diseases (WHO, 2021). Air pollution and smoking are the most important risk factors. Respiratory infections and genetic susceptibility. CRDs are typically permanent in nature and, to enhance lung function and quality of life, they must be managed for a long time. (Vogelmeier et al., 2017).

1.2  (GBD)

The  (GBD) Study is a comprehensive, global epidemiological research initiative led by the Institute for Health Metrics and Evaluation (IHME). The study systematically estimates the burden of diseases, injuries, and risk factors in countries and regions at an age and sex level using standardized methods. The GBD study framework enables consistent estimation of disease prevalence, mortality, DALYs, YLLs and YLDs which can be used to measure the burden of disease in a consistent way over time and across populations. Comprehensive (GBD 2021 Diseases and Injuries Collaborators, 2024), with a standardized format and methodology that allows for the measurement of the burden of disease and injuries, the GBD database is a valuable tool for monitoring global health trends and supporting evidence-based policy and interventions (IHME, 2024).

1.2 Background of the study

NCDs have become a public health challenge across the globe, responsible for 74% of deaths occurring worldwide. ((WHO), (2021)). Cardiovascular disease is the leading NCD killer, accounting for approximately 17.9 million deaths annually, followed by cancers (9.3 million), respiratory diseases (4.1 million) and diabetes (1.5 million). Aside from the mortality rates of these diseases, millions of humans are disabled and experience poor quality of life. According to GBD 2023, there have been significant rises in the burden of NCDs, obesity, mental illness, and environmental risks leading to DALYs, especially in developing nations (Bhutta, 2025). Understanding the geospatial distribution of these diseases helps guide in creating effective public health policies. Clustering techniques are exploratory geospatial mapping tools that can identify regional clusters of health status indicators that can be used to target specific populations for interventions. (Nugent, 2018). These include direct health care costs such as hospitalizations and medicine, as well as indirect costs such as loss of productivity and reduced participation in the labour force. (Nugent, 2018). There are potentially two types of risk factors associated with NCDs: modifiable and non-modifiable. Modifiable factors are smoking, unhealthy diets, physical inactivity, and excessive alcohol use, while non-modifiable factors are age, genetics, and family history (Organization W. H., 2021).

NCDs are one of the most serious public health problems in the contemporary world. In most cases these are non-infectious, chronic diseases, which have known long-term development and complicated interactions between behavioural, environmental, genetic and socio-economic factors. The major categories of NCDs are Cardiovascular diseases, cancers, chronic respiratory diseases, and diabetes, and all these types of diseases contribute to the major portion of morbidity and mortality in the world (Wagner and Brath, 2011). There have been some significant changes in the epidemiology of the health conditions in the world in the last few decades. The reduction of mortality caused by communicable diseases has been recorded in most regions due to the development and improvement of infectious disease control, vaccination, and sanitation. Nevertheless, there are similar tendencies in the world including the aging of populations, high rates of urban development, lifestyle disorders, poor nutritional habits, physical inactivity, tobacco consumption, and environmental pollution that have increased the prevalence of NCDs. Thus, a double burden of disease is currently experienced in many countries, especially in the low- and middle-income countries, with non-communicable and communicable diseases coexisting and exerting growing pressure on the health care system (Taheri Soodejani, 2020; Li et al., 2022). Global systematic evaluations undertaken within the  (GBD) framework continue to reveal that NCDs are the greatest source of deaths, disability-adjusted life years (DALY), as well as years of life lost (YLL). The findings illustrate the pressing necessity to tackle the data-driven methods of comprehending the changing burden of NCDs over time, geography, and socio-demographic settings (GBD 2017 Causes of Death Collaborators) shown in Figure 1.1.

Figure 1.1: Leading causes of NCDs mortality and associated risk factors ((WHO), (2021))

Source:  Adapted from World Health Organization (WHO), Noncommunicable Disease Fact Sheet, 2021.

1.3 Problem Statement

NCDs, including CVDs, DM, and CRDs are the major causes of global mortality and disability (GBD, 2020). These diseases account for 74% of deaths and are more prevalent in low-and-middle-income countries (WHO, 2021). NCDs pose great challenges to the economic, social, and institutional development of people, mostly those of LMICs. With the growing disparity between the rich and the poor, the world faces many challenges in fulfilling the SDGs, particularly SDG 3 (promoting good health and well-being for all ages). NCDs have a huge influence on the financial aspects of individuals as well as on the countries. The GBD studies generally provide a regional disease mapping but are least focused on identifying cross-regional similarities in disease risk and associated mortalities. Even though countries in different regions may be facing the same threats, they have not been categorized on the basis of data-trends. This is where the need for filling the gap arises through decision-making using data.

Many countries of the world could not address this surge of NCDs due to social, economic, and system-related constraints. This situation subsequently leads to multiplicative effects of NCDs on premature mortality, a severe socio-economic burden (WHO, 2021). Thus, the health systems of LMICs have collapsed due to limited resources. Therefore, the disparity between LMICs and other countries is widening day by day. The World Health Organization (WHO) and the United Nations (UN) have to reallocate the resources to deal with this disparity but without the involvement of the affected countries, these issues cannot be resolved. In this scenario, suitable evidence-based strategies are needed that can help to reduce this burden on LMICs and other countries as well.  This study tries to help in this matter by providing such analyses that will pinpoint the riskiest countries that can be combined in a cluster due to similarities regarding NCDs-related prevalence and mortality (N Gibbs, 2020).

This research examines to address this gap by employing advanced clustering and geospatial mapping techniques to identify cross-regional similarities in NCDs risk factors and their associated mortalities and disabilities. This study will estimate global, regional, and country-wise trends for mortality and Disability-Adjusted Life Years (DALYs) of NCDs. Understanding the distribution and patterns of risk factors of NCDs, mortalities and disabilities will be required to plan actionable strategies. Furthermore, through this research, the effect that socio-economic factors have on the prevalence of NCDs will be brought to light. Based on the prevalence of NCDs and associated risk factors, this study will provide recommendations. The use of clustering and geospatial mapping techniques provides a novel approach to NCDs, allowing the stakeholders to detect hot and cold spots and similarities in cross-regions. Many countries in the world were unable to tackle this issue of NCDs owing to certain limitations within the realm of social, economic, and systems-related issues. The above-mentioned scenario then leads to the phenomenon of multiplicative effects of NCDs regarding early mortality and socio-economic impacts (WHO, 2021). Furthermore, according to new GBD findings, it has been suggested that traditional SDI classification models might lack the ability to explain certain contextual factors such as environmental risks, climate change, conflicts, and health care system functioning, among others (Bhutta, 2025).

1.4 Research Objectives

1. To estimate global, regional, and country-wise trends for mortality and DALYs of NCDs and their risk factors.
2. To develop data-driven clusters based on mortality and DALYs of NCDs and their risk factors, and to compare them with GBD regions.
3. To compare regional estimates of mortality, and DALYs of non-communicable diseases and their risk factors with data-driven cluster estimates.
4. To explore hot and cold spots of non-communicable diseases using geospatial mapping techniques and data-driven clusters.
5. To drive a new data-driven threshold for Socio-demographic Index (SDI) for better decision-making.

1.5 Research Scope

This study investigated the burden of three main NCDs CVDs, DM, and CRDs based on the available data from the GBD study 2021. The research spans a range of geographical levels, from the global, through GBD super-regions, to sub-regions and 204 countries, capturing disease burden from various populations. To ensure the comparability of rates across countries and regions with different population age structures, the study uses age-standardised rates. The analysis comes primarily from two indicators of burden, deaths and Disability-Adjusted Life Years (DALYs), which together provide a picture of the mortality and overall health effects of the selected diseases. Temporal trend analysis of burden of disease was done using GBD estimates from 1990 to 2021 to investigate historical trends. Additionally, the Auto Regressive Integrated Moving Average (ARIMA) model was used to make future predictions regarding disease burden trends from 2022 to 2050, giving insights on likely future trends of these diseases.

Besides temporal analysis, the study uses the K-means clustering algorithm to create country clusters by looking at similarity of disease burden for different countries. The clusters are then compared with the current GBD regional categorisation to assess if countries within a cluster are consistent with traditional geographical regions. In addition, hotspot analysis is carried out to identify regions with a particularly high and low disease burden and Socio-demographic Index (SDI) analysis is performed to examine the relationship between disease burden and socio-economic development by countries and regions. The study follows the Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER) to ensure transparency, reproducibly, and full reporting of health estimates from the GBD database. By following the GATHER recommendations, the reliability and scientific rigour of the study are increased through greater clarity in data sources, analytical methods and reporting.

1.6 Summary of Chapter

In this chapter, the burden of NCDs globally, and specifically on CVDs, DM and CRDs was presented. It explained their significant risk factors, significance of the GBD Study and the problem, objectives, scope and significance of the research. The literature review in the next chapter describes the current knowledge regarding NCDs, disease burden indicators, risk factors, and previous work conducted on trend analysis, disease clustering, hotspot analysis, SDI, and disease forecasting. It also recognizes research gaps which justify the present study.

CHAPTER-02

REVIEW OF LITERATURE

2.1 Introduction

This chapter discussed literature on  (GBD) and non-communicable diseases (NCDs) emphasizes that the global burden of chronic diseases is rising and that the need for standard health assessment tools is rising. Prior research has extensively relied on GBD databases in analyzing the trends in death, disability, and risk factors in different countries and regions. This chapter focuses on a literature review on topics associated with the GBD approach, NCDs, regional burden trends, and approaches used in analyzing disease burden, particularly cluster analysis and geo-spatial approaches.

2.2  (GBD)

The GBD study was initiated in the year 1990. It was formulated as an extensive and comprehensive way of measuring diseases, injuries, and risk factors globally. The GBD study is now being coordinated by the Institute for Health Metrics and Evaluation (IHME).The GBD framework is one of the most commonly used tools for global health research, offering consistent estimates of health across countries, age groups and time. The GBD study offers systematic estimations of the impact of many diseases, injuries, and risk factors based on standardized methods allowing comparison between countries and across time periods. The initiative has transformed into an extensive global health surveillance system operated mainly by the IHME. New GBD studies offer estimations for hundreds of diseases and risk factors in over 200 countries and territories globally, constituting one of the most significant health databases for research on population health (Vos et al., 2020; Murray et al., 2020).

A number of studies have employed data from the GBD database in the examination of trends and patterns of prevalence, incidence, mortality and risk factors for NCDs, communicable diseases and injuries on a global or regional level. Researchers have applied GBD data in analyzing trends, burden, geospatial distribution, clustering and predictions to enhance our knowledge of health inequalities and help identify sub-populations at risk (Vos et al., 2020). The existence of Cardiovascular diseases estimates within the GBD database allows comparing populations with different age structures and conducting temporal analysis.Nevertheless, even despite the great contribution to public health practice, the GBD study has been criticized due to certain weaknesses such as modelling, disability weighting and assumptions in DALYs calculations. There has been a debate on how imputation and estimation in the context of scarce data in some regions can influence the accuracy of burden estimates. In addition, arguments have been made against certain assumptions in the evaluation of disease severity by means of disability weighting (Anand & Hanson, 1998; Knutsson & Munthe, 2020).

Figure 2.1: Evolution of  studies

The GBD Study has developed from its first assessment in 1990 to the current update GBD 2021 as shown in figure 2.1. The geographical scope of the study has broadened, along with the number of diseases, risk factors and methodology over the years, to become one of the most extensive health surveillance systems in the world.

2.3 Non-Communicable Diseases (NCDs)

NCDs may be defined as those medical problems which usually take a lot of time to develop. NCDs are not communicable diseases since they develop due to genetic makeup, the environment, physical activity, and behaviour (World Health Organization [WHO], 2023). These NCDs are now considered a very important public health problem due to their increasing numbers in cases, deaths, complications, and socio-economic effect.The four major NCDs are heart disease, cancer, respiratory disease, and diabetes mellitus. According to the information provided by the World Health Organization, these are the leading causes of the most deaths related to NCDs globally every year (WHO, 2023). Heart disease is the leading cause of death worldwide, of all these NCDs.Respiratory disease and diabetes are the third and fourth causes of death after cancer.

The prevalence of NCDs has increased considerably in recent decades globally due to many reasons including demographic changes, urbanization, population aging, sedentary behaviour, poor nutrition, hazardous alcohol use and environmental factors (GBD 2019 Diseases and Injuries Collaborators, 2020). Few healthcare professionals, late diagnosis, lack of investments in preventive measures and disparities in access to health care services are some of the factors that contribute to the unusually high burden of NCDs in low- and middle-income countries (LMICs). Several studies show that more than three-quarters of all deaths associated with NCDs worldwide occur in LMICs (WHO, 2023).There are several modifiable behavioural and metabolic risk factors that are common forNCDs. Behavioural risk factors include smoking, unhealthy diet, lack of exercise, and excessive drinking of alcohol. These types of behaviour are linked to disorders such as high blood pressure, obesity, high blood sugar, and high cholesterol, all of which are NCD risk factors (Murray et al., 2020).The role of environmental and occupational exposures in the ethology of chronic respiratory conditions and some cancers also is significant.

Now the increasing burden of NCDs is a matter of great concern to the global health organizations and policy makers because of their contribution to death, and significant contribution to disability and loss of productivity. In the  (GBD) study, the Disability-Adjusted Life Year (DALY) framework has been used to show that a large proportion of the total burden of DALYs are due to NCDs both regarding premature deaths and years lived with disability (YLD) (Vos et al., 2020). This burden is growing and is placing a strain on the healthcare system, particularly in low resource settings.

The growing global burden of NCDs required that NCD prevention and control should be placed under Sustainable Development Goal 3 (SDG-3) that deals with healthy lives and promoting well-being for all ages. One of the major SDG-3 targets is to achieve by 2030; “Ensure healthy lives and promote well-being for all at all ages, by strengthening the prevention and treatment of NCDs.” Therefore, knowing the temporal and geographical variation of NCD burden is crucial for planning of health care services, development of policies and allocation of resources. There are a number of researchers who have used GBD data to explore trends and spatial patterns of NCDs in countries and regions. Clustering has been applied to find high-burden populations, geospatial mapping and trend analysis have been applied to understand the inequities in disease distribution (Mensah et al., 2018). Most of the earlier studies, however, adopted the traditional geographic classification and did not employ the region grouping approaches based on data, so, there is need for further studies using the clustering-based approaches for region grouping.

2.3.1 Top Ten Non-Communicable Diseases in the World

According to the Global Burden of Disease study 2021, the majority of disease and mortality in the world are said to be caused by NCDs. The ten top categories of NCDs are illustrated in Table 2.1 as responsible for a considerable burden of the world’s health burden (Vos et al., 2024; WHO, 2023).

Table 2. 1:  Top Ten NCDs Globally

Source:Information based on GBD 2021 & WHO reports (Vos et al., 2024; WHO, 2023).

The impact of NCDs is usually described regarding the ‘big 4’ non-communicable chronic diseases as defined by the World Health Organization. The GBD Study 2021 reported that Cardiovascular diseases were the leading cause of death and burden of disease worldwide, followed by neoplasms, chronic respiratory diseases, and diabetes mellitus. Other important categories of NCDs that cause disability, reduced quality of life and health costs include chronic kidney diseases, mental and neurologic diseases, digestive diseases, musculoskeletal diseases and substance use disorders (Vos et al., 2024).These include Cardiovascular diseases and diabetes mellitus, which are becoming very prevalent and are strongly associated with modifiable behavioral and metabolic risk factors such as poor diets, lack of physical activity, obesity, high blood pressure and high sugar levels. These illnesses also play an important role in DALYs, YLDs and mortality on a global level.NCDs have created a lot of burden for the health systems, particularly in developing countries, due to limited access to prevention and/or treatment services (WHO, 2023). Cardiovascular diseases, diabetes mellitus and Chronic respiratory disease were chosen as the main focus of the present study, due to their significant role in the disease burden in the world and their similarities in risk factor profile.

2.4 GBD Regions and Super Regions

According to the  (GBD) Study, countries are grouped hierarchically by geographical areas, which include super regions and regions. The classification streamlines the comparison of disease burden, mortality, disability and risk factors between geographically and epidemiologically comparable populations. In GBD 2021, countries and territories are grouped into seven super regions and twenty-one regions based on geographical proximity and health-related characteristics (GBD 2021 Collaborators, 2024) are shown in Table 2.2.

Table 2. 2: Super and Sub-regions from GBD

Source: Adapted from GBD 2021 Collaborators (2024).

The GBD framework categorises countries into seven super regions and twenty-one regions for ease of comparison to disease burden between regions and countries (Tables 2.2). This hierarchical structure allows scientists to study regional differences in deaths, disability-adjusted life years lost (DALYs), and exposure to risk factors, and subsequently, to compare regions with similar geographical and epidemiological characteristics. The current study will build on the existing regional classification of the GBD and examine the potential of using alternative data-driven classifications via clustering to categorize countries based on their burden profile for Cardiovascular disease and diabetes, irrespective of geographical region.

The countries are classified in a number of hierarchical levels under GBD, such as global level, super-regional level, regional level, national level. The super regions are geographically and epidemiologically similar regions, and countries within individual regions have relatively similar health and development profiles. Some examples of GBD super regions are Sub-Saharan Africa, South Asia, Southeast Asia and Oceania, North Africa and the Middle East, High-income regions and Latin America and the Caribbean. These classifications allow disease burden to be analysed in           a more structured way over large geographical regions (Vos et al., 2020).

The GBD regional classification system is commonly employed in epidemiological studies due to its capacity to offer a consistent and uniform approach to comparing disease burden measures between countries over time. Researchers have relied on these regional clusters to estimate the variability of mortality, prevalence, DALYs, YLLs and YLDs for communicable diseases, NCDs and injuries (GBD 2019 Diseases and Injuries Collaborators, 2020). Regional classification also can help locate the areas with a disproportionate disease burden, as well as to guide policy development and healthcare planning.

A number of studies have found significant regional variation in disease burden in GBD regions. In accordance with the quality of the health care system, early diagnosis, and availability of the preventive strategies, the mortality rate of the communicable diseases is quite high in the low-income area as compared to the high-income area. Nevertheless, the mortality rate of both communicable and NCDs is higher in low and middle-income areas due to poor health care system, socio-economic inequality, environmental issues, and absence of disease prevention (Murray et al., 2020).

There is also significant regional variation in the distribution of NCDs based on differences in demographic transition, urbanisation, lifestyle behaviours and socioeconomic development across the GBD regions. For instance, the prevalence of Cardiovascular diseases in rapidly urbanizing areas has risen, and chronic respiratory diseases are associated with environmental and occupational exposures in several LMICs (WHO, 2023). Geographic classification systems are important to consider in understanding disease burden patterns because there are geographic variations.

2.5 NCDs Trends in GBD Regions

NCDs are now the leading cause of mortality and morbidity worldwide. NCDs represent a substantial proportion of the total global disease burden, constituting a considerable percentage across all geographic regions. In recent decades, the prevalence of NCDs has increased more than ever before owing to various factors such as population growth, aging, urbanization, and the epidemiological and lifestyle transitions (GBD 2019 Diseases and Injuries Collaborators, 2020). The GBD project has provided a wealth of data on trends in the burden of NCDs over time and across geographic regions, revealing disparities among nations and geographic regions. However, several studies based on the GBD database revealed that the burden of the NCDs categories has increased. Although there has been an increasing burden of NCDs and their disabilities in many LMICs, there have been developments in health care systems and technology which have helped to reduce the mortality rate in high-income regions (Murray et al., 2020). The rising burden will pose significant challenges to health care systems and economics across the globe.

Cardiovascular diseases are the foremost reason for NCD morbidity and DALYs in the world. As per previous literature, Cardiovascular disease has had a high burden in some regions of the world such as Eastern Europe, Central Asia and some parts of Sub-Saharan Africa due to risk factors like smoking, hypertension, poor diet and obesity (Mensah et al., 2018). Several high-income regions, however, have noted a decline in mortality rates owing to the improved preventive measures, early diagnosis and treatment.Likewise, diabetes has been shown to have a steady rise in prevalence in many GBD regions, especially in South Asia, the Middle East and North Africa. The major factors contributing to the increasing diabetes burden in these regions are urbanisation, sedentary lifestyle, obesity and dietary changes (Saeedi et al., 2019). There have also been significant regional variations in diabetes-related mortality and DALYs, with low and middle-income countries facing poor disease management and access to health care.

Similarly, chronic respiratory diseases have also caused morbidity and mortality globally, particularly in regions that experience higher levels of environmental and occupational exposures. Chronic respiratory diseases are believed to be greatly affected by smoking, air pollution, biomass fuel exposure, and occupational exposures in some LMICs (GBD Chronic Respiratory Disease Collaborators, 2020). Differences in environmental exposures and health care systems by region have a large impact on disease outcomes in different populations. The GBD framework has further demonstrated substantial inequalities in NCD burden across super regions and countries. High-income regions generally show lower mortality rates and improved disease management, while developing regions continue to experience increasing disease burden because of limited healthcare resources and growing exposure to behavioral and metabolic risk factors (Vos et al., 2020). These differences suggest a geographic variation in the rate of epidemiological transition.

A temporal trend analysis conducted with GBD data has revealed that while Cardiovascular diseases mortality rates have declined in some regions for some NCDs, the number of cases, deaths and DALYs attributable to these NCDs has continued to rise worldwide as a result of population growth and aging (Murray et al., 2020). This trend implies that the healthcare system could have more chronic disease management issues in the future.Several researchers have adopted more sophisticated methods like clustering analysis, trend modeling and Geospatial mapping to analyse the pattern of NCDs at the level of GBD regions. High-burden clusters and geographic hot spots have been identified in these studies that are not in line with traditional regional classifications (Roth et al., 2020). These results highlight the need for data-driven methods to gain insight into disease distribution and to identify vulnerable populations.

2.6 Research Gaps in Existing Literature

Global health research on NCDs, including Cardiovascular diseases, chronic respiratory diseases, and diabetes mellitus, is being increasingly pursued to tackle the burden of disease problem posed by these NCDs. While some studies have focused on the estimation of burden of disease using GBD approach, there are important areas that still need to be explored from a methodological and analytical perspective.Most of the studies of NCD burden have focused on geographic regions and super regions, which are predetermined entities (GBD 2021 Diseases and Injuries Collaborators, 2024; Vos et al., 2020). These categorisations help region-to-region comparisons, but there can be significant variation in disease burden, health care provision, and risk-factor burden within regions between different countries. Thus, the traditional geographical classification may not reflect the true similarities in some countries having similar epidemiological features. There are several studies reporting changes in mortality or DALYs of individual diseases over time, but relatively few studies have used clustering techniques to identify groupings of countries that share similar disease burden and risk factors (Roth et al., 2020; Murray et al., 2020).

 Previous analyses are mostly carried out on geographical regions instead of burden-based clusters, which means that countries that have similar epidemiological patterns but are geographically far away are not identified. Most studies have focused on Cardiovascular diseases, diabetes mellitus, and chronic respiratory diseases, individually. Likewise, risk-factor analyses are typically made without taking into account DALY and mortality estimates (GBD Risk Factors Collaborators, 2024). Very limited evidence exists that incorporates more than one NCD, mortality, DALYs and related risk factors within a single framework. Such an integrated approach could provide a more comprehensive understanding of disease burden patterns. While a few studies have done geo-referencing of disease burden, very few have used spatial statistical methods to determine significant hotspots and cold spots of NCD burden (Moran et al., 2017; Roth et al., 2020). Targeted areas of geographic risk are therefore poorly understood, and in combination with data-driven cluster classifications, are left under-explored. To measure socioeconomic development in relation to health outcomes, the Socio-demographic Index (SDI) is widely used within GBD studies. The categories of SDI, however, are fixed and may not capture the full spectrum of changes in disease burden over time and across countries in recent years (GBD 2021 Collaborators, 2024). However, few studies have tried to set up thresholds based on SDI data that reflect the actual patterns of NCD burden, which opens up opportunities for more evidence-based classifications.

2.6 Summary of Chapter

In this chapter literature on NCDs, their main risk factors and patterns of burden at the global level were reviewed. The review identified the rising burden of Cardiovascular diseases, diabetes mellitus and chronic respiratory diseases, in each of the GBD regions and drew attention to the impact of socio-demographic, behavioural, metabolic and environmental factors. Most current studies were based on traditional geographical classification and generally did not have an integrated data-driven approach to the identification of burden patterns, hotspots and socio-demographic variation. The gaps serve as the basis of the present study. The research methodology is described in the next chapter, including the data sources, study variables, analytical framework, clustering techniques, geospatial analysis, and statistical analysis to reach the study objectives.

CHAPTER-03

METHODOLOGY

3.1 Introduction

The research methodology used in this paper to analyse burden of non-communicable diseases (NCDs) and associated risk factors, in this study, is presented. In this paper, the research methodology adopted to analyze the burden of NCDs and associated risk factors is presented using the  (GBD) 2021 database. The methodology was systematically designed to meet the research goals that involved data preprocessing, trend analysis, K-means clustering, comparison with GBD regional classifications, hotspot analysis, Socio-demographic Index (SDI) analysis and forecasting for the future using Auto Regressive Integrated Moving Average (ARIMA) model. The analytical methods used included the identification of temporal trends, geographical patterns, and similarities in disease burden on the country level, regional level and global level. The study adheres to the Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER) to make health estimates from the GBD database transparent, reproducible, and comprehensive. Following the GATHER guidelines helps to make the study more reliable by ensuring that the sources of the data are clearly presented, that the data is preprocessed, the analytical methods are clearly presented and that the reporting of the data is clearly set out. To conduct the research in this study, the whole process of research is presented in Figure 3.1 as follows: Furthermore, in addition, it is worth noting that the proposed research methodology is aligned with the GATHER reporting guidelines as presented in Figure 3.2, showing that each step of the research meets the recommended principles for transparent health estimates reporting.

Figure 3.1: Methodology Diagram

Figure 3.2: Methodology workflow of the study and Gather Reporting requirements

Each methodological step is described in detail in the following sections, which include data collection, data preprocessing, exploratory data analysis, trend analysis, clustering, geospatial hotspot analysis, SDI analysis and forecasting procedures.

3.2 Source of Data

The databases utilized in this study are part of the GBD database compiled by the Institute for Health Metrics and Evaluation (IHME). The GBD database contains comprehensive and standardized estimates of diseases, injuries, and risk factors thereof on a global, regional, and country scale.The GBD database was selected for its reliability in providing accurate and comparable estimates for the disease burden indicators over various periods. The dataset contains details on mortality and DALYs, and their risk factors. Secondary data collected from the official IHME database were used in this research. The collected data were from several countries around the world.

3.3 Description of Dataset

It describes the structure, coverage, and features of the dataset that was used in this research. This was a dataset that had been extracted from the GBD database and includes standardized estimates related to NCDs and their associated risk factors. The design of the dataset facilitated comparison of disease burden indicators across the world and other levels. In this analysis age-adjusted rates were used to make a comparison between nations with different demographic compositions. Age standardization reduces demographic differences and makes a comparison easier. The data sets consist of complete information on disease burden indicators as well as the risk factors associated with NCDs. They include epidemiological parameters like mortality and DALYs, behavioral, metabolic and environmental risk factors and SDI indicators.

3.3.1: Summary of Datasets

Several data sets were involved in the study on various kinds of NCDs. The data sets used in the research are summarized in Table 3.1.

Table 3. 1: Dataset used in Research

The selected dataset has been individually obtained and processed for disease specific study. The datasets contain standard burden measures along with their risk factors under the GBD system. The data sets have been subsequently combined, filtered, and arranged based on the requirements of the study.

3.3.2 Disease Burden Indicators

The analysis aims to examine the key disease burden indicators that are often considered in epidemiological and public health studies in relation to the effect of NCDs. The selected disease burden indicators cover all relevant aspects related to mortality, disability and health burden. The description of disease burden indicators are shown in detail in Table 3.2.

Table 3. 2:Description of Disease Burden Indicators

3.3.3 Risk Factors Variables

The datasets consist of numerous risk factors for developing NCDs, including both behavioral, metabolic, and environmental ones. They were classified according to the classification system for GBD risk factors. The main risk factors analyzed within the study. The classification of the risk factors are shown in Figure 3.2.

Figure 3.3: GBD Risk Factor Framework

This was done to establish the role played by the risk factors in disease burden trends and similarities between countries.

3.3.4 Description of Variables

The data sets that have been extracted include demographics, geography, time, disease variables, and statistical factors necessary for pre-processing, clustering, trend analysis and geospatial visualization. The key variables extracted are listed in Table 3.3.

Table 3. 3:Description of variables

The selected variables were filtered and arranged based on the study goals. The same variables were then used in the process of pre-processing, normalizing, clustering, analyzing trends, and geospatial visualization.

3.4 Data Preprocessing

The datasets from the GBD were preexisting and standardized. Therefore, there was no need for any extensive data cleaning. However, various data preprocessing procedures such as data filtering, variable selection, country name standardization, datasets merging, and normalization were carried out before performing cluster analysis and trend analysis.

3.4.1 Normalization and Standardization

Measures of continuous variables were analyzed such as mortality rates, DALYs, and risk factors measures.  In cases where this was important, variables are standardized or normalized so that they are comparable across regions and countries. This move helped to do correct clustering based on distance and ensure that variables with bigger scales did not dominate the analysis. For transformation, the z-score standardization method was used, shown in Equation (3.1).

(3.1)

​ Where, x represents the Actual value of the variable,  represents the Mean of the variable and  represents the Standard Deviation.

3.5 Data-Driven Clustering

The clustering method was applied to discover countries that share common characteristics regarding the burden of NCDs and their risk factors. Clustering methods allowed countries to be clustered according to common characteristics related to disease burden measure, including mortality and DALYs. A data-driven clustering methodology was used to discover hidden regional patterns and compare them with current GBD regional groupings. However, traditional geographical classification systems might not completely capture the similarity in patterns of disease burden across nations. Therefore, clustering techniques were used to find groups that reflect epidemiological similarities rather than only geography. The cluster analysis was done separately on each NCDs dataset such as diabetes mellitus, CVD and CRD. 

The main aim of clustering was to find countries that share common attributes regarding their disease burden profile. Clustering helps to discover patterns in our multivariate dataset and conduct comparisons between various countries and regions. Some specific aims for the current study application of clustering analysis are:

• Identifying countries that have similar disease burden profiles.
• Comparing the data based regional groupings with the currently defined GBD regions.

The clustering framework is more flexible and a good approach as compared with traditional geographic classification.

3.5.1 Clustering Procedure

Clustering analysis was performed using the indicators of major disease burden as contained in the GBD database. The selection of variables is based on the epidemiological significance of the variable as well as the objective of the research. In order to reduce the effect of age distribution among the countries, Cardiovascular diseases rates were chosen. The most recent year of information available is considered to represent the current burden of the country. The variables chosen for clustering are Deaths Rate (represents mortality burden) and DALYs rate (represents overall disease burden). These variables show multiple dimensions of disease burden including mortality, disability and overall health loss. These selected variables were used to cluster the countries into groups using the K-means classification approach. The clustering procedure, with K-means algorithm and cluster visualization is described in the following sections.

3.5.2 K-Means Clustering Algorithm

The K-means Clustering Algorithm was employed as the main clustering method for this research work. K-means clustering algorithm is a well-known unsupervised machine learning algorithm that is often applied to group observations with similarities based on their features. K-means clustering algorithm clusters countries into K partitions by minimizing within cluster variance and maximizing between cluster variance. Countries clustered in the same cluster possess similar disease burden characteristics. 

The K-means algorithm consists of the following steps:

1. Choosing the number of clusters (K)
2. Random initialization of cluster centroids
3. Assign data points to the closest centroid.
4. Updating the centroids based on their respective assignments.
5. Repeating the entire procedure until reaching convergence.

The objective function of K-means is given in equation (3.2) as follow:

                                           (3.2)

Where:

The total within-cluster sum of squared errors  is minimized in the objective function, and this minimizes the size of the clusters and makes them as homogeneous as possible. The parameter  is the number of clusters, and the is the  cluster. The variable   stands for one of the observations allocated to the cluster  while ​ is the centroid(mean) of the cluster The expression  is the squared Euclidean distance of each observation  from its cluster centroid . The shorter this distance the closer the members of a cluster are to one another and the farther away the clusters are from each other the better.The optimal number of clusters was established by the Elbow method. The within-cluster sum of squares was computed for various values of K, and the optimal number of clusters was found where a decrease in WCSS started becoming slow and formed an elbow. According to the findings from the Elbow method, the countries have been clustered into four clusters.

3.5.3 Cluster Visualization

The cluster visualization process utilized scatter plots to highlight similarities between countries based on their DALYs rate standardization. Convex hull borders were drawn to distinguish between clusters. Various colours have been used for each cluster, enabling us to distinguish between high-burden and low-burden country clusters.

3.6 Trend Analysis

Trend analysis was performed to evaluate temporal differences in NCD burden globally, regionally, and nationally based on Cardiovascular diseases estimates generated from the GBD database that are shown in table 3.4. Trend analysis was aimed at understanding temporal changes in disease burden metrics as well as comparison of burden profiles between different geographic areas and high-burden nations. Trends analysis was performed separately for different groups of diseases such as diabetes, Cardiovascular diseases, and chronic respiratory diseases.

Table 3. 4: Filtering Criteria used for Trend Analysis

The filtered data sets were sorted by time to observe the evolution of the disease burden indicator metrics. Year-wise consolidation was done to show trends at the global, regional, and country levels.

3.6.1 Global Trend Analysis

A global trend analysis was done to determine the temporal changes in the burden indicators over the study period. Aggregation on a yearly basis was done by calculating the average burden rates per year. Graphical representations of the Cardiovascular diseases burden rate changes were plotted using line graphs.

The global trend analysis facilitated:

• Determining the burden trends over time.
• Detecting areas with increased burden rates.
• Determining the burden trends over time.

3.6.2 Regional Trend Analysis

Regional trend analyses were done by making use of GBD super-region estimates to study the trends in disease burden of different geographical locations. Line charts were plotted to observe the differences in burden patterns over different years in each region.

The regional trend analysis helped in:

• Comparing burden patterns of GBD super-regions and sub-regions.
• Finding out which regions had higher burden levels and which had lower burden levels.
• Understanding regional epidemiological transitions.

3.6.3 Country-level Trend Analysis

Trend analysis at the country level was done for selected high burden countries during the course of the study. Top burden countries were selected according to the disease burden, and year-wise aggregation was done for each country. Line plots were used for comparative plotting of temporal changes in disease burden indicators between countries.

The country-level analysis helped in:

• Identifying countries that have a continuously high burden.
• Comparing burden trends between countries.
• Evaluating long-term disease burden trends.
• Analyzing temporal changes in a country.

3.6.4 Visualization of Trends

The temporal trend of disease burden indicators was visualized using line plots at various geographic levels. The x-axis referred to the study period, while the y-axis referred to age-adjusted disease burden indicators per 100,000 population.

The following lines were plotted:

• Global trends
• Regional trends
• Country-level trends

The visualization helped in understanding: Burden increase and decrease patterns. Differences in disease burden among regions. Epidemiologic transition over time. Comparison of burden trends between countries and regions. Grids, legends, and markers were added for better clarity.

3.6.5 Forecasting of Disease Burden Using ARIMA

The future burden of NCDs was estimated through forecasting analysis using the Autoregressive Integrated Moving Average (ARIMA) model. ARIMA is a popular time series forecasting method which models the temporal patterns observed in time-series data and forecasts future values. The model was used to calculate the rates of death and DALYs for each disease selected by applying it to the historical data from 1990 to 2021. The trained model was used to forecast the disease burden for the next 28 years (2022 to 2050). The forecasting procedure consisted of the following steps:

• Historical Cardiovascular diseases rates (1990–2021) were ranked by year.
• The yearly mean burden rate was used as the input time series.
• The historical observations were fitted with an ARIMA (1,1,1) model.
• A forecast for 29 years (2022–2050) was generated for future values.

 Line graphs were used to visualize historical observations and forecasted estimates for comparison to look for observed and forecasted trends.

The ARIMA model can be represented in equation (3.2):

                                        (3.2)

Where,  is the forecasted disease burden at time t, c is the constant term,  is the autoregressive coefficient that models the impact of the past disease burden on the current disease burden,  is the moving average coefficient that models the impact of past forecast errors on the current disease burden, and  is the white noise error which represent the random error that occurs at time t. Forecasting was performed separately for Deaths and DALYs of CVDs, DM, and CRDs. The resulting projections provide insights into the expected future burden and support long-term public health planning.

3.6.6 Visualization of Forecast

The forecasting results were visualized using line plots showing both the observed (1990–2021) and forecasted (2022–2050) Cardiovascular diseases burden rates. A vertical reference line was added for the year 2021 to show the change between historical observations and projections. The solid lines indicated observed values and the dashed lines indicated the forecasted estimate. For comparison of historical trends and predicted disease burden, this visualization was used. The ARIMA (1,1,1) model was chosen because it was simple enough to be valid for the historical disease burden data, but still offered a good forecasting performance. A first-order autoregressive term is included in the model, as well as a first-order moving average term, to account for the short-term temporal dependence.

3.7 Geospatial Analysis

A geospatial analysis was performed to understand the spatial pattern of the burden of diabetes disease per country using clusters from the dataset. In particular, the aim was to map patterns of the burden of high, medium, and low risk across different countries and to find differences between them. In the case of spatial mapping, a world shapefile called “ne_110m_admin_0_countries.shp” from the Natural Earth database was used. The shapefile was loaded in Python using the GeoPandas package to get the polygons representing the countries.

The country-level data was extracted from the processed database. The aggregated non-country levels, such as the Global level, regional level, Super-Regional level (for example, Global, South Asia, Sub-Saharan Africa, and others), were excluded to ensure that both datasets are consistent with each other. The cleaned data was integrated with the shape file using country names as key variables. Spatial join made it possible for clustering analysis results to be mapped out according to the geographic boundaries of each country. The cluster labels obtained from the K-means clustering algorithm were used to provide the hotspot classification. The countries with higher cluster labels (cluster values ≥ 3) were considered as hot spots where disease burden was relatively high, while countries with lower cluster labels (cluster values ≤ 1) were considered as cold spots where disease burden was relatively low. The remaining cluster of countries were grouped as the moderate spots. This classification was used to enable visualization and comparison of the geographical distribution of disease burden between countries.

The geospatial visual representation was done by employing GeoPandas and Matplotlib in Python. Choropleth mapping was conducted where in each country is represented using different colors based on the classification obtained, wherein red indicates hot zones, yellow denotes moderate zones, while blue corresponds to cold zones. In addition, the countries which possessed the highest and lowest cluster values were identified and mapped using boundary highlighting techniques. The countries were also labeled in the graphic for clear identification of extreme burden values. The output of the geospatial analysis gave a pictorial depiction of the global burden of diabetes, which helped in identifying geographical differences of burden values. It facilitated comparison of the results obtained from cluster analysis and geographical analysis.

3.7.1 Hot and cold spot identification

Spatial cluster analysis findings give crucial information regarding the geographical pattern of diabetes burden among different countries. The cluster labels from the K-means clustering analysis were used to identify hot and cold spots. Countries with higher cluster labels were deemed hot spots as they had a relatively high disease burden, while countries with lower cluster labels were deemed as cold spots. Countries with intermediate cluster labels were grouped in the “moderate spots” category. This classification allowed for geographical variation in disease burden to be recognized, and for countries with comparable disease burden characteristics to be compared.

The categorization of countries as hot and cold spots points out the presence of spatial inequality regarding the burden of diabetes on a global scale. Hot spot countries could require special interventions aimed at bringing down the burden of disease. The spatial clustering findings overall contribute to our knowledge on geographic inequalities and aid decision-making using evidence-based practice for global and regional health management.

3.8 Socio Demographic Index (SDI)

The SDI analysis was performed to determine the association between socio-demographic development and disease burden by considering multiple health outcomes such as DALYs and deaths. SDI is a combined index that determines the state of development of a country based on income per capita, years of education, and total fertility rate. In this study, the index of development was used to determine the influence of development levels on the variation in disease burden among countries. For the analysis, data on disease burden was obtained from specific GBD data sources.The analysis was done individually for different health indicators, such as DALYs and deaths, so that a thorough understanding of disease burden trends could be achieved. Only Cardiovascular diseases data and the latest years’ data were considered for each indicator, and other aggregate levels apart from the country level, such as global and regional levels, were not considered to ensure consistency with the SDI scores. The estimates at the country level were then aggregated for each location. Predictive Data-Driven SDI Threshold. The chosen indicators (DALYs and mortality) were normalized through the use of z-score standardization to make sure that each variable is comparable to other variables and ready for clustering. K-Means clustering method was used to cluster the countries depending on the values of their normalized burden. The countries were divided into four groups of burdens and then ranked based on their means. The groups were named as Low Burden, Medium Burden, High Burden, and Very High Burden. To analyze further, a binary classification was established wherein the high and very high burden groups were placed under one category of high burden, whereas the low and medium burden groups were placed under one category of non-high burden. The SDI variables were matched to the cluster file by adding the country-level variable for SDI and then merging the two files together. This way, the association between SDI and disease burden was determined. The missing SDI variable was not included in the analysis. In order to establish the thresholds related to SDI that correspond to high burden, a Decision Tree Classifier analysis was conducted. The independent variables were represented by SDI whereas the dependent variables were classified into two groups – countries with high burden and those with non-high burden. The Decision Tree Classifier analysis involved setting a restricted number of levels and leaves to avoid over-fitting.

The Decision Tree Classifier was used to generate meaningful SDI thresholds related to disease burden. The model split the countries into homogeneous groups recursively at each decision node based on the optimal split according to the SDI value. This recursive partitioning produced a tree structure that created SDI cut-off values from the data separating countries with high disease burden from those with lower disease burden. This tree yielded transparent and easily interpretable decision rules which can be employed to classify countries on the basis of their socio-demographic characteristics.The resulting decision tree offered interpretable thresholds for SDI that distinguish countries according to their chances of having a high burden of disease. The rule-based decision rules were derived from the decision tree to enable the interpretation of the link between socio-demographic development and health outcomes. All in all, the analysis of SDI was conducted on several disease burden parameters (DALYs and mortality rates), thus giving an overall perspective on the impact of socio-economic development on health outcomes. The SDI analysis also made it possible to establish the SDI cut-offs using empirical data.

3.8.1 Development of new SDI Threshold

Socio-demographic thresholds that correspond to high burden of disease, a Decision Tree Classifier methodology was used where SDI was the independent variable while the dependent variable consisted of the different disease burden categories. The first step involved classification of countries according to disease burden categories through clustering of standardized DALYs and mortality rates. Clusters were then categorized into two categories, namely high burden and non-high burden countries. The Decision Tree classifier was fitted to determine the most suitable SDI cut-off values for distinguishing between countries with higher disease burdens and those with lower ones. Restrictions on the number of variables allowed for each node and other hyperparameters were imposed to minimize overfitting. The tree structure that emerged from the process yielded empirical SDI cut-offs that correspond to socio-demographic boundaries of higher disease burden. The SDI thresholds that were obtained were applied in exploring the disparities in health burden between different nations. This was possible through the identification of the socioeconomic levels beyond which there was a significantly higher risk of mortality and DALYs.

3.9 Software and Tools

The analyses were carried out in Python programming language (version 3.x), using a Jupyter Notebook interface that offered an interactive computing environment for data management, analysis, and visualization. A number of Python packages were utilized in the different steps of data analysis. Data manipulation involved use of the Pandas package to clean, filter, aggregate, and merge GBD and SDI datasets. Numerical operations were done using the NumPy package where necessary.For the purpose of statistical analysis and machine learning, Scikit-Learn was employed. It was applied for standardizing the data, implementing K-Means clustering technique to group countries according to their disease burden level, and using the Decision Tree classifier technique to find out SDI levels that are responsible for high disease burden levels. For visual representation, Matplotlib was used to make graphs from the outputs of the clustering and decision tree algorithms. As well, GeoPandas was employed for geographical visualization of statistics.

3.10 Summary of chapter

The entire methodology of the study was presented in this chapter to attain the objectives of the study. It explained the data collection, data pre-processing, and data analysis methods that were used on the GBD datasets such as trend analysis, K-means clustering, geospatial mapping, forecasting with the ARIMA model, and SDI threshold analysis with the Decision Tree Classifier. The study also followed the GATHER guidelines to ensure transparency, reproducibility and standardized reporting of health estimates. The application of Python programming, GIS methods and specialized analytical libraries also facilitated the processing and analysis of the GBD data at a scale that supported the processing of large amounts of data, providing reliable and repeatable results. The results from these analyses are provided in the next chapter, including trends in the disease burden of the selected NCDs over time and projections into the future, clustering patterns in the data, comparisons with the GBD regional classifications, hotspot and cold spot distributions, and a set of newly derived SDI thresholds for the selected NCDs.

CHAPTER-04

RESULTS AND DISCUSSION

4.1 Introduction

In this chapter results and discussion of the conducted analyses to achieve the research objectives are presented. The burden of Cardiovascular diseases (CVDs), diabetes mellitus (DM) and chronic respiratory diseases (CRDs) was assessed by Cardiovascular diseases Deaths and Disability-Adjusted Life Years (DALYs) rates retrieved from the  (GBD) 2021 dataset. The analysis was conducted at the global, GBD super-regional and country levels to examine temporal changes in disease burden over the period 1990–2021, and future trends were projected from 2022 to 2050 using the AutoRegressive Integrated Moving Average (ARIMA) model. Furthermore, K-means clustering was used to derive data-driven country clusters based on disease burden and to compare them with the existing GBD regional classification. To explore the spatial distribution of disease burden, geospatial hotspot analysis was performed. The association of socio-demographic development and disease burden was examined using Decision Tree-based Socio-demographic Index (SDI) threshold analysis. The results are presented and discussed in the context of the research aims and compared to relevant published literature where applicable.

4.2 Global, Region, and Country-level Trends of Non-Communicable Disease (NCDs)

In this section, the global, regional and national patterns of age-adjusted mortality rate and DALYs due to various NCDs such as CVDs, DM, and CRDS. Region and country-level differentials are explored to look out for the regions with comparatively high and low disease burden. This section shows temporal trends in the burden of CVD, DM and CRD at global, regional and country levels. The Cardiovascular diseases rate estimates were analysed to study the temporal trend in disease burden after adjusting for differences in age distribution across time. Trends from 1990 to 2021 have been assessed through GBD data, whereas the trends from 2022 to 2050 have been predicted by using forecasting methods.

4.2.1 Cardiovascular disease Burden (CVDs)

This section shows temporal trends in the burden of CVD at global, regional and country levels. The Cardiovascular diseases rate estimates were analysed to study the temporal trend in disease burden after adjusting for differences in age distribution across time. Historical trends (1990-2021) have been evaluated based on GBD data, and future trends (2022-2050) have been projected using forecasting techniques. The analysis gives insight into long-term trends of cardiovascular mortality and DALYs gives the possibility to find possible future trends in the burden of disease. 

4.2.1.1 Global Trends in Mortality due to Cardiovascular disease (CVDs)

The Cardiovascular diseases death rate for CVDs between 1990 and 2021 and their projection up to 2050 are shown in Figure 4.1.

Figure 4.1: Cardiovascular diseases Death Rate due to Cardiovascular disease Observed (1990-2021) and Forecasted (2022-2050)

The observed trend above indicates a substantial reduction in the Cardiovascular diseases death rate for the world throughout the period under analysis. In 1990, the Cardiovascular diseases death rate stood at 37.9 deaths per 100,000 population, which was the highest rate of death in the study region. After this point, the death rate has been on a consistent downward trend reaching the minimum value of 21.8 deaths per 100,000 population in 2020. From 2020 to 2021, there has been an incremental growth in the death rate to 22.7 deaths per 100,000 population. From 2022, the projected mortality rate begins to move upwards reaching 23.6 deaths per 100,000 population, followed by a gradual decrease until 2030. From 2030 to 2050, the projected mortality rate is to stay relatively stable at 25. 5 deaths per 100,000 population. Global death rate from CVDs in relation to the age-standardization rate showed a significant decrease from 1990 to 2021.Such a negative trend can be considered to have achieved something in the field of prevention, early diagnosis, and treatment of Cardiovascular diseases worldwide.

The recent reports of the  project reveal the decrease in mortality from Cardiovascular disease because of better access to medicines and medical care as well as better management of hypertension and hyperlipidemia (GBD 2021 Cardiovascular diseases Collaborators, 2024).The highest mortality burden was seen in 1990, at around 37.9 deaths per 100,000 population. Most countries in the early 1990s lacked access to modern cardiovascular interventions, few people were aware of cardiovascular risk factors and healthcare systems had limited effectiveness in providing preventive care. This led to high rates of cardiovascular deaths during this time period (Mensah et al., 2019). From the mid-1990, to around 2015, there was a strong downward trend in deaths. Such a decrease could be attributed to significant advances in tobacco prevention and control, blood pressure treatment, cholesterol reduction medications and acute cardiovascular care. The decrease in smoking prevalence and the control of cardiovascular risk factors was found to have a significant role in cardiovascular death reduction in the world over time (Roth et al., 2020; Zhang et al., 2025).

The lowest mortality rate was observed around 2020. There was a slight rise, however, in 2021 compared with 2020. This rise could be attributable partly to the impact of the COVID-19 pandemic on health care delivery, postponed diagnosis and treatment of Cardiovascular diseases, and the growing rates of obesity, diabetes and metabolic risk factors in many parts of the world (Wang et al., 2024). The projected trajectory indicates that mortality rates will rise at a moderate rate after 2021 and then level off at about 25.5 deaths per 100,000 people by 2050. The age-standardised mortality rates are still significantly lower than in 1990, but the predicted rise suggests that it will be increasingly difficult to further reduce the burden of Cardiovascular disease in the future. Major cardiovascular risk factors such as high systolic blood pressure, high LDL cholesterol, obesity, diabetes, poor diet and physical inactivity could, in future decades, continue to drive cardiovascular deaths (GBD Risk Factors Collaborators, 2024). The overall results indicate that in the last 30 years there has been a significant improvement in Cardiovascular disease prevention worldwide. However, ongoing investments in prevention strategies, risk factor control programmes and health care systems will be crucial to maintain these gains and to avoid further rising cardiovascular disease burden (Roth et al., 2020; GBD 2021 Cardiovascular diseases Collaborators, 2024).

The highest mortality burden was noted during 1990 when the death rate was around 37.9 deaths per 100,000 population. In the early 1990s, there were fewer countries with access to modern cardiac intervention, and the awareness of cardiovascular risk factors and effectiveness of preventive health care systems was not as high. As a result, the mortality rates for Cardiovascular diseases (CVDs) were still relatively high in this period (Mensah et al., 2019). A sharp drop in mortality was seen from the mid-1990s until around 2015. This decrease reflects significant advances in tobacco control, the treatment of hypertension, the increasing use of cholesterol-lowering medications, and improvements in acute cardiovascular care. Reductions in smoking prevalence and better cardiovascular risk factor control have been reported to be important drivers of the decrease in cardiovascular deaths globally (Roth et al., 2020; Zhang et al., 2025). There was a minimum mortality rate reported in the year 2020. But there was a slight rise in 2021 over 2020. This rise may be partly attributed to the impact of the COVID-19 pandemic on healthcare, delayed diagnosis and treatment for Cardiovascular diseases and the increasing trend of obesity and diabetes as well as metabolic risk factors in many regions (Wang et al., 2024). For the coming decades the role of major cardiovascular risk factors like high systolic blood pressure, high LDL-cholesterol, obesity, diabetes, unhealthy diets and physical inactivity in cardiovascular mortality may still play a role (GBD Risk Factors Collaborators, 2024). In conclusion, the data suggest major progress in the prevention of Cardiovascular disease on a global scale over the past 30 years. However, continued investment in prevention strategies, risk factor control programs, and health systems will be crucial to maintain these gains and avoid an increase in the burden of Cardiovascular disease in the future (Roth et al., 2020; GBD 2021 Cardiovascular diseases Collaborators, 2024).

4..2.1.2 Global Trends in DALYs due to Cardiovascular disease Burden (CVDs)

The Cardiovascular diseases DALYs rate for CVDs between 1990 and 2021 and their projection up to 2050 are shown in Figure 4.2. The global Cardiovascular diseases DALYs rate due to CVDs over 1990–2021 and projected estimates for 2022–2050 are shown in Figure 4.2. The trend in DALYs rates shows a marked decrease in the DALYs rates over the period of the analysis. The highest value for the DALYs rates is recorded for the year 1990, where it was approximately 809 DALYs per 100,000 population. Subsequently, there is a declining trend till the minimum value of 485 DALYs per 100,000 population in the year 2020. There is a minor increase compared to the previous values from 2020 to 2021 with the rate of DALYs per 100,000 population increasing to approximately 507 DALYs per 100,000 population. It can be expected that the DALYs per 100,000 population will keep on increasing after 2021 to a level of around 580 by early 2030s, remaining more or less.

Figure 4.2: Global Cardiovascular diseases DALYs Rate due to Cardiovascular disease Observed (1990-2021) and forecast to 2050.

A significant decrease in the global Cardiovascular diseases DALYs rate from 1990 to 2021 shows great progress in the health outcomes related to Cardiovascular diseases. DALYs are indicators of the impact of mortality and disability, so the downward trend is a sign of a decline in Cardiovascular disease mortality and disability (Roth et al., 2020). The heaviest burden ever recorded in 1990 coincided with a time when Cardiovascular diseases had higher mortality rates, poor treatment methods, low knowledge about the risk factors of Cardiovascular diseases and underdeveloped healthcare systems. Cardiovascular disease (Mensah et al., 2019) burden has significantly decreased due to the advances in healthcare delivery, advanced medical technologies and rising availability of prevention strategies. The significant decline between 1990 and 2015 may also be due to a reduction in risk factors such as smoking, high blood pressure, and high levels of cholesterol. Better risk management and preventive health services have been recognised as critical to reducing the burden of CVD in previous GBD research (GBD 2021 Cardiovascular diseases Collaborators, 2024). A slight increase in 2021 may be due to indirect effects of the COVID-19 pandemic, which may impact health care systems, delay treatment for heart disease, and decrease access to preventive cardiovascular care. This has been emphasized by research that examined the trends in Cardiovascular diseases immediately following the pandemic (Wang et al., 2024).

However, since the increase in DALYs rates is anticipated to occur after 2021, Cardiovascular diseases are still expected to constitute a critical threat to global public health, despite past achievements. Such an increase might be connected with the increased number of cases of obesity, diabetes mellitus, aging populations, sedentary lifestyles, unhealthy eating habits, and constant exposure to various environmental and metabolic risks (GBD Risk Factors Collaborators, 2024). The flattening of the DALYs rates around 2030 suggest that if there were to be any further improvements in the burden of Cardiovascular diseases going forward, they would be at a slower pace than in the previous years. Despite having Cardiovascular diseases rates significantly lower compared to those from 1990, public health measures along with successful cardiovascular risk factor management would be necessary to prevent an increase in the burden of Cardiovascular diseases in the future (Roth et al., 2020). One should note that there has been tremendous success over the last 30 years at reducing the burden of Cardiovascular diseases. Nevertheless, as per the forecasts, there is still a need for further efforts so that the results are not lost in the next few years.

4.2.1.3 Super-Region Trends in Mortality due to CVDs

Figure 4.3 Shows age-standardised mortality rates for Cardiovascular diseases for six GBD super-regions between 1990 and 2021. Substantial regional differences in cardiovascular mortality were apparent during the study period. Some of the regions with the highest numbers of deaths are Central Europe, Eastern Europe and Central Asia. Death rates in this region rose from 63 deaths per 100,000 people in 1990 to more than 72 deaths per 100,000 people in 1994. Since then the number of deaths has been steadily decreasing to about 40 deaths per 100,000 people in 2021. Despite the reduction, this region continued to be the one with the highest mortality rate throughout the whole period.

Figure 4.3: Cardiovascular diseases Death Rate due to Cardiovascular disease across GBD Super Region (1990-2021)

North Africa and the Middle East were ranked second regarding the burden of cardiovascular mortality. The mortality rate was declining gradually from about 59 deaths per 100,000 individuals in 1990 to about 39 deaths per 100,000 individuals in 2021. Despite the general decline in mortality rates, mortality rates were quite high compared to those in the high-income group and South Asia. There was a considerable decline in mortality rates in Southeast Asia, East Asia, and Oceania; from about 45 deaths per 100,000 individuals in 1990 to about 24 deaths per 100,000 individuals in 2021. The South Asian region also recorded mortality reductions for heart diseases. Mortality rates declined from about 28 per 100,000 people in 1990 to about 17 per 100,000 people in 2021. The high-income group had the lowest mortality burden. Their rates went down from around 26 deaths per 100,000 population in 1990 to nearly 10 deaths per 100,000 population in 2021, which was the least rate recorded in all super regions. There were relatively stable rates throughout the whole period under review in the Sub-Saharan region, ranging from 22 to 24 deaths per 100,000 people. The Latin America & Caribbean region was marked by certain fluctuations in the mortality rates that dropped from approximately 30 deaths per 100,000 people in 1990 to almost 29 deaths per 100,000 people in 2021. Overall, the results reveal that there is a consistent trend of decrease in Cardiovascular diseases CVD mortality in all GBD super-regions except for one over time. The extent to which there was an improvement was not uniform among the regions.

These shifts may be related to the health care system reforms, access to medications used to treat heart diseases, good management of high blood pressure, actions taken to prevent people from using tobacco products, and increased awareness about cardiovascular risks among the population. It was already established that the progress made in preventative cardiology, detection of diseases at an early stage, and enhanced therapy of patients suffering from hypertension and dyslipidaemia is a crucial element that helped reduce cardiovascular mortality globally (Roth et al., 2020; GBD 2021 Cardiovascular diseases Collaborators, 2024). Nonetheless, there still persist certain regional inequalities. High mortality in Central Europe, Eastern Europe, and Central Asia may be caused by the fact that the population of these regions shows risky behaviour and suffers from metabolic risks, including smoking, alcohol consumption, high blood pressure, obesity, and poor access to preventative medical care (Mensah et al., 2019; GBD Risk Factors Collaborators, 2024). In contrast, low mortality in the region of high-income countries results from effective healthcare systems and cardiovascular interventions (Mensah et al., 2019; GBD Risk Factors Collaborators, 2024).

4.2.1.4 Super-Region Trends in DALYs due to CVDs

Figure 4.4 depicts the rate of Cardiovascular diseases DALYs caused by Cardiovascular diseases in various GBD super-regions from 1990 to 2021. According to Figure 4.X, the results obtained showed marked differences in the effect of Cardiovascular diseases among the selected regions during the whole period of analysis. In contrast, Central Europe, Eastern Europe and Central Asia had high rates of DALYs caused by Cardiovascular diseases. These rates rose from roughly 1,290 DALYs per 100,000 people in 1990 to 1,570 DALYs per 100,000 people in 1994, and then decreased to 820 DALYs per 100,000 people in 2021.

Figure 4.4: Cardiovascular diseases DALYs Rate due to Cardiovascular disease across GBD Super Region (1990-2021)

North Africa/Middle East had the second largest DALYs burden for heart disease. The age-standardised DALYs rate has been declining steadily from approximately 1220 DALYs per 100,000 population in 1990 to 790 DALYs per 100,000 population in 2021. Even though there was a considerable decrease in burden, the region still suffered from higher burden than seen in high-income and Latin America and the Caribbean regions. The same can be said about Southeast Asia, East Asia, and Oceania where a sharp fall in DALYs rates was observed: from about 960 DALYs per 100,000 population in 1990 to 520 DALYs per 100,000 population in 2021. An aspect of the region of South Asia was distinct from other regions. Between the 1990s and early 2000s there was little change in the DALYs rate, ranging between 650 and 710 DALYs per 100,000 people. Yes, this rate started to increase significantly from 2010, and was approximately 740 DALYs per 100,000 people in 2013, followed by a slight decline. In 2021, the DALYs rate was still rather high with 680 DALYs per 100,000 people. This indicates that the health of the population continues to be adversely affected due to the aforementioned problems in South Asia. The high-income regions always had the smallest number of DALYs burdens out of all the super-regions. There was a significant reduction in the DALYs rate, dropping from approximately 530 DALYs per 100,000 people in 1990 to approximately 230 DALYs per 100,000 people in 2021, which is much lower than the rates in other regions. Latin America and the Caribbean also had a decrease in their DALYs rate. It fell from around 620 DALYs per 100,000 people in 1990 to about 370 DALYs per 100,000 people in 2021. The region of Sub-Saharan Africa showed a slight improvement over the years. The DALYs rate stayed at 490-530 DALYs per 100,000 people over the years and was around 500 DALYs per 100,000 people in 2021.

Findings show a substantial decrease in Cardiovascular disease DALYs for most GBD super-regions between 1990 and 2021. However, this positive trend was observed to differ markedly across regions, indicating that important variations still exist in the prevalence of Cardiovascular diseases and their outcomes worldwide. The decline in DALY rates is likely attributable to improvements in the prevention of Cardiovascular diseases, improved treatment of patients with hypertension and dyslipidaemia, availability of essential drugs and health system improvements overall. Reductions in major cardiovascular risk factors, particularly smoking, hypertension and hypercholesterolaemia, have been identified as key factors driving the decline in Cardiovascular diseases mortality and morbidity (Roth et al., 2020; GBD 2021 Cardiovascular diseases Collaborators, 2024). But there are still some sharp variations across the regions that need to be considered. For instance, the continuing high burden of DALYs in Central Europe, Eastern Europe and Central Asia and South Asia, for example, may be a sign of ongoing prevalence of issues associated with behavioural risk factors, metabolic conditions, age-related problems, and inadequate access to health care. However, the large drop in higher income countries is likely due to the strong structure of their health systems, risk factor management programs, and cardiovascular treatment programs (Mensah et al., 2019; GBD Risk Factors Collaborators, 2024).

4.2.1.5 Regional Distribution of Mortality and DALYs due to CVDs

The Cardiovascular diseases DALYs and death rates from Cardiovascular diseases by region in 2021 are shown in Table 4.1. There was some regional variation of the Cardiovascular disease burden. The highest Cardiovascular diseases DALYs rate was in Oceania (1075.5 DALYs per 100,000 population) then in Central Asia (947.9 DALYs per 100,000 population) and Eastern Europe (926.4 DALYs per 100,000 population). High-income Asia Pacific (146.5 DALYs per 100,000 population), Australasia (148.2 DALYs per 100,000 population) and Western Europe (193.7 DALYs per 100,000 population) had the lowest rates of DALYs per 100,000 population.

Table 4. 1: Regional Distribution of Deaths and DALYs Rate

A similar pattern was observed for mortality rates. The Cardiovascular diseases mortality rate was highest in Central Asia (47.4 deaths per 100,000 population), followed by Eastern Europe (44.1 deaths per 100,000 population) and Oceania (43.7 deaths per 100,000 population). In North Africa and the Middle East, deaths per 100,000 population were also high, at around 39.5 deaths per 100,000 people. On the other hand, the lowest death rates were observed in High-income Asia Pacific countries (6.3 deaths per 100,000 population), Australasia (7.2 deaths per 100,000 population) and Western Europe (9.7 deaths per 100,000 population). Cardiovascular burdens were relatively high in the remaining regions, namely South Asia (DALYs rate of 678.7 per 100 000 population) and Southeast Asia (DALYs rate of 740.9 per 100 000 population). Likewise, the Caribbean, Central Sub-Saharan Africa and Central Europe have higher DALYs and mortality rates than the world average. The DALYs rates range from about 241 to 380 per 100,000 people, and the mortality rates are between 10 and 19 deaths per 100,000 people in the Latin American regions, which generally show moderate cardiovascular burden.

The findings reveal significant geographical disparities in Cardiovascular disease burden between regions of the GBD. Increased rates of key cardiovascular risk factors such as hypertension, obesity, poor dietary habits, smoking and limited access to preventive health services – may explain the higher burden seen in Oceania, Central Asia and Eastern Europe. Eastern Europe and Central Asia have been the focus of previous studies that have reported high and disproportionately high cardiovascular mortality and disability burden compared to many other parts of the world (Roth et al., 2020; GBD 2021 Cardiovascular diseases Collaborators, 2024).This was in contrast with the lowest rates of mortality and DALYs observed in High-income Asia Pacific, Australasia and Western Europe. The health-care systems are generally robust, cardiovascular risk factors are better controlled, knowledge of evidence-based therapies is high, and public health interventions are effective in these regions. The regional differences highlight the need for tailored prevention programmes and investments in healthcare in high-burden areas, to help reach the broader global target of halving Cardiovascular disease mortality and disability by 2030 (Mensah et al., 2019; GBD Risk Factors Collaborators, 2024)

4.2.1.6 Country-wise Trends in Mortality due to CVDs

The Cardiovascular diseases death rates due to Cardiovascular disease among the ten countries with the highest mortality burden between 1990 and 2021 are shown in Figure 4.5

Figure 4.5: Cardiovascular diseases Death Rates due to Cardiovascular disease Top 10 Highest burden countries

The mortality pattern in each country during the study period had a lot of variation. Nauru had the highest cardiovascular mortality rates of all countries. The death rate rose significantly from about 95 deaths per 100,000 population in 1990 to peaking at almost 125 deaths per 100,000 population in the mid-2000s. After this, there was a gradual declining period of mortality rates, but Nauru continued to be the country with the highest mortality rates for the majority of the study. There was a significant drop following 2015, when the death rate stood at around 70-78 per 100,000 people by 2021. The Marshall Islands also had a chronic high cardiovascular mortality burden. The mortality rate rose steadily from the 1990 rate of about 83 deaths per 100,000 population to about 91 deaths per 100,000 population in 2014-2015. Like Nauru, there was also a significant decrease after 2015 which saw deaths per 100,000 population drop to around 68 in 2021. The same pattern of rising deaths and falling rates was seen in Tuvalu, with mortality rates climbing from around 79 deaths per 100,000 population in 1990 to nearly 90 deaths per 100,000 population before dropping precipitously after 2015 to around 62 deaths per 100,000 population in 2021. Solomon Islands had the largest changes between the two years. Mortality was relatively stable in the 1990s and the first few years of the 2000s, with a significant rise in mortality post 2010. In 2015, the death rate was nearly 79 deaths per 100,000 population, but decreased significantly to about 56 deaths per 100,000 population by 2021. The level of variability was moderate in Egypt during the study period. Mortality rates rose steadily from about 67 deaths per 100,000 people (1990) to about 80 deaths per 100,000 people (2009-2010), but dropped significantly after 2015, to about 48 deaths per 100,000 people (2021).

The mortality trends in Afghanistan, the Syrian Arab Republic, Turkmenistan, Uzbekistan and the Federated States of Micronesia were also negative in the second half of the study period. While death rates differed from one of these countries to another, they all showed significant declines since 2015. In general, mortality rates were lower than in the early years of the study period, with most countries having rates between 35 and 50 deaths per 100,000 population in 2021. At the end of the study period, the mortality rate in Uzbekistan and Turkmenistan was among the lowest with about 36-38 deaths per 100,000 population in 2021. Overall, the findings suggest that in these most burdened countries, the death rates for Cardiovascular diseases have not dropped significantly over the last 15 years, but that significant reductions have been reached in recent years.The most striking declines were post-2015, indicating that a lot has happened in many high-burden settings in the area of Cardiovascular disease prevention, diagnosis and management.

Countries with a high mortality burden like Nauru, Marshall Islands and Tuvalu could have a high prevalence of metabolic and behavioural risk factors such as obesity, hypertension, diabetes, physical inactivity, and unhealthy dietary habits. Pacific Island countries have been listed as among the highest prevalence for obesity and non-communicable disease risk factors worldwide (Roth et al., 2020; GBD Risk Factors Collaborators, 2024) and play a major role in contributing to high cardiovascular mortality rates. The significant reduction in deaths following 2015 could be due to better access to health services, better control of cardiovascular risk factors, greater access to necessary medicines, and greater public health efforts on non-communicable disease.But the death rates in the countries are still substantially above the world’s average, implying that the burden of heart diseases is still substantial. If the achievements are to be maintained and further reductions are made in the number of cardiovascular deaths in the high-burden countries, then more efforts in covering prevention programs, risk factor reduction, and health infrastructure development will be required (Mensah et al., 2019; GBD 2021 Cardiovascular diseases Collaborators, 2024).

4.2.1.7 Country-wise Trends of DALYs due to Cardiovascular diseases (CVDs)

Figure 4.6 Represents the Cardiovascular diseases DALYs rate for CVDs in the seven GBD super-regions from 1990 to 2021. There have also been notable variations in the incidence of Cardiovascular disease in different geographical locations during the period under review.

Figure 4.6: Cardiovascular diseases DALYs Rates due to Cardiovascular disease Top 10 Highest burden countries

The highest rates of DALYs were constantly noted for Central Europe, Eastern Europe and Central Asia. The rate of DALYs increased from around 1290 per 100,000 in 1990 up to the maximum value of 1570 in the mid 1990s and decreased to 820 per 100,000 in 2021. The second highest rates of DALYs were observed in North Africa and the Middle East that gradually decreased from around 1,220 DALYs per 100,000 population in 1990 down to ~790 DALYs per 100,000 population in 2021. A similar trend was also observed in Southeast Asia, East Asia and Oceania, where the DALYs rates decreased from 960 to 530 DALYs per 100,000 population. The most significant absolute decline of Cardiovascular disease burden is observed in high income regions with steady rates’ decrease from about 530 DALYs per 100,000 people in 1990 to nearly 230 DALYs per 100,000 people in 2021. Latin America and the Caribbean also experienced a decline of DALYs rates from around 620 DALYs per 100,000 to 370 between the years of study. The burden was rather stable in South Asia, despite some fluctuations. It slightly increased after 2005 and reached around 740 DALYs per 100,000 population around 2012. All regions saw increases, with Sub-Saharan Africa having the worst performance. The DALYs rate did not change significantly throughout the study period, ranging between 490 and 530 DALYs per 10,000 people, with only minor declines until 2021.

The outcomes of the research show some significant variations among GBD super-regions regarding the burden of Cardiovascular diseases. These variations can be attributed to some socio-economic differences, availability of health care, epidemiological transition (ET), demographic patterns and exposure to risk factors for CV diseases. Central and Eastern Europe/Central Asia Cardiovascular disease burden was the highest in this particular region throughout the period under analysis. It is easy to explain the high rate of DALYs in the early 1990s through the increased prevalence of hypertension, smoking, poor nutrition, excessive alcohol consumption and lack of CV health care. Subsequent decline may have been due to improvements in health care services, risk factor and Cardiovascular disease management strategies (Roth et al., 2020; GBD 2021 Cardiovascular diseases Collaborators, 2024). The region of North Africa and the Middle East. An area of north Africa and the Middle East. This area showed a downward trend, but the burden of Cardiovascular diseases was still high. The region has been identified as having a high burden of Cardiovascular diseases and a few factors have been identified to play a significant role in this increase which is due to rapid urbanization, increasing prevalence of obesity, diabetes, and dietary transitions. These risk factors still have a large impact on cardiovascular health outcomes, despite increasing access to healthcare (GBD Risk Factors Collaborators, 2024). The area of Southeast Asia, East Asia and Oceania is referred to as the Southeast Asian region. A significant decrease in this region could be linked to better health service provision, better management of hypertension and Cardiovascular diseases, and effective implementation of public health interventions to major cardiovascular risk factors (Mensah et al., 2019). High-Income Regions Cardiovascular DALYs rates decreased most significantly and consistently in high income areas. The possible reasons that can contribute to the reduction are increased healthcare provision, high rates of screening, control of blood pressure and cholesterol, reduced smoking rates, and improved access to evidence-based care for CVD (Roth et al., 2020). The geographical location is Latin America and the Caribbean. The continued reduction in cardiovascular burden implies better health services and disease management practices. Despite these moderate rates of DALYs in many countries in the region, however, Cardiovascular diseases remain a public health problem, as evidenced by their persistence (GBD 2021 Cardiovascular diseases Collaborators, 2024).

Unlike South East Asia, there were mixed trends in South Asia where increases in burden were observed since 2005. The pattern can be explained by the high number of births, urbanization, prevalence of diabetes and obesity, as well as persisting problems in achieving optimal control of the risk factors for cardiovascular conditions. South Asia is considered one of the hotspots for the burden of Cardiovascular diseases (GBD Risk Factors Collaborators, 2024), and thus it significantly contributes to the burden of cardiometabolic diseases. Sub-Saharan Africa There have not been any changes in the region of interest within the period of the study. The decline in DALYs rates is due to the scarce amount of healthcare resources, late diagnosis, low coverage of the treatment, and increasing impact of metabolic and behavioral risk factors. Additionally, competing health concerns and scarce resources may impede the rapid decrease of the burden of Cardiovascular diseases in the region (Mensah et al., 2019). The findings overall indicate that the burden of Cardiovascular disease has declined in most of the super-regions since 1990. However, there is significant variation in the extent of progress made.

4.2.2 Diabetes Mellitus Disease (DMs)

This section reviews the worldwide, regional and country-level patterns of Cardiovascular diseases mortality and DALYs due to diabetes mellitus over the study period. The analysis also offers a basis for understanding the key geographical trends and variations of diabetes burden, as well as the comparative burden in different regions and countries.

4.2.2.1 Global Trends in Mortality due to Diabetes Mellitus Disease (DMs)

World standard mortality rate due to diabetes mellitus causing mortality from diabetes mellitus during the years 1990 to 2021, as well as estimated future trends until 2050 are presented. The trend observed in the study period is in Figure 4.7.

Figure 4.7: Global Cardiovascular diseases Death Rate due to Diabetes Mellitus Disease Observed (1990-2021) and Forecast (2022-2050)

The mortality increased from about 4.07 deaths per 100,000 population in 1990 to about 4.29 deaths per 100,000 population in 1995. This mortality trend showed an increase, which continued and led to an increase to the maximum level in 2003 with a rate of about 4.41 deaths per 100,000 population. The rate of mortality reduced slightly afterwards, to reach the lowest rate of about 4.20 deaths per 100,000 population in 2012. The mortality rate began increasing again after 2012, reaching another peak of about 4.40 deaths per 100,000 population in 2020. Afterward, the mortality rate reduced slightly to reach a level of about 4.37 deaths per 100,000 population in 2021. According to the forecasted trend, the mortality rate would slightly decline and reach a rate of about 4.34 deaths per 100,000 population by 2030. Thereafter, the forecasted mortality rate remains fairly stable until 2050. Overall, these results indicate that there was a reasonable level of stability in mortality of diabetes globally throughout the entire period under consideration.

The variations in the global Cardiovascular diseases diabetes mortality rate reflect the reality that despite development in the healthcare system and improvement in diabetes treatments, diabetes continues to be an issue in public health in the whole world. The death rate of diabetes has not considerably reduced within the past three decades as compared to several other non-communicable diseases (NCDs) (GBD 2021 Diabetes Collaborators, 2024). The increase that occurred in the 1990s and early 2000s might have been attributed to the widespread presence of type 2 diabetes in the world. Such an increase was highly associated with aging of populations, urbanization, obesity prevalence, inactive lifestyle, and unhealthy diets (IDF, 2021). The decrease that was experienced between approximately 2003 and 2012 might be attributed to diabetes awareness and early diagnosis, more access to glucose-lowering drugs, and management of diabetes complications. Public health measures were put in place in many nations to promote prevention and control of diabetes; the measures led to declines in mortality rates temporarily (Zheng et al., 2018). These measures could have contributed to the reduction in deaths from diabetes. The renewed increase after 2012 suggests that the greater prevalence of diabetes and metabolic risk factors were compensated by improvements in the management of the disease.

Physical inactivity, being overweight or obese and unhealthy diets are some of the risk factors for the growing number of diabetes cases in the world (GBD Risk Factors Collaborators, 2024). It is possible that the second surge of deaths in 2020 is due to the COVID-19 pandemic. People with diabetes are at increased risk of developing complications from the infection, and disruption of health care delivery services may have impacted diabetes management adversely (Khunti et al., 2021). The predicted plateauing of diabetes-related deaths until 2050 creates grounds for predicting gradual decline in deaths from diabetes. Even as major progress is being made in the treatment of diabetes, expect to see an increase in numbers being diagnosed with the disease in years to come, meaning that any considerable decline in mortality rate will not be easy. Therefore, efforts aimed at helping people avoid obesity, eating bad food, doing exercise, and reducing metabolic risks will still play a crucial role in preventing diabetes in the future (IDF, 2021; GBD 2021 Diabetes Collaborators, 2024). On the whole, the findings indicate that the death rates from diabetes have remained rather stable during the last three decades, with rises and falls reflecting the complex interaction of progress in diabetes care and the increasing number of risk factors for diabetes.

4.2.2.2 Global Trends in DALYs due to Diabetes Mellitus Disease (DMs)

The age-adjusted DALYs rate caused by diabetes mellitus in the entire world between 1990 and 2021 together with its projections up to 2050 is in figure 4.8. The trend shows an increase in the number of DALYs caused by diabetes.

Figure 4.8: Global Cardiovascular diseases DALYs Rate due to Diabetes Mellitus Disease Observed (1990-2021) and Forecast (2022-2050)

Globally, The DALYs rate increased from about 149 DALYs per 100,000 population in 1990 to about 209 DALYs per 100,000 population in 2021, which is an increment of 39%. The increase was rather steady between 1990 and 2013, and then became faster to reach 2021. The predicted increase implies that the burden of diabetes will keep on increasing in the coming decades. The DALYs rate will increase from about 210 DALYs per 100,000 population in 2022 to almost 257 DALYs per 100,000 population in 2050. The growth rate is not especially fast, but it slowly goes up in the forecasted time period. Generally, it can be concluded that the burden of diabetes mellitus has been increasing all along worldwide and does not demonstrate any signs of stabilization during both the observed and the forecasted periods. The never-ending nature of the increase of DALYs rates implies that diabetes mellitus keeps on being one of the most burdensome diseases in the world. In comparison with the relatively stable mortality rates observed in the period under consideration, DALYs become a better measure of the health consequences of diabetes as they incorporate both death and disability issues (Murray & Lopez, 1996). The increasing trend between 1990 and 2021 can be accounted for by the increasing incidence of diabetes prevalence worldwide. Factors like urbanization, aging population, physical inactivity, malnutrition, and obesity rates have become some of the significant contributors to the prevalence of diabetes mellitus worldwide, especially Type 2 diabetes (International Diabetes Federation [IDF], 2021).

It might be that the increase in DALYs rate around 2013 might be associated with the progress made in medicine because it led to a more considerable life expectancy of diabetic patients, but at the same time, many patients were suffering from other diseases, including Cardiovascular diseases, renal diseases, neuropathy, and visual impairment (Zheng et al., 2018). Predicting the development of the problem up to 2050, the problem is going to persist, and the impact of the problem will keep increasing. The reasons for this growth may be associated with obesity rates, aging population, lack of physical activities, and metabolic risks. One of the risk factors identified in the  studies recently is high body mass index and high fasting plasma glucose (GBD Risk Factors Collaborators, 2024).No decrease in the DALYs rates shows the problem which is related to the issue. In spite of the developments in the treatment of the disease, which have led to the improvement in survival and reduction of complications, apparently, they were not enough to compensate for the increase in the number of cases of diabetes (GBD 2021 Diabetes Collaborators, 2024).

On the other hand, the projection about the increase in DALYs points at the importance of implementation of public health initiatives aimed at reducing risk factors. Healthy eating, physical activity, prevention of obesity, diagnosis and treatment will be crucial to prevent the further increase in diabetes mellitus (IDF, 2021). In conclusion, it can be stated that the problem of diabetes mellitus will be one of the biggest challenges for the future years. The constant growth in DALYs rates shows the increasing importance of disability due to diabetes and necessity of implementing the prevention initiatives on the global level.

4.2.2.3 Region Trends in Mortality due to Diabetes Mellitus Disease (DMs)

The figure 4.9 below the age-standardized death rates due to diabetes mellitus across GBD super-regions for the period of time ranging from 1990 to 2021. The above trends reveal that there is considerable inequality in diabetes related mortality around the world.

Figure 4.9: Cardiovascular diseases Death Rate due to Diabetes Mellitus disease across GBD Super Region (1990-2021)

The Sub-Saharan Africa region remained on the top of the list of regions having high death rates during all years starting from about 9.2 deaths per 100,000 individuals in 1990 up to nearly 11.0 deaths per 100,000 individuals in 2021. The increase remained rather consistent with a sharp rise noted after the year 2010, thus confirming that the region is still suffering from increasing mortality from diabetes. Another region that recorded a high mortality rate is Latin America and the Caribbean region, recording a drop from 9.3 deaths per 100,000 individuals in 1990 to approximately 8.5 deaths per 100,000 individuals in 2021. There was also a remarkable upward trend in the North Africa/Middle East region, rising from approximately 6.1 per 100,000 individuals in 1990 to 7.5 per 100,000 individuals in 2021, with most of the increase being experienced during 2000-2015. Similarly, the prevalence rate increased from approximately 5.5 per 100,000 individuals in 1990 to 6.5 per 100,000 individuals in 2021 in South Asia due to type 2 diabetes prevalence in the region.The mortality rate related to diabetes in High-income regions has decreased from approximately 3.6 deaths per 100,000 individuals in 1990 to approximately 2.3 deaths per 100,000 individuals in 2021.

The downward trend might be attributed to factors such as improvements in healthcare infrastructures, early diagnoses, effective medications, and other public health services that exist in those countries (NCD Risk Factor Collaboration, 2016). Southeast Asia, East Asia, and Oceania were relatively constant with death rates near to 3.5 deaths per 100,000 people, with slight increments after the year 2015. Meanwhile, Central Europe, Eastern Europe, and Central Asia registered relatively constant low death rates increasing slightly from 2.0 up to around 3.6 deaths per 100,000 people by 2021. The marked difference in the condition of Sub-Saharan Africa compared to that of richer nations indicates how important access to healthcare facilities, economic power, and public health institutions are in managing diabetes. The reason behind the persistently high death rate in poorer areas is associated with late diagnosis, lack of access to insulin and drugs, and poor healthcare infrastructures (IDF, 2021). Besides, the increasing trends in the prevalence of diabetes in North Africa, the Middle East, and South Asia have been linked to the rapid transition taking place there (Cho et al., 2018). This disparity shows the need for tailored interventions in this sector. It is important to enhance primary healthcare delivery systems, provide affordable diabetes medication, and implement preventive strategies to prevent high mortality rates from diabetes, especially in regions with increasing diabetes deaths (World Health Organization [WHO], 2023).

4.2.2.4 Regional Trends in DALYs due to Diabetes Mellitus Disease (DMs)

The age-adjusted DALYs rate associated with diabetes mellitus among GBD super-regions between 1990 and 2021 are shown in Figure 4.10.

Figure 4.10:  Cardiovascular diseases DALYs Rate due to Diabetes Mellitus disease across GBD Super Region (1990-2021)

The highest DALYs rate was registered in Latin America and the Caribbean region for all years analyzed, beginning from approximately 322 DALYs per 100,000 people in 1990 and finishing with 350 DALYs per 100,000 people in 2021. It should be mentioned that the DALYs rate slightly declined from 1995 to 2005, but then it began increasing and reached its maximum value in 2021. However, the most dramatic change can be seen in the North Africa and the Middle East region when the DALYs rate increased from 205 DALYs per 100,000 people in 1990 to 350 DALYs per 100,000 people in 2021. Particularly, the rapid increase, especially after 2005, equates the North Africa and the Middle East region to Latin America and the Caribbean by the DALYs rate. The highest DALYs rate was registered in Latin America and the Caribbean region during the whole period analyzed, beginning from approximately 322 DALYs per 100,000 people in 1990 and reaching more than 350 DALYs per 100,000 people in 2021. In addition, it should be mentioned that the considerable growth, particularly after 2005, has caused the rates of DALYs in North Africa and the Middle East to match those of Latin America and the Caribbean in relation to the DALYs rates.

The DALYs rate in Latin America and the Caribbean was highest throughout the whole period observed, starting with the figure of about 322 DALYs per 100,000 population in 1990 and reaching over 350 DALYs per 100,000 population in 2021. It is also worth mentioning that there was a slight fall between 1995 and 2005, and then the growth occurred, reaching its peak in 2021. In addition, it should be noted that the greatest rise occurred in North Africa and the Middle East, where the DALYs per 100,000 population grew from about 205 in 1990 to about 350 in 2021. However, the DALYs rates in Sub-Saharan Africa were quite high at 258 DALYs per 100 000 in 1990 but had increased gradually, and attained 320 DALYs per 100 000 in 2021. Despite this, the rate of increase was quite low, but the burden was very high in the region. The same can be said about the DALYs rates in South Asia which have been increasing over the years, from an estimate of 175 DALYs per 100,000 population in 1990 to an estimate of 230 DALYs per 100,000 population in 2021, due to the rapid epidemiological transition taking place in countries such as India, Pakistan and Bangladesh, owing to urbanization and increase in obesity (Sun et al., 2022).

The DALYs rates in the high-income regions and in Southeast Asia, East Asia, and Oceania have been relatively low and quite stable, at around 135–180 DALYs per 100,000 in 2021. Stability of higher incomes may be linked to an efficient system of healthcare, presence of technologies of treatment of diabetes, and constant investments into prevention of the disease through implementation of programs in public health care (Magliano et al., 2021). The lowest DALYs rates in 1990 were observed in the territories of Central Europe, Eastern Europe and Central Asia (108 DALYs per 100,000 population in 1990); however, at the same time, the region is experiencing consistent and statistically significant growth of DALYs rates with the DALYs per 100,000 population reaching 190 in 2021, that indicates a significant diabetes burden which was underestimated before. Contrast between low-income and high-income regions clearly shows the influence of socio-economic determinants of diabetes burden and access to healthcare and risk factors. An acceleration of DALYs growth in LMICs after 2010 was caused by global epidemiological transition, that was related to increasing level of urbanisation, sedentary lifestyle and poor nutrition, which led to a rapid growth of type 2 diabetes burden in LMICs (Ogurtsova et al., 2022). In addition to this, diabetes is a condition that is linked to several complications like heart disease, end-stage kidney disease, and lower limb amputation, which increase the DALYs in regions with poor health care facilities (GBD 2021 Diabetes Collaborators, 2024). Such findings highlight the role of public health measures in such regions. In areas of high gain regarding DALYs, early detection and provision of medicines along with changes in lifestyles can help reverse the trend of diabetes mellitus. (WHO, 2023).

4.2.2.6 Region Distribution of Mortality and DALYs due to Diabetes Mellitus Disease (DMs)

Table 4.2 the Cardiovascular diseases DALYs and mortality rate due to diabetes mellitus in 21 GBD regions in 2021. There was a great variability in the burden of diabetes in different regions. Oceania had the greatest burden regarding DALYs at about 904 DALYs per 100,000 people compared to other regions. Second in the list of greatest burden of DALYs is Southern Sub-Saharan Africa where it was estimated at 542 DALYs per 100,000 population. Central Latin America is the third region with the greatest burden where it was estimated at 443 DALYs per 100,000 people.

Table 4. 2: Region Distribution of Deaths and DALYs Rate

This pattern could also be seen in case of death rates. The age-standardized death rate of diabetes mellitus was highest in the region of Oceania, which recorded 27 deaths per 100,000 population. Second highest mortality was seen in Southern Sub-Saharan Africa with 19 deaths per 100,000 people, while third highest mortality was seen in Central Sub-Saharan Africa (12 deaths per 100,000 people) and fourth highest mortality was reported in Central Latin America (11 deaths per 100,000 people). On the contrary, lowest mortality was reported in the High-income Asia Pacific (0.95 deaths per 100,000 population), Western Europe (2.14 deaths per 100,000 population), Australasia (2.24 deaths per 100,000 population) and East Asia (2.31 deaths per 100,000 population).The burden of diabetes was considered moderate in South Asia and Southeast Asia, where the number of DALYs per 100,000 individuals was at around 264 and 275, respectively, and the mortality rate was at less than 8 deaths per 100,000 individuals. Similarly, the region of North Africa and Middle East had relatively high numbers of DALYs and deaths per 100,000 individuals.

The burden of diabetes measured by mortality rates ranged from intermediate in most Latin American regions to high in some, while the burden of disability was low everywhere in European high-income regions. Overall, the findings confirm a significant regional variation in the burden of diabetes mellitus across GBD regions. Regions with the highest burden were generally Oceania and Sub-Saharan African regions, with generally low burden in high-income regions, specifically High-income Asia Pacific, Australasia and Western Europe. Diabetes burden in Oceania and Sub-Saharan Africa is particularly high, and could relate to a combination of metabolic, behavioural and healthcare-related factors.The high prevalence of obesity, low physical activity, dietary modifications and lack of early detection and diabetes management facilities are known to be the major factors contributing to the development of deaths and complications associated with diabetes in such countries. The previous  studies have revealed that Pacific Island countries have among the highest prevalence rates of diabetes in the world (GBD 2021 Diabetes Collaborators, 2024; NCD Risk Factor Collaboration, 2023), as well as the highest diabetes DALYs and diabetes mortality rate.

The relatively low burden in High-income Asia Pacific, Western Europe and Australasia could be attributed to more effective glycaemic control and diabetes complications management, higher quality of healthcare facilities and widespread screening programs. Such regions have greater access to health care facilities, diabetes medications and preventive care measures. The differences between the regions highlight the importance of further development of diabetes prevention efforts and increased access to healthcare in high burden regions to reduce diabetes mellitus (DM) related early mortality and disabilities (International Diabetes Federation, 2024; GBD Risk Factors Collaborators, 2024).

4.2.2.7 Country wise Trends in Mortality due to Diabetes Mellitus Disease (DMs)

Figure 4.11 illustrates the Cardiovascular diseases mortality rates due to diabetes mellitus between 1990 and 2021 for 10 most affected countries. From the trends noted, the rate of mortality from diabetes was highest among small island countries, hence making these populations bear the greatest burden of the disease, particularly in the Pacific region.

Figure 4.11: Cardiovascular diseases Death Rates due to Diabetes Mellitus Disease (DMs)

Top 10 Highest burden countries

Amongst all countries considered, Fiji experienced the largest death rate. This country had the death rate of 51 per 100,000 of population in 1990, which rose sharply to 71 per 100,000 of population in 2005 and then stabilized at the slightly lower level of 69 per 100,000 of population by 2021. However, the high death rate persisted in Fiji even in recent years, being much higher compared to that of other countries in the top 10, indicating high prevalence of diabetes in the country. An increase in the death rate was steady and consistent in Kiribati, where it increased from around 38 deaths per 100,000 population in 1990 to about 50 deaths per 100,000 of population in 2021 – the second highest in the end of observation period. Death rate per 100,000 of population also increased significantly in Marshall Islands from around 25 in 1990 to about 44 in 2021, which is one of the highest increases amongst the top 10 countries. Nauru had the death rate of around 31 in 1990 and 38 per 100,000 population in 2021. Nevertheless, there has been an increase in the diabetes rate of Bahrain from about 28 per 100,000 inhabitants in 1990 to nearly 40 per 100,000 inhabitants in 2005; the growth has stabilized at about 32 per 100,000 inhabitants in 2021, probably because of better health care facilities in the country.

The Federated States of Micronesia, Niue, Solomon Islands, American Samoa and Tonga also saw death rates of between about 20 and 35 deaths per 100,000 people throughout this time period. However, their rates were relatively low among the top 10 groups of countries, but all of these countries had death rates significantly higher than the world’s average. The trends were relatively stable in American Samoa and Tonga and were steadily increasing in Solomon Islands over the period. This clustering of the highest diabetes mortality rates in small Pacific Island countries does not come as a surprise, given the extensive documentation of the particular susceptibility of Pacific populations. All these factors have resulted in high prevalence of type 2 diabetes and its complications in these countries in particular (Mishra et al., 2021). The fact that Bahrain was identified as one of the top 10 countries with the highest burden of diabetes adds to the larger picture of growing diabetes burden in the Middle East and Gulf region, which is partly attributable to the high prevalence of obesity, inactivity and swift economic and dietary changes (Cho et al., 2018).

High mortality rates persisted in all these countries, and in a few, like Fiji or Bahrain, were slowly declining during the later years, a welcome sign that some improvements in diabetes management have been made, however not enough to make a difference in the number of diabetes deaths. The limited availability of essential medicines including insulin and the poorly developed primary health care systems are major challenges to the improvement of outcomes in these settings (Basu et al., 2019).

4.2.2.8 Country wise Trends in DALYs due to DM

Figure 4.12: The 10 countries with the highest burden of diabetes mellitus are shown on , where the age-standardised DALYs rates are plotted. The DALYs data, as with the mortality data above, shows Pacific Island countries to be the highest burdened with diabetes, with rates much higher than the global average, and a significant and cumulative impact of diabetes on early deaths and disabilities in these countries.

Figure 4.12: Cardiovascular diseases DALYs Rates due to Diabetes Mellitus Disease (DMs)

Top 10 Highest burden countries

Fiji had the highest DALYs rate for the entire period, with an estimated rate of around 1,400 DALYs per 100,000 population in 1990 before steeply increasing to nearly 1,950 DALYs per 100,000 population by around 2010 and remaining relatively stable at around 1,940 DALYs per 100,000 population through to 2021. This very high and sustained burden exemplifies the severe burden of diabetes on mortality and disability in Fiji. The Marshall Islands had among the highest increases in DALYs of the top 10 countries, from approximately 930 DALYs per 100,000 population in 1990 to approximately 1,560 DALYs per 100,000 population in 2021, a nearly twofold increase over 30 years. Kiribati also showed a large and steady increase, from around 1,110 DALYs per 100,000 people in 1990 to around 1,550 DALYs per 100,000 people in 2021, becoming the third country with the highest burden at the end of the observed period. Overall, Nauru experienced a steady rise from around 950 DALYs per 100,000 population in 1990 to around 1,320 DALYs per 100,000 population in 2021, followed by a slight plateau from 2000 to 2010, and then a continued rise.

Modern increases were observed in both American Samoa and Micronesia (Federated States of) with rates becoming closer in the future, around 1080-1100 DALYs per 100,000 population, showing the high burden of diabetes across Micronesian and Polynesian populations, and its increasing prevalence. The other four in the top 10 Niue, Cook Islands, Tonga, and Samoa had lower DALYs rates, between about 650 and 800 DALYs per 100,000 population in 1990. This was the case for all four countries, though, with rates increasing from around 950 to 1,080 DALYs per 100,000 population between 2000 and 2021. Of particular note, Tonga and Cook Islands had lower DALYs rates in 1990, but experienced consistent and unbroken progress throughout the entire period, corresponding to a rise in the DALYs rate from about 1,100 to 1,300 in 2021. In 1990, Samoa had a low DALYs rate, but it steadily increased to approximately 970 DALYs per 100,000 population in 2021. The strong association of the highest DALYs rates with Pacific Island countries is consistent with established research that indicates these populations are also genetically vulnerable to efficient energy storage – the “thrifty genotype” hypothesis, which posits that they may have a genetic disposition to this function which becomes a liability in the settings of high calorie intake and high carbohydrate content of modern diets (Neel, 1962, cited in Mishra et al., 2021)

In addition, genetic risk, the nutrition transition in Pacific Island nations (defined as the shift from a diet of traditional whole foods to that of processed foods and added sugars), lack of physical activity, high obesity prevalence and limited access to preventive health services have set the stage for the emergence and worsening of type 2 diabetes and its complications (Poblete et al., 2022). These very high DALYs rates are due to a high prevalence of diabetes as well as the burden of diabetes-related complications such as Cardiovascular disease, chronic kidney disease, diabetic neuropathy and vision loss, compounded by delayed diagnosis and poor access to comprehensive diabetes care in small island and developing states (SIDs) (Mishra et al., 2021). The small size and geographic isolation of many of these countries further impede their capacity to develop and support these health care systems and make them more vulnerable to dependence on foreign health care assistance and support (NCD Alliance, 2023). These findings highlight the urgent and sustained international attention needed to deal with the diabetes crisis in Pacific Island countries. The most critical strategies to help arrest and reverse the increasing DALYs burden in these uniquely vulnerable groups are culturally tailored prevention programmes, primary healthcare system capacity building, enhanced access to insulin and other essential diabetes drugs and strengthening national surveillance systems (WHO, 2023; IDF, 2021).

4.2.3 Chronic Respiratory Disease (CRDs)

This subsection analyses the global, regional, and country-specific trends in the mortality and Disability-Adjusted Life Years (DALYs) of chronic respiratory disease. The time-series methods are used to estimate the changes in the number of deaths and the total disease burden related to chronic respiratory diseases that enable the determination of any long-term variations at various levels of geography. Such analyses give a dynamic perspective of how there is a change in burden of chronic respiratory disease with time and contribute to interpretation of the effects of health interventions and demographic changes in various areas. The trends in the results demonstrate the rises or falls in the chronic respiratory mortality and DALYs, depending on the region and country. There are countries that show a steadily high burden, whereas other countries have a decline over time, and this is indicative of better access to healthcare or prevention of disease or improved socio-economic status. Through such patterns, one can find out the regions, in which the public health efforts have been successful and where further resources or policy action can be necessary.

4.2.3.1 Global Trends in Mortality due to Chronic Respiratory Disease (CRDs)

Figure 4.13 shows the age-standardised death rate from chronic respiratory diseases (CRDs) for the world overall between 1990 and 2021, and projected to 2050. The mortality trend for CRDs differs from what has been observed for diabetes mellitus, where there has been a steady upward climb over time, meaning that the substantial progress in the prevention, management and treatment of CRDs over the past 30 years is evident.

Figure 4.13: Global Cardiovascular diseases Death Rate due to Chronic Respiratory Disease (CRDs) Observed (1990-2021) and Forecast (2022-2050)

Globally, the age-standardised mortality rate for CRDs decreased significantly by almost 48% from 25 deaths per 100 000 population in 1990 to 13 deaths per 100 000 population in 2021. During this period the rate of decrease continued at a high level with no levelling off or trend change over the 30 years (1990 to 2021). The death rate reached its peak between 2000 to 2015. It dropped from about 21 deaths per 100,000 population to about 15 deaths per 100,000 population. This shows that deaths decreased more rapidly during this period. A projection of the trend over the period 2022 to 2050 has indicated that this trend will continue, but at a slower rate of decline. The current death rate of around 13 deaths per 100,000 population is projected to drop to about 11 deaths per 100,000 population by 2050.This projected reduction is still good progress, but the reduced percentage reduction compared to the observed period may be due to a declining proportion of the remaining burden of CRD-related mortality being concentrated among populations with poor access to health care or with ongoing exposure to environmental risk factors. Several converging factors may explain the downward trend in CRD death rates since 1990. Progress to reduce global smoking rates has been a major driver of reductions in mortality, especially in regions with stronger implementation of comprehensive tobacco control policies which have been most successful in high-income countries (GBD Chronic Respiratory Disease Collaborators, 2020).

Another important point is that, the observed decrease in deaths due to CRD has been influenced by better air quality due to environmental legislation, less occupational dust and fume exposure, and better pharmacological control – such as the introduction of bronchodilators and the use of inhaled corticosteroids. (Collaborators, 2020) In addition, the increased access to pulmonary rehabilitation programmes, the better diagnosis of respiratory conditions and the increased availability of supplemental oxygen therapy in low and middle-income countries during this period are likely responsible for the improvement in survival among those with advanced CRDs (Soriano et al., 2020). Furthermore, global household air pollution (HAW) reduction has contributed significantly to the reduction of CRD mortality, especially for women and children in South and Southeast Asia (Stanaway et al., 2018), as the transition to cleaner cooking and heating fuels in many low-income countries has helped to reduce HAW. Although the overall trend is positive, the moderate level of decline towards 2050 reflects the continued presence of important risk factors responsible for maintaining a burden of CRD deaths. Household and ambient air pollution, indoor and outdoor exposures at work, tobacco use, and an increasing urbanization of asthma are important and ongoing policy and public health issues that will need continued attention and investment (GBD Chronic Respiratory Disease Collaborators 2020). In the future, further progress will be required in tobacco control in LMICs, reducing ambient and household air pollution and equipping all regions with access to essential respiratory medicines and health services, if the death toll of CRDs is to continue to decrease.

4.2.3.2 Global Trends in DALYs due to CRDs

Chronic Respiratory Diseases (CRDs) age-standardised DALYs rates are shown as a global rate from 1990 to 2021, and a projection to 2050 shown in Figure 4.14. The global DALYs rate for CRDs mirrors the trend of improving health outcomes for chronic respiratory conditions at the global level, similar to the mortality trends described above, with a substantial and ongoing downward trend over the time period observed.

Figure 4.14: Global Cardiovascular diseases Death Rate due to Chronic Respiratory Disease (CRDs) Observed (1990-2021) and Forecast (2022-2050)

The global age-standardised DALY rate for CRDs fell by almost half from approximately 524 DALYs per 100,000 population in 1990 to ~267 DALYs per 100,000 population in 2021 (a reduction of nearly 49% over the 30 years period observed). This was a steady and gradual run without significant breaks out, reversals or periods of stalling. Between 2005 and 2015, the DALYs rate declined by about 267 DALYs per 100,000 persons, indicating a faster decline in death and disability from CRDs during the later time period. The forecast trend for the period 2022 to 2050 shows a continued decline in the global DALYs rate for CRDs, but at a slower pace than in previous decades. The rate will decrease from approximately 265 DALYs per 100,000 in 2022 to 230 DALYs per 100,000 in 2050. This trend is positive, although the gradual off of the slope of the forecast curve shows that more effort will be needed to further reduce the burden of CRDs since the easiest wins could have already been achieved.

The large decrease in CRD-related DALYs between 1990 and 2021 can be attributed to several convergent trends. The first key contributing trend is the decrease in the prevalence of smoking in the general population since smoking remains the predominant avoidable risk factor for COPD, which causes the largest number of CRD-related DALYs globally (GBD Chronic Respiratory Disease Collaborators, 2020). Several policies such as taxes on tobacco products, restriction on advertisements, smoke-free public areas, and tobacco cessation programs in various countries throughout this period have made significant contributions to decreasing the overall burden of CRDs (WHO, 2023). In addition, better air quality in ambient environments and improved household environments, especially due to decreased use of solid fuels used for cooking and heating purposes in South and Southeast Asia and sub-Saharan Africa, have minimized exposure to environmental risks in those regions (Stanaway et al., 2018). Improvements in the clinical management of CRDs have been another key factor behind the drop in DALYs, as they have made disease control easier, slowed down the progression of diseases, and decreased the amount of disability among people suffering from CRDs. Improved access to inhalable bronchodilators, corticosteroids, and combinations of both for patients diagnosed with COPD and asthma, as well as increased access to lung function rehabilitation and oxygen therapy, have all had a beneficial effect on the overall outcome and contributed to the drop in disability experienced by affected individuals (Soriano et al., 2020). Moreover, greater spending on early detection and monitoring of chronic respiratory conditions in low- and middle-income countries has allowed for more successful management of these diseases.

Although there is a clear trend in the positive direction, the expectation of a more gradual decline towards 2050 highlights some serious issues that still need to be addressed. Constant exposure to cigarette smoke, polluted air, and occupational risks in many LMICs ensures that there is considerable ongoing burden of disability and premature death associated with COPD and related illnesses (Stanaway et al., 2018). Additionally, increasing rates of asthma cases in fast-growing urban environments in connection with allergens, pollutants, and altered microbial environments is an emerging problem that may hinder future reduction in disease burden (Collaborators, 2020). Thus, the expected trend of decreasing disease burden is highly dependent on efforts in controlling smoking habits, reducing air and home pollution, and providing respiratory care improvements to everyone equally.

4.2.3.3 Super-Region Trends in Mortality due to CRDs

Figure 4.15 shows the death rate from respiratory diseases that has been adjusted for the age of the people across different areas of the world known as GBD super-regions from 1990 to 2021. The number of deaths from respiratory diseases went down in all areas during this time. However, the amount that the death rate from respiratory diseases decreased was very different in each area. This means that some areas saw a decrease in deaths from chronic respiratory diseases while others saw a smaller decrease in deaths from chronic respiratory diseases.

Figure 4.15: Cardiovascular diseases Death Rate due to Chronic Respiratory Disease across GBD Super Region (1990-2021)

The place with the deaths over time was Southeast Asia, East Asia and Oceania. The number of deaths per 100,000 people in Southeast Asia, East Asia and Oceania went down a lot from about 66 deaths per 100,000 people in 1990 to 21 deaths per 100,000 people in 2021. This represents a decrease of 68 per cent in Southeast Asia, East Asia and Oceania. The number of deaths really started going down after 2005 which means Southeast Asia, East Asia and Oceania made a lot of progress in controlling deaths from CRD. South Asia always had the highest number of deaths. In the 1990s and early 2000s, the deaths per 100,000 inhabitants in South Asia fluctuated at around 40 deaths per 100,000 individuals. However, it then gradually decreased to reach 32 deaths per 100,000 individuals in 2021. Even though the number of deaths did not go down much as in Southeast Asia, East Asia and Oceania, South Asia still had fewer deaths over time. Sub-Saharan Africa had fairly steady but slowly dropping death rates. In relation to the death rate in the region, there was a fall from about 11 deaths per 100,000 persons in 1990 to approximately 8 deaths per 100,000 persons in 2021, meaning that the deaths reduced by 27%. Despite the fall in the death rate, Sub-Saharan Africa still reported more deaths compared to regions with high income. There was a reduction in the deaths in the region from 11 deaths per 100,000 persons in 1990 to 7 deaths per 100,000 persons in 2021. The deaths remained more or less constant over the years.

There was a decline in deaths in central Europe, eastern Europe, and central Asia. The mortality rate reduced from an estimated 10 deaths per 100,000 individuals in 1990 to 4 deaths per 100,000 individuals, in 2021. It represents a decline of 60 percent, with the greatest reduction occurring after 2000. There was a decline in deaths due to CRD in Latin America and the Caribbean, from about 9 deaths per 100,000 individuals in 1990 to an estimated 4 deaths per 100,000 individuals in 2021. The decline occurred gradually over the years. Wealthy areas maintained their death rates throughout the period under review. The number of deaths per 100,000 individuals reduced from an estimated 6.5 deaths in 1990 to an estimated 3.5 deaths in 2021, representing a 46 percent reduction. Latin America and the Caribbean and other areas experienced deaths higher than wealthy areas. The wealthy areas maintained their death rates throughout the period under review. Deaths due to CRD in areas were lower than other parts of the world. The number of people dying from respiratory diseases is going down around the world. This is because we are getting better at preventing, diagnosing and treating these diseases. We have healthcare systems now and we have more effective treatments. We have stronger public health programs. These things have probably helped to reduce the number of deaths from respiratory diseases in most parts of the world (World Health Organization [WHO] 2023; GBD Chronic Respiratory Disease Collaborators, 2024).

Chronic respiratory diseases are going down fast in Southeast Asia, East Asia and Oceania. This is probably because more people can get healthcare and we have vaccines. We are also getting better at treating infections and we are not exposed to as many bad things in the environment. Some studies have shown that when countries get richer and they have healthcare they can reduce the number of deaths from chronic respiratory diseases (Soriano et al. 2020; WHO, 2023). In South Asia the number of people dying from respiratory diseases is going down but not as fast. This is a problem for public health in this region. People are still breathing air at home and there is a lot of pollution from factories. Many people are. They do not have access to special doctors who treat respiratory diseases. We know that bad air and smoking are making people sick with diseases in South Asia (Salvi & Barnes 2009; GBD Risk Factors Collaborators, 2024). Chronic respiratory diseases are still an issue in South Asia and we need to do something about chronic respiratory diseases. We need to think about respiratory diseases and how they affect people. Chronic respiratory diseases are a concern and we should focus on chronic respiratory diseases. The progress made in Sub-Saharan Africa and North Africa and the Middle East is not very impressive. This is probably because people in these places have a hard time getting the medical care they need. They are also exposed to things in their environment and have to deal with dangers at work. People in these places can get different kinds of breathing problems. In countries where people do not have a lot of money it is difficult for them to get the care they need and this makes it hard to diagnose problems early. (Adeloye et al., 2015; WHO, 2023).

In countries not many people die from breathing problems. This is likely because these countries have been working on their medical care systems for a time. They also have rules to control tobacco and keep the air clean. People in these countries can get the help they need with breathing problems. Other studies have shown that the number of deaths from respiratory diseases in countries (Mannino & Buist 2007; GBD Chronic Respiratory Disease Collaborators, 2024) can be reduced if we can control tobacco and help people manage their diseases. Between 1990 and 2021, the total number of people dying from long term breathing problems has dropped in all parts of the world. However, some places are still doing worse than others. South Asia and Southeast Asia, East Asia and Oceania are still seeing a lot of deaths, from breathing problems. This tells us that we need to come up with plans to help people in these places. We need to reduce the things that can cause breathing problems and make the medical care systems stronger.(GBD Risk Factors Collaborators, 2024; WHO, 2023).

4.2.3.4 SuperRegion Trends in DALYs due to CRDs

Figure 4.16 illustrates the Disability-Adjusted Life Years (DALYs) rates resulting from CRD in the seven GBD super-regions between 1990 and 2021. From the findings, there were notable differences between the regions in the burden of CRDs during the period under consideration. Although DALYs rates declined in most super-regions, there were marked differences in the extent to which rates decreased. The highest DALYs rates were recorded in Southeast Asia, East Asia, and Oceania, whereas the highest burden levels persisted in the high-income super-regions.

Figure 4.16: Cardiovascular diseases Death Rate due to Chronic Respiratory Disease across GBD Super Region (1990-2021)

The Southeast Asia, East Asia and Oceania region was found to have the highest burden of DALYs in 1990 (c. 1150 DALYs per 100,000 population). Nevertheless, there was a marked decrease in the DALYs rate during the subsequent thirty years to 2021, to around 355 DALYs per 100,000 population. The above figure represents a reduction in the DALYs burden of respiratory disease of just under 69%, representing good improvement for respiratory disease in the region. South Asia is the second highest DALYs affected region with a decrease from about 840 DALYs per 100,000 population in 1990 to around 630 DALYs per 100,000 population in 2021, corresponding to a reduction of nearly 25%. The other regions have decreased from 250 DALYs per 100,000 people in 1990 to about 95 DALYs per 100,000 people in 2021. In the High-Income Region, a decrease was seen from 170 DALYs per 100,000 people to about 90 DALYs per 100,000 people. The reduction was slight in Latin America and the Caribbean, where there was a reduction from 195 DALYs per 100,000 people in 1990 to approximately 100 DALYs per 100,000 people in 2021. Slight decline was noted in North Africa & Middle East with DALY rates dropping from 245 to around 160 DALYs per 100,000 population. The decrease is comparatively low in Sub Saharan Africa (270 DALYs per 100 000 people in 1990 to 195 DALYs per 100 000 people in 2012).

However, the implications of the results point towards an advancement globally in the morbidity of chronic respiratory diseases in all super-regions. Nevertheless, some regional differences can be noted, particularly when it comes to high DALYs in South Asia and Southeast Asia, East Asia, and Oceania relative to the High-income super-region. The drop in the DALYs rates of chronic respiratory diseases in the majority of regions could be attributed to the enhanced provision of medical services, disease prevention, early diagnosis, and treatment of chronic respiratory diseases. Further, lower exposures to some environmental and occupational risks, including air pollution indoors and tobacco smoking, also account for better health status regarding the respiratory diseases in different regions across the globe (GBD 2021 Risk Factors Collaborators, 2024). In addition, the significant drop in the DALYs rate in Southeast Asia, East Asia, and Oceania can be attributed to the economic development of those regions, the improvement in the healthcare system, and taking measures to decrease respiratory disease risks.

Despite the significant decline of the rate mentioned, Southeast Asia, East Asia, and Oceania remain the region with the largest DALYs burden rate of all regions in 2021. This situation can be explained by the large population of the region under discussion, exposure to the environmental risk factors including ambient air pollution, occupation hazards, smoking and long-term effects of COPD and asthma (World Health Organization [WHO], 2023). It was found out that air pollution and smoking become the main risk factors for respiratory disability and mortality in developing countries (Soriano et al., 2020). In South Asia, it can be seen that there is a relatively high number of DALYs cases because of environmental and socioeconomic problems. Air pollution within the houses from solid fuels, fast-paced urban development, lack of proper living conditions, and inability to receive healthcare services are some of the most serious factors affecting this problem in South Asian states (GBD Chronic Respiratory Disease Collaborators, 2020). Even though there has been some improvement in the situation in this region, it can be noticed that the rate of DALYs decline is slower than in other regions.

The lowest DALYs were observed in high income regions in the overall time period. It is probable that this can be attributed to better health care services, smoking restrictions in place, better environmental laws in force, along with better provision of interventions. Early diagnosis and management of respiratory disease patients would also have facilitated a better quality of life and disability reduction for these patients (WHO, 2023). A slower decline in Sub-Saharan Africa and North Africa and Middle East may indicate ongoing issues of access to health care facilities and exposure to the environment, and population growth. Many countries in low- and middle-income brackets suffer from poor diagnosis and management of their respiratory diseases, thus causing an ongoing problem of disability (GBD 2021 Risk Factors Collaborators, 2024). The overall finding of this study is that the burden of morbidity from chronic respiratory diseases has decreased significantly in all GBD super-regions in the past 30 years. Concurrently, the high DALYs rates which continue in South and Southeast Asia, East Asia and Oceania regions indicate the need for targeted public health interventions to reduce exposure to risk factors among patients. These include the prevention and control of chronic respiratory conditions, and effective health services.

4.2.3.5 Region Distribution of Mortality and DALYs due to Chronic Respiratory Disease

The Cardiovascular diseases DALYs and mortality rate caused by chronic respiratory diseases among the 21 GBD regions in 2021 are shown in Table 4.3. Some of the geographical differences have been observed with regard to the prevalence of chronic respiratory diseases. According to Table 4.3, in all of the regions, the largest DALYs of chronic respiratory diseases has been registered in Oceania, reporting about 677.4 DALYs per 100,000 population, followed by South Asia, with DALYs of approximately 627.7 per 100,000 population. Another region with a high burden of disease is East Asia, which has 376.7 DALYs per 100,000 population. On the other hand, the least DALYs have been observed in High-Income Asia Pacific (46.6 DALYs per 100,000) and Andean Latin America (52.9 DALYs per 100,00).

Table 4. 3: Region Distribution of Deaths and DALYs Rate

There were also some similarities in mortality rates. The highest mortality rate from chronic respiratory diseases occurred in Oceania, where the death rate was 32.8 deaths per 100,000 people, followed by South Asia and East Asia, which had 31.6 deaths per 100,000 people and 23.1 deaths per 100,000 people, respectively. Higher mortality rates also occurred in Southeast Asia (12.5 deaths per 100,000 people) and Central Sub-Saharan Africa (11.4 deaths per 100,000 people). Contrariwise, the region with the lowest mortality rate is High-Income Asia Pacific, with a mortality rate of about 1.7 deaths per 100,000 people. There were also moderate burden levels in many regions. In North Africa & Middle East, the figures were approximately 159.2 DALYs and 6.9 deaths per 100,000 people, respectively. Similarly, High-Income North America had approximately 153.9 DALYs and 5.3 deaths per 100,000 people. Likewise, burden at intermediate levels was noted for Western Sub-Saharan Africa, Central Asia and Tropical Latin America. On the whole, these findings highlight that there is a higher burden of chronic respiratory diseases in some areas of Oceania, South Asia, East Asia and Sub-Saharan Africa.

These observed discrepancies could be due to variations in exposure to certain major risk factors such as smoking, household air pollution, ambient particulate matter pollution, workplace exposures, and lack of healthcare services access. According to previous literature, there is a high incidence of chronic respiratory diseases like COPD, and others where high levels of air pollution and smoking are observed (GBD Chronic Respiratory Diseases Collaborators, 2024; Soriano et al., 2020). The exceedingly high prevalence rate observed in Oceania and South Asia might be due to their exposure to the environment, lack of access to health services, and high prevalence of respiratory risk factors. In addition, the relatively lower mortality rate and DALYs noted in High-income Asia Pacific, Australasia, and certain parts of Europe can be attributed to the fact that they have efficient health systems, strong tobacco control policies, strict air pollution policies, and adequate management of chronic respiratory conditions. It is, therefore, necessary to put in place measures that would reduce the impact of respiratory risk factors and increase healthcare access in such burdened areas (GBD Risk Factors Collaborators, 2024; World Health Organization, 2024).

4.2.3.5 Country wise Trends in Mortality due to CRDs

The Cardiovascular diseases mortality rates resulting from CRDs in the ten most affected countries between 1990 and 2021 are shown in Figure 4.17. From the results, significant differences have been observed in the mortality trends among the different countries despite a general decrease in mortality from CRDs during the specified period. While several countries had notable reductions in mortality, other countries had stable mortality rates or small declines in mortality. China registered the highest mortality rate for most of the years within the initial part of the study period, while countries like Vanuatu andKiribati had considerably low mortality rates.

Figure 4.17: Cardiovascular diseases Death Rates due to Chronic Respiratory Top 10 Highest-Burden countries

China had the highest mortality rate of 87 per 100,000 people in 1990. Over the next 30 years, death rates plunged, falling to nearly 24 deaths per 100,000 in 2021. The reduction of CRD mortality is estimated to be 72%, which places China among the top countries with the largest fall in CRD mortality during the study period. Nepal also saw a huge drop in mortality rates, from roughly 70 deaths per 100,000 people to 52, a 26% decline. Papua New Guinea’s mortality rates declined from 56 deaths per 100,000 people in 1990 to 44 deaths per 100,000 people in 2021. Also, the Democratic People’s Republic of Korea (DPRK) reduced its mortality rates from nearly 57 deaths per 100,000 in 1990 to around 38 deaths per 100,000 by 2021. Myanmar also had an intermediate decline in mortality rates, from approximately 54 deaths per 100,000 in 1990 to approximately 34 deaths per 100,000 in 2021.The mortality rates in India showed fluctuations during the study period. Mortality rates were fairly stable throughout the 1990s and early 2000s, and have begun to fall since 2010. The mortality rate fell from close to 41 deaths per 100,000 people in 1990 to 34 deaths per 100,000 people in 2021.

Lesotho’s mortality rates followed a similar trend of fluctuation, with mortality rates declining from around 38 deaths per 100,000 people to 26 deaths per 100,000 people. Mortality rates in Vanuatu have also been low and have continued to decline from approximately 35 deaths per 100,000 population in 1990 to nearly 24 deaths per 100,000 population by 2021. There were some variations seen in the case of Pakistan. In the case of Pakistan, the mortality rate rose slightly from the late 1990s up to early 2000s, to about 35 deaths per 100,000 population and further decreased to about 26 deaths per 100,000 population by 2021. However, there were no smooth trends in Kiribati, as the death rate rose from around 24 deaths per 100,000 population in 1990 to around 32 deaths per 100,000 population in the mid-2000s, before declining to around 27 deaths per 100,000 population by 2021. As seen above, all ten countries with the highest burden experienced decreases in their mortality rates due to chronic respiratory diseases between 1990 and 2021.

The decrease seen in China can be credited to the marked advancements made regarding healthcare facilities, disease management, economy, and increased availability of health services in the last few decades. Governments’ work on health programmes related to respiratory diseases, living conditions and household air pollution also could have been a significant factor in this regard (World Health Organization [WHO], 2023). Despite continued concern about ambient air pollution, better diagnosis and treatment methods could probably have decreased the number of premature deaths due to chronic respiratory diseases. The reasons for the observed declines in mortality rates for Nepal, Myanmar and the Democratic People’s Republic of Korea are similar. However, despite all the advances, the mortality rates of these countries remain quite high when compared to others. As previous research shows, inadequate access to healthcare services, exposure to smoke from biomass fuel, occupational dangers, and tobacco use are some of the key factors determining respiratory mortality rates in many LMICs (GBD Chronic Respiratory Disease Collaborators, 2020). It could be because of the constant exposure to environmental risks, rapid population growth, increasing urbanization and other issues regarding the provision of healthcare services in these two countries. Despite these two countries, air pollution, tobacco smoke and other occupational risk factors remain a major threat to the number of deaths due to chronic respiratory diseases (GBD 2021 Risk Factors Collaborators, 2024). High exposure to household air pollution and other environmental pollutants also contribute to the problem of chronic respiratory diseases in India.

Fluctuating trends in Pakistan and Kiribati could indicate that the reduction in respiratory death rates has been somewhat inconsistent in those areas. It may be because of the changes in accessing health care, monitoring the number of cases of such diseases, and demographic shifts, among others. There have been other developing nations where similar trends were observed (Soriano et al., 2020). The general decreasing trend in mortality rates in all high-burden nations demonstrates the benefits brought about by improvements in the field of respiratory healthcare, antismoking policies, and various public health initiatives. However, the comparatively high mortality rate recorded in some countries points out the need for improvement in this regard. Efforts aimed at reducing the incidence of smoking, air pollution, improvement in the health sector, as well as early detection and treatment of chronic respiratory diseases, would be crucial in any attempt to further reduce the mortality rate. The findings from the study stress the significance of having the proper policies in place in the countries suffering from chronic respiratory diseases. While significant gains have been made after 1990, these diseases continue to be among the leading causes of early death in many developing nations.

4.2.3.6 Country wise Trends in DALYs due to CRDs

Figure 4.18 below shows the DALY rates for chronic respiratory diseases (CRDs) among the ten countries which experienced the greatest burden during the period from 1990 to 2021. It can be seen that there was great disparity in the burden of CRDs among countries, despite an apparent downward trend observed in almost all cases. Nepal was consistently ranked first with regard to DALYs for chronic respiratory diseases during the time span under consideration, while the Solomon Islands ranked last. There were also some countries where the prevalence of disabilities and deaths caused by chronic respiratory diseases remained quite high even in 2021.

Figure 4.18: Cardiovascular diseases DALYs Rates due to Chronic Respiratory Top 10 Highest-Burden countries

The highest DALYs ratio was observed in Nepal during 1990, with an estimated value of 1,440 DALYs per 100,000 people. Despite the steady decline in the rate of DALYs, in 2021, the ratio remained high at nearly 1,000 DALYs per 100,000 people. That means the drop is roughly 31 percent. The second highest ratio has been recorded in Papua New Guinea; the number is approximately 1,140 DALYs per 100,000 people in 1990 compared to 870 DALYs per 100,000 in 2021. The respective reduction is close to 24%. In the case of the Democratic People’s Republic of Korea, the reduction in DALYs rates was even more pronounced from almost 1,020 DALYs per 100,000 people in 1990 to 700 DALYs per 100,000 people in 2021. On the other hand, in India, the decrease happened from 860 DALYs per 100,000 to just over 660 DALYs per 100,000. Generally speaking, the disease burden caused by CRDs is still substantial in the aforementioned countries even though there have been some reductions. As for Myanmar, there were some changes during the specified period of time. DALYs rates stayed stable throughout the decade of 1990, however, afterwards they started decreasing consistently. The burden decreased from approximately 1,080 DALYs per 100,000 people in 1990 to just below 620 DALYs per 100,000 people in 2021, which is a 43% drop. Similar tendencies could be observed in Kiribati when DALYs rates decreased from about 810 DALYs per 100,000 people in 1990 to roughly 580 DALYs per 100,000 people in 2021.

For Lesotho, the DALYs burden rates had a unique trend pattern compared to other nations in the world. The DALYs rates remained more or less constant in the 1990s but suddenly shot upwards at the beginning of the next decade. The burden rates hit around 740 DALYs per 100,000 people in 2005, after which it gradually came down to just above 640 DALYs per 100,000 people in 2021. Even though there was a drop in the DALYs rates, they still remained higher than what was observed earlier in the study period. Vanuatu experienced a moderate decline in the DALYs rates as the DALYs dropped from around 730 DALYs per 100,000 people in 1990 to about 510 DALYs per 100,000 people in 2021. Pakistan showed a small decline in DALYs as DALYs rates dropped from approximately 640 DALYs per 100,000 people to about 490 DALYs per 100,000 people. The Solomon Islands had been the country with the lowest DALYs burden in all three countries for all the years selected, as DALYs decreased from around 590 DALYs per 100,000 people in 1990 to about 460 DALYs per 100,000 people in 2021. It can therefore be concluded that there have been huge decreases in DALYs due to CRD in the majority of the countries with a high burden between 1990 and 2021.

The remarkably high rates of DALYs noted for Nepal during the entire study period could be due to continuous exposure to important risk factors for respiratory diseases such as household pollution, smoking, occupation, and inadequate provision of specialized respiratory care facilities. The use of solid fuel in cooking and heating is known to be one of the major contributing factors to the burden of chronic respiratory diseases in many South Asian countries (GBD Chronic Respiratory Disease Collaborators, 2020). In the same way, high numbers of DALYs experienced in Papua New Guinea, the Democratic People’s Republic of Korea, and India may be attributed to the continuous presence of socio-environmental risk factors in these places. Risks such as air pollution, smoke from biomass, population increase, and uneven provision of health care services affect the prevalence and death rates related to respiratory disorders in these countries (GBD 2021 Risk Factors Collaborators, 2024). Regarding the case of India, the increasing risks of exposure to respiratory hazards because of industrialization and urbanization can be noted.

Lastly, the peculiar increase in DALYs rate in Lesotho within the first years of the twenty-first century may be attributed to the interaction between chronic respiratory disease and other health problems like infections, poverty, and poor access to health care facilities. In fact, past studies have found out that nations struggling with multiple health challenges tend to lag behind in managing chronic diseases (Soriano et al., 2020). The steady decrease in DALYs in Vanuatu, Pakistan, Kiribati, and the Solomon Islands could signify improved outcomes regarding the health condition of people in the past three decades. Nevertheless, the constant occurrence of moderate DALYs points out that respiratory diseases are a significant issue from a public health perspective and exposure to certain elements, such as tobacco smoke, can hinder any further improvements in the future. Generally, the reduction of DALYs in all selected countries can be viewed as a good sign and can be regarded as progress in solving the given issue. Nevertheless, the substantial burden that is placed on numerous countries emphasizes the need for making additional efforts to alleviate the impact of risks, to maintain the quality of the air, and to provide timely health care to people.

4.3 Data-Driven Cluster of Countries from NCDs

In this subsection, the K-means algorithm was implemented for the CVDs, diabetes mellitus and CRDs dataset with the objective of seeking patterns concerning death and DALYs based on the selected countries. In detailing the distinct burdens of NCDs from the death and health perspectives, the focus was on the cluster analysis for the deaths and DALYs variables separately.

4.3.1 Data-Driven Clusters Based on Cardiovascular diseases Death Rates for Cardiovascular disease (CVDs)

Figure 4.19 shows the Cardiovascular diseases death rates for CVDs for countries clustered by this indicator. The standardized mortality data were standardized and clustered using the K-means algorithm, which yielded four distinct clusters. The figure shows that there is a clear separation between clusters based on relative burden of CVD mortality. The majority of countries with standardized values below the mean are countries with comparatively low Cardiovascular diseases death rates. The trends in these countries were relatively homogeneous, suggesting better cardiovascular health outcomes or more effective disease prevention or management. Countries in Cluster 0 (red) have moderate mortality burden of CVDs. The values in the standardised scale are more or less clustered around the mean of the population with little variation, indicating that there are intermediate levels of cardiovascular mortality in different geographic areas. The countries with relatively high standardized death rates are in cluster 3 (purple). This cluster has fewer countries, but there’s a consistent pattern of high mortality, which suggests high levels of Cardiovascular disease.

Figure 4.19: Cluster Based on Cardiovascular diseases Death’s rate for Cardiovascular disease

Mortality-based clustering

• Cluster 0: 65 countries
• Cluster 1: 5 countries
• Cluster 2: 95 countries
• Cluster 3: 39 countries

The highest-risk group (green) contains very few countries with very high death rates and is included only if the standardized death rate is very high. The clustering solution shows these countries as clear outliers because they have a very high load of cardiovascular mortality among all the clusters. The distance between Cluster 1 and the other clusters highlights the performance of the data-driven clustering approach in clustering countries that have higher than usual CVD mortality rates. Overall, the clustering pattern suggests that there is a high level of heterogeneity in the mortality rates for Cardiovascular diseases across countries. Countries were not classified by geography, but rather by similarities in the mortality profile, reflecting the value of using data-driven methods for defining populations with similar disease burdens. These findings align with previous  (GBD) studies, which showed significant cross-national differences in cardiovascular mortality attributable to demographic characteristics, access to health services, behavioral risk factors and socioeconomic status (GBD 2021 Diseases and Injuries Collaborators, 2024; Roth et al., 2020).

The full classification of 204 countries is shown in Appendix Table A.1, based on Cardiovascular diseases rates of Cardiovascular disease deaths. The clustering results show that there are considerable differences between countries in mortality burden. Cluster 0 comprises countries with moderate levels of cardiovascular mortality burden, between 307.843 and 551.828 deaths per 100,000 population in Nepal and Liberia respectively. Within this group are several countries in South Asia, Africa, the Middle East and Eastern Europe, which present similar mortality patterns despite their geographic difference. The five countries in Cluster 1 are Nauru, Marshall Islands, Tuvalu, Solomon Islands and Micronesia (Federated States of). This cluster has a cardiovascular mortality burden of 1,258.546 to 2,241.439 deaths per 100,000 population, the highest of any cluster. The fact that these countries are all Pacific Island nations suggests that they have some very similar mortality attributes which set them apart from all of the other countries analysed.

Countries in Cluster 2 have the lowest cardiovascular mortality rates and are also the largest. The death rates range from 73.571 deaths per 100,000 population in Israel to 305.643 deaths per 100,000 population in Gabon. Many high-income nations, including Japan, France, Australia, Switzerland, Norway, Canada, Republic of Korea, and Netherlands, fall into this cluster, and have relatively low mortality rates which could relate to better health systems, effective prevention programmes and improved management of cardiovascular risk factors. The countries in cluster 3 have very high cardiovascular mortality rates, ranging from 571.243 deaths per 100,000 population in the Bahamas to 1,002.843 deaths per 100,000 population in Egypt. The range of countries in this cluster includes those in all the different GBD regions, suggesting that higher cardiovascular mortality is not limited to one geographical region, but is present in several countries with similar epidemiological characteristics. Overall, Appendix Table A.1 shows that countries with comparable cardiovascular mortality burdens are spread across the continents and GBD regions. This discovery indicated that data-driven clustering is a better method to identify similarities of disease burden than traditional geographic classification. The findings echo earlier GBD analyses that found that shared demographic, behavioural, metabolic and health-care related factors are key drivers of cardiovascular mortality, not just geographic location (Roth et al., 2020; GBD 2021 Diseases and Injuries Collaborators, 2024).

4.3.1.1 Comparison of data-driven clustering and GBD Regions

Table 4.4 compares the highest- and lowest-burden countries in each of the clusters of data-driven countries with the GBD sub-regions and super regions. The comparison shows that countries in the same cluster do not have to be located in the same geographic area. Rather, the clustering algorithm categorized countries by their similarity in Cardiovascular diseases rates of Cardiovascular disease (CVD) deaths, without taking into account their geographic location. The highest death rate in Cluster 0 was observed in Palestine while the lowest death rate in Cluster 0 was observed in Albania. Both countries are in different sub-regions of GBD and super regions, but were grouped together due to similar overall characteristics of the CVD mortality profile after normalisation. Countries across different geographical areas can have similar disease burden patterns as well, such as Palestine (North Africa and Middle East) and Albania (Central Europe).

Table 4. 4: Comparison of data-driven clustering and GBD regions

Micronesia (Federated States of) and Northern Mariana Islands are both part of the sub-region Oceania of the super region Southeast Asia, East Asia and Oceania for Cluster 1. MONIA had a significantly higher death rate than Northern Mariana Islands, despite being in the same geographical classification. The cluster mainly consists of island countries with relatively high levels of cardiovascular mortality. Cluster 2 is for Guinea, the region is Western Sub-Saharan Africa and the death rate is the highest, while Israel is in the North Africa and Middle East region, with the lowest death rate. The clustering method captures the statistical similarity in disease burden, as observed through their clustering together as opposed to geographic proximity. This discovery indicates that countries in different continents could have similar epidemiological profiles in different health systems and socioeconomic situations. Likewise, Cluster 3 includes Nauru with the highest death rate, and Egypt with the lowest death rate within the cluster. Nauru is part of the Oceania sub-region and Egypt is part of the North Africa and Middle East region.

Both are located far away from the other countries, but both are in the highest-burden cluster with exceptionally high levels of cardiovascular diseases and cardiovascular mortality compared with most other countries. In general, the comparison shows that the data-driven clustering method is not truly based on the GBD region boundaries. Rather, it describes countries that share comparable disease burden characteristics in various regions of the world. This illustrates the benefit of data-driven clustering over traditional regionalization based on geography by highlighting epidemiologically similar groups that are hidden when countries are grouped by geographic location. The patterns of disease burden observed in previous studies based on GBDs have been found to be influenced not only by geographic location, but also by common risk factors and shared demographic characteristics, lifestyle factors, and access to health services (Murray et al., 2020; GBD 2021 Diseases and Injuries Collaborators, 2024; Roth et al., 2020).

4.3.2 Data-Driven Clusters Based on Cardiovascular diseases DALYs Rates for Cardiovascular disease cluster (CVDs)

The data driven clustering of 204 countries by age-standardised DALYs rates for Cardiovascular diseases (CVDs) is presented in Figure 4.20. There were four separate clusters, each with significant differences in disease burden among countries. The clustering algorithm clustered countries by their similarity in DALYs rates, not based on their geographical location, meaning that countries with similar disease burden could be clustered in different GBD regions. Cluster 1 (green) are countries with the highest rates of standardised DALYs. This cluster has a fairly small number of countries but is clearly distinct from the other clusters suggesting an extremely high burden of Cardiovascular disease in this cluster. This high-burden cluster includes countries like Micronesia (Federated States of), Afghanistan, Syrian Arab Republic, Turkmenistan, Uzbekistan, Iraq, Sudan, Fiji, Libya, Belarus, Ukraine, Tajikistan and Vanuatu. These countries have much higher DALY rates, indicating a heavy burden of premature death and disability from Cardiovascular disease. Cardiovascular diseases have been recognized as a major contributor to disability and premature death in the  studies, especially in low- and middle-income countries that are undergoing socioeconomic or political instability, have a high burden of metabolic risk factors and have limited health service capacity (Roth et al., 2020; GBD 2021 Diseases and Injuries Collaborators, 2024).

Another group with a high burden (Cluster 3) had relatively lower DALYs rates than Cluster 1. This cluster contains countries that are still suffering a high burden of Cardiovascular disease, including Nauru, Marshall Islands, Tuvalu, Solomon Islands and Egypt. The widespread obesity, diabetes, hypertension, unhealthy dietary habits and the lack of healthcare capacity are some of the reasons several Pacific Island countries are among the highest burdened world over. This result is consistent with other global epidemiological research which has shown that Oceania has one of the highest burdens of Cardiovascular diseases (CVDs) in the world (Murray et al., 2020). Countries in Cluster 0 (red) have medium to high DALYs rates with a high percentage from developing countries spread across Asia, Africa, Latin America and Eastern Europe. Representative countries include Palestine, Philippines, Republic of Moldova, Lithuania, Algeria, Pakistan, India, Malaysia, South Africa, Zimbabwe and Paraguay. These countries are different geographically, but have similar levels of cardiovascular disability burden. The clustering illustrates that the similarities in epidemiology are due more to disease burden and risk factors associated with the disease than to geographical location.

The largest number of countries are in cluster 2 (blue) which has the lowest standardized DALYs rates. The cluster includes most high-income countries including France, Japan, Republic of Korea, Australia, Switzerland, Netherlands, Norway, Belgium, Canada, Denmark, Spain and Israel. The countries in this group typically have good health care systems, good preventive measures, early disease detection, and better management of cardiovascular risk factors, which leads to relatively low DALYs rates (Mensah et al., 2018). The overall pattern in Figure 4.3.2 is clear-cut separation between high, moderate and low burden countries. The data-driven clustering was able to classify countries with similar cardiovascular DALYs profiles without regard to geographic location, which is not true for traditional regional classifications. This showcases the value of using machine learning clustering to define clusters of epidemiologically similar populations and to move towards the goal of implementing regionally defined clusters for global health evaluation.

Figure 4.20: Cluster Based on Cardiovascular diseases DLAYs Rates for Cardiovascular disease

DALYs-based clustering

• Cluster 0: 75 countries
• Cluster 1: 34 countries
• Cluster 2: 90 countries
• Cluster 3: 5 countries

The complete list of 204 countries grouped into four clusters based on Cardiovascular diseases DALYs rates for Cardiovascular diseases is given in Appendix Table A.2. The appendix builds on Figure 4.20 by showing the exact country breakdown for each cluster and the significant variation in the burden of Cardiovascular disease around the world.The countries in Cluster 0 have moderate-to-high DALYs rates, ranging from 25.082 DALYs in Palestine to 13.541 DALYs in Albania. The cluster contains countries from various continents: Asia, Africa, Europe, Latin America and Caribbean. Similar burden of Cardiovascular diseases is observed in geographically diverse countries like Pakistan, India, Kazakhstan, Algeria, Lithuania, Malaysia, Zimbabwe, South Africa, Paraguay and Cuba, etc. This observation implies that a common demographic and metabolic risk factor is more important than geographical ones for the disease burden.

The most heavily burdened countries in Cluster 1 have DALYs rates between 47.354 in Micronesia (Federated States of) and 26.035 in Northern Mariana Islands. Most of the countries in this cluster are in conflict areas, Oceania, the Middle East and Central Asia, including Afghanistan, Syria, Iraq, Sudan, Yemen, Ukraine, Belarus, Tajikistan, and Papua New Guinea. The clustering of countries with high levels of cardiovascular disability burden due to political instability, limited healthcare infrastructure, and high cardiovascular risk factors confirms previous findings that these factors are significant contributors to cardiovascular disability burden (Roth et al., 2020).The countries in Cluster 2 have moderately low rates of DALYs, ranging from 13.473 in Guinea to 3.431 in Israel. This cluster is home to the largest number of countries and many developed countries including Japan, France, Australia, Switzerland, Canada, Denmark, Netherlands, Norway, Belgium, Republic of Korea and Singapore. These lower cardiovascular DALYs in these countries may be related to better health care systems, better cardiovascular risk factor prevention programmes and better management of cardiovascular risk factors.

Only five countries are in Cluster 3 (Nauru, Marshall Islands, Tuvalu, Solomon Islands and Egypt), where DALYs rates range between 77.789 and 49.931. This cluster is small but includes the countries with very high cardiovascular disability burden. The high proportion of Pacific Island countries in this cluster is consistent with the high prevalence of states of obesity, diabetes, hypertension and poor food choices affecting cardiovascular health, as well as the fact that this cluster also has elevated rates of cardiovascular mortality. The burden is also high in Egypt, underscoring the fact that despite varying socioeconomic conditions, different geographic regions can have similar epidemiological conditions.In general, Appendix Table A.2 shows that countries with similar cardiovascular DALYs profiles can be found in the same GBD clusters across GBD regions. These findings also confirm that data driven clustering is a good approach to identify epidemiologically similar countries and set the groundwork for the subsequent comparison with GBD regional groupings.

4.3.2.1 Comparison of data-driven clustering and GBD Regions

Table 4.5 shows some comparisons between the representative countries of each of the clusters and their respective GBD sub-regions and super-regions to test if the data-driven clusters correspond to traditional geographic classifications. Liberia has the highest DALYs rate in Cluster 0 while Nepal has the lowest DALYs rate in the South Asia region among the countries in the Sub-Saharan Africa super-region. While both countries are clustered together based on information in the data, they are in different GBD regions. This illustrates that the cardiovascular burden of the countries can be different, but from the same geographical background.

Table 4. 5: Comparison of data-driven clustering and GBD Regions

Nauru and Federated States of Micronesia both are located in Cluster 1 in the Southeast Asia, East Asia and Oceania super-region in Oceania. They are clustered in the same cluster due to their geographic proximity and similarly high cardiovascular DALYs burden burden, which indicates that the clustering result is consistent with the conventional classification of the GBD. Gabon is in Cluster 2, the highest DALYs rate, and Israel is in Cluster 4, the lowest DALYs rate, both of which are part of different sub-regions in the DALYs pyramid.Within Cluster 2 (highest DALYs rate), there are several countries in Central Sub-Saharan Africa, including Gabon, and within Cluster 4 (lowest DALYs rate), there are several countries in the North Africa and Middle East region, including Israel. Although the countries have different GBD super-regions, both countries are in the same low-burden cluster, indicating that the clustering algorithm sorts countries by disease burden and not by geographic location.

Likewise, the regions of North Africa and Middle East (Cluster 3) include Egypt in Latin America and Caribbean, while Caribbean (Cluster 3) includes Bahamas. These countries are neither geographically close to each other, nor in the same highest-burden cluster, but have similar levels of cardiovascular DALYs. The general impression from the comparison is that data-driven clustering does not necessarily stick to the traditional GBD regional pattern. Some clusters (especially Cluster 1) are mostly defined by a stated geographic region; others are a mixture of countries across GBD sub-regions and super-regions. The findings show that disease burden patterns are not only determined by spatial proximity but also by common epidemiological properties and risk factor profiles, suggesting that data-driven clustering could be a valuable tool to identify countries with similar Cardiovascular disease burdens that may be more appropriately targeted by similar public health interventions (GBD 2021 Diseases and Injuries Collaborators, 2024; Murray et al., 2024).

4.3.3 Data-Driven Cluster Based on Cardiovascular diseases Death Rates for Diabetes Mellitus

The clusters of countries were identified based on the Cardiovascular diseases death rates for diabetes mellitus shown in Figure 4.21. The clustering algorithm was able to successfully cluster countries based on similarities in diabetes mortality without being based on their geographical location, indicating there was a high level of heterogeneity in the diabetes burden of death globally. Three countries (Fiji 69.009, Kiribati 49.595, Marshall Islands 43.881) have the highest burden of diabetes mortality, Cluster 3. These countries are all in the Oceania region and have very high ASDRs compared to all other clusters. These increases in mortality burden may be attributed to the high prevalence of obesity, diabetes complications, and delayed diagnosis, among other factors, which have been reported in many Pacific Island countries (GBD 2021 Diabetes Collaborators, 2024; International Diabetes Federation [IDF], 2025).

Figure 4.21: Cluster based on Cardiovascular diseases Death’s rate for Diabetes Mellitus

Mortality-based clustering

• Cluster 0: 68 countries
• Cluster 1: 111 countries
• Cluster 2: 22 countries
• Cluster 3: 3 countries

Cluster 2 is the second highest mortality burden ranging from 19.299 to 38.192. This cluster includes countries such as Nauru, Bahrain, Micronesia (Federated States of), Niue, Solomon Islands, American Samoa, Tonga, Eswatini, Cook Islands, Palau, Mauritius, Tuvalu, Samoa, Trinidad and Tobago, Lesotho, Tokelau, Papua New Guinea, Vanuatu, Qatar, Guyana, Grenada, and South Africa. Within this cluster there are several African, Caribbean and Middle Eastern countries, though many countries are located in Oceania. It indicates that there is an association between increased diabetes mortality and common epidemiological features, not just proximity. The moderate mortality group (cluster 0) includes countries with mortality rates between roughly 8.144 and 18.033. Dominica, Haiti, Namibia, Gabon, Mexico, Egypt, Pakistan, Palestine, Zimbabwe, Guatemala, Iraq, Bangladesh, Kenya, Nigeria, Ethiopia and other countries scattered throughout Africa, Asia, Latin America and the Caribbean. The widespread distribution is an indication that moderate diabetes mortality is similar across countries with different socioeconomic development and disease burden. The lowest burden of diabetes mortality is in Cluster 1, which also contains the highest number of countries. Countries with death rates between 0.446 and 8.121 include Japan, Singapore, Finland, Monaco, Switzerland, Norway, Canada, Australia, France, Germany, Republic of Korea, the United Kingdom, China, India, Brazil and other European countries. This cluster is dominated by high-income countries, indicating that these countries have more robust healthcare systems, better glycaemic control, better early diagnosis, effective diabetes management programmes and more accessibility to prevention health services (IDF, 2025; GBD 2021 Diseases and Injuries Collaborators, 2024). In general, Figure 4.3.3 shows that the mortality rate due to diabetes is highly heterogeneous among countries and that the countries can be clustered into different groups that share similar mortality profiles, regardless of their geographic location. Based on these results, common metabolic risk factors, healthcare access and socioeconomic factors seem to be more determinants of diabetes mortality than traditional regional classification.

All 204 countries are classified into 4 groups according to their Cardiovascular diseases diabetes death rate in appendix table A.3. The numerical data per country is provided in the appendix to complement the graphical presentation and to allow for a more detailed understanding of the patterns of mortality at country level. The countries in the moderate diabetes mortality cluster (0) include Dominica (18.033) to Kenya (8.144). The countries come from a variety of GBD regions: Sub-Saharan Africa, North Africa and the Middle East, South Asia, Southeast Asia, Latin America and the Caribbean, and Oceania. These include Pakistan, Egypt, Palestine, Mexico, Zimbabwe, Bangladesh, Ethiopia and Nigeria. This diversity helps to confirm that countries that have similar diabetes mortality rates occur in varying geographical and socioeconomic contexts. The countries in cluster 1 have the highest number of countries and the lowest mortality burden. The death rates vary from the most to the least in the United Republic of Tanzania (8.121) to Singapore (0.446). This cluster contains a high percentage of European nations as well as Australia, Canada, Japan, Republic of Korea, New Zealand, the U.S. and a number of middle income nations including India, Brazil, Thailand, Malaysia and China. This cluster is dominated by developed healthcare systems, which is a result of better diabetes prevention, early diagnosis and clinical management.

Countries from Nauru (38.192) to South Africa (19.299) are in Cluster 2 because they have high deaths. Most of the countries are located in Oceania, such as Micronesia, Niue, Solomon Islands, American Samoa, Tuvalu, Samoa, Tokelau, Papua New Guinea and Vanuatu, but a few African, Caribbean, and Middle Eastern countries are also included. This is in line with previous studies that found Pacific Island countries have some of the highest diabetes burdens in the world. The countries with the highest diabetes mortality rates from the data set are found in cluster 3, which includes Fiji (69.009), Kiribati (49.595) and Marshall Islands (43.881). Their clustering as an independent cluster reflects the particularly high burden compared to all other countries. Overall, Appendix Table A.3 shows that there is a large variation in diabetes deaths between countries, and also illustrates that countries with similar mortality patterns tend to be in different regions of the GBD. The results indicate the usefulness of data-driven clustering for identifying countries with similar epidemiological features other than geographical classification. There were four mortality patterns identified by the clustering analysis of diabetes mellitus. The highest mortality burden was found in Cluster 3, which comprised only Pacific Island countries, while countries with similar high mortality rates were spread across Oceania, Africa, Caribbean and Middle East, and were part of Cluster 2. The countries in Cluster 0 had moderate mortality rates, and Cluster 1 included most countries with comparatively low diabetes death rates, particularly high-income countries. The results suggest that the geographical effects on diabetes mortality are more closely related to the shared metabolic risk factors and healthcare access and socioeconomic factors than to mere geographical location. The findings also show that data-driven clustering is a more meaningful approach to grouping countries by diabetes mortality profiles than the traditional regional approach.

4.3.3.1: Comparison of data-driven clustering and GBD Regions

When compared to the geographical classification in GBD, it is clear that the main difference between the two is that the clusters derive from similarity in diabetes mortality as opposed to geographical location that is clearly shown in table 4.6. Cluster 0 contains Dominica (18.033) which has the highest mortality in the cluster from the Caribbean (Latin America and Caribbean super-region), and Kenya (8.144) which has the lowest mortality in the cluster from Eastern Sub-Saharan Africa (Sub-Saharan Africa super-region). The mortality profiles of both countries are very similar, despite their different geographical location and are grouped in the same cluster.

Table 4. 6: Comparison of data-driven clustering and GBD Regions

The countries in Cluster 1 include the United Republic of Tanzania (8.121) in Eastern Sub-Saharan Africa and Singapore (0.446) in Southeast Asia. This illustrates that there can be similar patterns of diabetes mortality across countries with very different socioeconomic and regional composition. Nauru (38.192) from Oceania is the country with the highest mortality, and South Africa (19.299) from Southern Sub-Saharan Africa is the with the lowest. These countries are in different super-regions of the GBD, but have similar high mortality burdens. Fiji (69.010) and Marshall Islands (43.881) are the only two countries in Cluster 3, both in the Southeast Asia, East Asia and Oceania super-region in Oceania. This highlights the strong clustering of the highest diabetes mortality in Pacific Island populations. Overall, this comparison suggests that the clustering result from the data-driven approach better reflects the similarity in disease burden when compared to the traditional geographical classification method. Diabetes mortality profiles can be similar for countries from different GBD regions, but neighbouring countries can be in completely different clusters based on their epidemiological characteristics.The International Diabetes Federation (2025) also notes that Pacific Island countries are considered diabetes hotspots despite having similar classification as countries which have significantly lower disease burden. This further confirms the value of clustering based on machine learning algorithms for the identification of population clusters that are epidemiologically similar, not just geographically.

4.3.4 Data-Driven Cluster Based on Cardiovascular diseases DALYs Rates for Diabetes Mellitus

The distribution of countries is shown by Cardiovascular diseases DALYs rate for diabetes mellitus into four data-driven clusters in Figure 4.22. There was a clear separation by clustering which indicated there was significant variation in the global distribution of disability and premature mortality due to diabetes. The countries in Cluster 1 (green) are the ones with the highest burden of diabetes, and the countries in this cluster include just four countries from the Pacific Islands: Fiji, Marshall Islands, Kiribati and Nauru. The DALYs rates are significantly higher than all other clusters with Fiji having the highest DALYs rate per 100,000 population (1,940.37), followed by Marshall Islands (1,562.17), Kiribati (1,550.32) and Nauru (1,316.31). The distinct pattern of distribution of these countries from the rest suggests that diabetes is a very high burden in Pacific Islanders.

Figure 4.22: Cluster Based on Cardiovascular diseases DALYs rates for Diabetes mellitus

DALYs-based clustering Results

• Cluster 0: 63 countries
• Cluster 1: 4 countries
• Cluster 2: 118 countries
• Cluster 3: 19 countries

The countries in Cluster 3 (purple) also have relatively high rates of DALYs but lower than those in Cluster 1. The majority of countries in this cluster are small island and Oceania countries such as American Samoa, Micronesia, Niue, Cook Islands, Tonga, Samoa, Solomon Islands, Palau, Tokelau, Tuvalu and Vanuatu. However, there are also a number of non-island countries, such as Bahrain, Mauritius, Guyana, Eswatini, Qatar, Grenada, and Trinidad and Tobago, which fall into this cluster, indicating that high burden of diabetes is not solely a product of geographic isolation but also linked to metabolic risk factors and lifestyle determinants. Countries with moderate-to-high DALYs rates are in Cluster 0 (red). This cluster comprises numerous countries from Sub-Saharan Africa, the Caribbean, the Middle East, South Asia and Latin America. The countries in these clusters have a significant diabetes burden, but overall rates of DALYs are much lower than in Clusters 1 and 3. The intermediate burden is seen in countries like Lesotho, Haiti, Dominica, Mexico, Iraq, Pakistan and South Africa. Cluster 2 (blue) shows the highest number of countries, and is the lowest burden of diabetes DALYs of all clusters. This cluster consists of a large number of high income countries from Europe as well as some countries from East Asia and a few middle income countries such as France, Ireland, Austria, Norway, Denmark, Japan, Republic of Korea, Australia, Canada and the United Kingdom. The small size of this cluster indicates the relative low level of disability due to diabetes, perhaps because of better health systems, earlier detection of diabetes, better glycaemic control, and better prevention measures.

The overall pattern of diabetes burden is clearly gradient by country in Figure 4.3.4. The pattern of clustering shows that PIOs have the highest diabetes burden while most high-income countries are aggregated in the lowest burden cluster. These results are similar to previous  studies, which found that diabetes burden is disproportionately high in Oceania and comparatively low Cardiovascular diseases DALYs rates in high-income countries because of better healthcare systems and more effective diabetes benefit management (GBD 2021 Diabetes Collaborators, 2024; World Health Organization [WHO], 2024).The countries allocated to the four diabetes DALYs clusters can be found on Appendix Table A.4 along with their corresponding GBD sub-region and super-region allocations. From the above table, the clustering approach successfully grouped countries with similar diabetes burden regardless of their geographical location.

Cluster 0 is a heterogeneous combination of countries from Southern and Eastern Sub-Saharan Africa, the Caribbean, Latin America, the Middle East, South Asia and Southeast Asia. Although they are quite geographically diverse, these countries have similar intermediate DALYs rates for diabetes between about 337 and 655 per 100,000 population. The smallest cluster but the highest risk is Cluster 1, which includes only four Pacific Island countries. The exclusive composition of Fiji, Marshall Islands, Kiribati and Nauru confirms the exceptionally severe diabetes burden affecting Pacific Island populations, in keeping with previous epidemiological evidence identifying Oceania as one of the world’s highest diabetes burden regions.

Cluster 2 has the highest number of countries from all super-regions across GBD and includes countries from Europe, Asia, Sub-Saharan Africa, Latin America, and North America. The majority of high-income countries fall into this particular cluster due to the relatively lower rates of diabetes DALYs. The geographic diversity of the cluster indicates that low diabetes prevalence is related more to socioeconomic progress and healthcare rather than geographic location.The third cluster is largely made up of countries from the Pacific Islands region along with some Caribbean and Middle Eastern countries. Despite being spread out geographically, these countries display high levels of diabetes DALYs which shows how data-driven clustering can be used to determine similarities between countries.

4.3.3.2: Comparison of data-driven clustering and GBD Regions

Table 4.7 presents the list of representative countries using the data-based clustering technique in relation to their corresponding GBD sub-region and super-region. It is clear from the results that countries clustered into the same clusters do not have to be located in one geographical region. Rather, the countries were categorized based on similarities in the age-adjusted diabetes DALYs rate and not their geographical locations.

Table 4. 7: Comparison of data-driven clustering and GBD Regions

Cluster 0 contains countries with a moderately high level of diabetes DALYs rate. Among these countries, Lesotho had the highest DALYs rate (654.683) and is part of the Southern Sub-Saharan Africa sub-region of the Sub-Saharan Africa super-region. On the other hand, Bolivia (Plurinational State of) had the lowest DALYs rate (337.201) and is part of the Andean Latin America sub-region of the Latin America and Caribbean super-region. Notwithstanding their continental differences and the GBD regions that they are associated with, the two countries had a sufficiently similar profile for the disease burden to be placed in the same cluster.It implies that factors influencing diabetes burden are similar rather than being determined by geographical location alone. Cluster 1 comprised only countries from Oceania. Among these, Fiji had the highest DALYs rate (1940.367), while Nauru had the lowest DALYs rate (1316.314) within the cluster. The two countries were from the Southeast Asia, East Asia, and Oceania super-region. Their extremely high DALYs rate reflects the ongoing diabetes burden in Pacific Island nations caused by factors such as obesity, dietary changes, lack of healthcare, and high metabolic risk factors (GBD 2021 Risk Factors Collaborators, 2024; International Diabetes Federation [IDF], 2025).

The second cluster had moderately high rates of diabetes DALYs and consisted of countries belonging to various GBD regions. South Sudan was the country with the highest rate of DALYs (324.673), while it belonged to the sub-region of Eastern Sub-Saharan Africa; on the other hand, Belarus had the lowest rate of DALYs (96.178), which belonged to the sub-region of Eastern Europe in the super-region of Central Europe, Eastern Europe, and Central Asia.The third cluster had high presence of nations and population islands of Oceania that had extremely high burden of diabetes. American Samoa had the highest DALYs rate (1170.747) among all while Grenada showed the lowest DALYs rate (679.530). Despite Grenada being part of the Caribbean sub-region of Latin America and the Caribbean super-region, it still belonged to this cluster due to the similarity in diabetes burden with the Pacific Island nations.

From the above comparison, it is clear that the cluster analysis method provides a complement to the traditional classification of GBD regions regarding the formation of groups based on the similarities in the burden of diabetes within the countries as opposed to geographical locations. Even as the GBD regions continue to be relevant in the reportage of regional health issues, the new clusters provide evidence of epidemiologic connections among nations which might not have been obvious before. This further shows the relevance of using machine learning to determine the similarities in the burden of diseases among countries.

4.3.5 Data-Driven Cluster Based on Cardiovascular diseases Death Rates for Chronic Respiratory Disease

Four clusters have been identified based on the age-standardized death rate of chronic respiratory diseases (CRDs), which is shown in Figure 4.23. The clustering process shows the extent of heterogeneity regarding mortality due to chronic respiratory diseases among the 204 countries considered for this study. Similarities in mortality rate led to the grouping of countries, regardless of their geographic region, implying that the clustering method was more concerned with the similarity of the burden of diseases than geographic location.

Figure 4.23: Cluster Based on Cardiovascular diseases Deaths rates for Chronic respiratory

Mortality-based clustering

• Cluster 0: 26 countries
• Cluster 1: 115 countries
• Cluster 2: 5 countries
• Cluster 3: 58 countries

Countries belonging to Cluster 0 are those that show higher Cardiovascular diseases mortality rates due to chronic respiratory diseases. According to data presented in Appendix Table A.5, Kiribati had the highest Cardiovascular diseases death rate (27.633), Lesotho (26.265), Vanuatu (24.039), and Pakistan (23.917). These countries included China, Bangladesh, Bhutan, Cambodia, Afghanistan, Indonesia, and some countries in the Pacific Islands. These countries are grouped together to show where exposure to important risk factors for respiratory diseases such as tobacco use, household solid fuel pollution, particulate matter pollution and lack of health care. This is consistent with other findings of previous GBD analyses that have identified high CRD mortality in South Asia, parts of Sub-Saharan Africa, and countries in Oceania (GBD 2021 Diseases and Injuries Collaborators, 2024; GBD 2021 Risk Factors Collaborators, 2024). Cluster 1 comprises the biggest cluster and consists of countries that have relatively low to moderate Cardiovascular diseases death rates. As seen from Appendix Table A.5, the cluster spans from Mongolia (5.874) to Kuwait (0.866). Cluster 1 consists of a number of high-income countries including the United States, the United Kingdom, Germany, France, Canada, Australia, Japan, Switzerland, Norway, Sweden, Singapore, and other European countries. Relatively low mortality recorded in these countries may be associated with the presence of adequate healthcare infrastructure, early diagnosis, proper treatment of chronic respiratory diseases, measures to reduce the number of smokers, and reduced exposure to various environmental risks. These results are in line with the international data indicating the relationship between lower CRD mortality and health systems in the countries, along with implementation of various public health strategies (World Health Organization [WHO], 2023; GBD 2021 Diseases and Injuries Collaborators, 2024).

The second cluster comprises only five countries and is distinguished as having the largest mortality burden out of all clusters. The country with the highest mortality rate was Nepal (52.019), followed by Papua New Guinea (43.657), Democratic People’s Republic of Korea (37.972), Myanmar (34.372), and India (33.360). The segregation of this cluster from other clusters in Figure 4.3.5 shows that the mortality from chronic respiratory diseases in this cluster is very high compared to the global average. It implies that these countries have similar epidemiology, such as high exposure to indoor and outdoor pollution, combustion of biomass fuels, smoking, occupational risks, and obstacles in accessing healthcare services for treating respiratory diseases. Indeed, there have been cases of high mortality from chronic respiratory diseases identified in previous GBD assessments in South Asia and Western Pacific regions (GBD 2021 Risk Factors Collaborators, 2024). The third cluster includes the group of countries with an intermediate mortality level between the low-mortality group of Cluster 1 and high-mortality groups of Clusters 0 and 2. As seen from the Appendix Table A.5, it starts with Democratic Republic of Congo (13.062) and ends with Malaysia (6.149). This fact is reflected by the variety of the countries included in this cluster showing that there is moderate CRD mortality in several geographical areas irrespective of the level of socio-economic development. The above statement implies that identical mortality rates can be observed in areas with similar distribution of respiratory risk factors rather than geographic closeness.

In general, clustering shows distinct gradation of chronic respiratory disease mortality among countries. Despite Cluster 1 including countries with relatively low CRD mortality rate, Clusters 0 and 2 include those countries which experience higher mortality from this condition and Cluster 2 includes those countries with maximum CRD mortality rate. Also, the fact that countries of different continents are in one cluster proves the effectiveness of clustering using computer algorithms when compared to geographical clustering. The four different clusters for chronic respiratory diseases revealed distinct mortality patterns. The second cluster had the highest burden of mortality, having only five countries with very high Cardiovascular diseases mortality rate and the country with the highest mortality being Nepal. The third cluster had the burden of high mortality, especially from South Asia, Sub-Saharan Africa and Oceania. The first cluster was the largest cluster and included mostly countries with low mortality rates, most of which were high income countries. The fourth cluster had intermediate mortality in different geographical locations. In general, the data confirm that through data driven clustering, countries can be classified based on their disease burden rather than geographical region.

4.3.5.1: Comparison of Data-Driven Clustering and GBD

The comparison between the data-based clustering and the (GBD) regional categorization is provided in Table 4.8 for cardiovascular disease mortality caused by chronic respiratory diseases. The difference from the GBD regional classification is that while the latter takes into consideration only the physical proximity of the countries, the data-based categorization relies on the countries’ mortality load similarity. That means that countries belonging to different GBD sub-regions and super-regions often fall into one cluster.

Table 4.8: Comparison of Data-Driven Clustering and GBD

Cluster 0 consists of countries with a higher mortality and comprises nations from Oceania, South Asia, Southeast Asia, Central Latin America, and Sub-Saharan Africa. Among those countries included in Cluster 0, the highest mortality rate was in Kiribati (27.633), while Kiribati is a nation from the Oceania sub-region of the Southeast Asia, East Asia, and Oceania super-region, but the lowest mortality was in Rwanda (13.382) that falls in the Eastern Sub-Saharan Africa sub-region of the Sub-Saharan Africa super-region. The Appendix Table A.5 also shows that there are many regions of GBD regions where countries in this cluster are located.Cluster 1 comprises the greatest number of countries, having relatively low mortality burden levels. The highest mortality level in Cluster 1 was found in Mongolia (5.874), a country located in the East Asia sub-region, whereas the lowest mortality rate was found in Kuwait (0.866) belonging to North Africa and Middle East GBD region. Cluster 1 encompasses countries from high-income, European, Latin American, North African and Middle Eastern, Southeast Asian, and Sub-Saharan African regions as seen in Appendix Table A.5; this means that countries with relatively low mortality burdens are spread over several GBD regions.

Cluster 2 consists of only five countries, all coming from either South Asia or Southeast Asia, East Asia, and Oceania GBD super-region. The highest mortality burden is found in Nepal (52.019), while the lowest value is in India (33.360). All other countries, namely Papua New Guinea, Democratic People’s Republic of Korea, and Myanmar, are found in the neighbouring Asian and Oceanic GBD regions. Such grouping implies that these countries share similar epidemiological attributes of extremely high mortality burden from chronic respiratory diseases.In Cluster 3, the countries are characterized by medium mortality levels and show great geographical diversity. The country with the highest mortality level (13.062) is the Democratic Republic of the Congo, which is located in the Central Sub-Saharan Africa sub-region. At the same time, the country with the lowest mortality level (6.149) is Malaysia, which is located in the Southeast Asia sub-region. As seen in Appendix Table A.5, countries in this cluster belong to such GBD regions as Sub-Saharan Africa, Oceania, South Asia, Central Asia, North Africa and the Middle East, High-income regions, Latin America, and Southeast Asia. This implies that countries with similar mortality patterns are geographically dispersed across multiple GBD regions.In conclusion, the results of the comparison reveal that the data-driven clustering method does not reproduce the geographical structure of the GBD regional classification. On the contrary, the countries are clustered into groups based on their similarities regarding mortality patterns for chronic respiratory diseases, irrespective of their location. Some clusters are made up of countries mostly within one GBD super-region, but other clusters contain countries from several regions.

4.3.6 Data-Driven Cluster Based on Cardiovascular diseases DALYs Rates for Chronic Respiratory Disease

The clustering analysis based on the ratios of Cardiovascular diseases DALYs for chronic respiratory diseases revealed four clear clusters, indicating large variations in the burden of disease across countries in Figure 4.24. Unlike geographic classification, which groups regions geographically close to each other, the clustering method has clustered countries according to disease burden regardless of geographical closeness, suggesting the presence of similar risk factors among these countries.

Figure 4.24: Cluster Based on Cardiovascular diseases DALYs rates for Chronic respiratory

DALYs-based clustering

• Cluster 0: 56 countries
• Cluster 1: 120 countries
• Cluster 2: 21 countries
• Cluster 3: 7 countries

Cluster 2 included those nations that had DALYs in large numbers, namely, Vanuatu, Pakistan, Solomon Islands, Bangladesh, Lao People’s Democratic Republic, Bhutan, China, Cambodia, Madagascar, Somalia, Afghanistan, Rwanda, and Democratic Republic of the Congo. Most of these countries belong to South Asia, Southeast Asia, Oceania, and Sub-Saharan Africa, where chronic respiratory diseases continue to be an important public health issue. The inclusion of many nations from the Pacific Islands in this group could be because of the impact of various factors, such as tobacco use, indoor pollution, changes in the environment, and limited health care facilities. In the same way, other nations, such as Pakistan, Bangladesh, Afghanistan, and Somalia, have been observed to suffer from high incidence of respiratory diseases because of urbanization, biomass fuel usage, poor air quality, and lack of prevention of respiratory diseases (GBD Risk Factors Collaborators, 2024; WHO, 2023).

The number of countries belonging to Cluster 1 was the highest, along with the inclusion of most wealthy countries and some middle-income countries. Denmark reported the highest DALYs rate for Cluster 1 (148.181), while the lowest DALYs rate was recorded in Singapore (30.927). This cluster was noted to show comparatively lower disease burden because of the efficient healthcare systems, application of effective tobacco control laws, use of cleaner fuel at home, better occupational safety laws, and more access to diagnostic and chronic respiratory care services. Research evidence suggests that nations with more effective public health measures and air quality regulations tend to report lower DALYs for chronic respiratory diseases (Soriano et al., 2020; GBD 2021 Chronic Respiratory Disease Collaborators, 2024).

Cluster 0 was associated with countries with a relatively high DALY rate and included countries such as Zimbabwe, Indonesia, Kazakhstan, South Sudan, Guinea Bissau, Kenya, Bahrain, Egypt, South Africa, and the United States. The cluster analysis shows that countries sharing a similar chronic respiratory disease burden tend to be from distinct GBD regions, underscoring the added significance of using a data-driven cluster approach rather than geographical classification. Though the GBD regions provide a good geographical classification, cluster analysis using DALYs allows one to identify similar epidemiological groups that could benefit from similar preventive and health care measures. This gives way to prioritizing countries based on their burden of disease and developing appropriate health care strategies.

4.3.6.1:  Comparison of Data-driven Clustering and GBD

Table A6 is a comparison of countries classified using clustering against their respective GBD sub-regions and super-regions using cardiovascular diseases DALY rates of chronic respiratory diseases. It can be seen from the comparison that the use of clustering to classify the countries did so based on their epidemiological similarities rather than geographical ones, as is shown in the conventional GBD regional classification in table 4.9.

Table 4. 9: Comparison of Data-driven Clustering and GBD

Cluster 3 had the highest disease burden, with the highest Cardiovascular diseases DALYs rate in Nepal (995.697 per 100,000 population), followed by Papua New Guinea (870.370), Democratic People’s Republic of Korea (700.008), India (657.645), Myanmar (638.320), Lesotho (612.009), and Kiribati (581.056). The countries are located in different GBD subregions like South Asia, Oceania, East Asia and Southern Sub-Saharan Africa. Though they differ geographically, the grouping into the same cluster was made due to similarities regarding the high DALYs rate. This implies that the disease burden is more strongly driven by the risk factors common to all countries like air pollution, smoke, occupational risk, poverty, lack of healthcare services and others. It has been observed in other GBD studies as well.The most recent data from the GBD Study show that CRDs are still clustered in low- and middle-income countries with a high prevalence of risk factors (GBD 2021 Diseases and Injuries Collaborators, 2024; Soriano et al., 2020).

Similar to Cluster 1, the nations in Cluster 2 also had higher DALYs rates compared to the nations in Cluster 1, although lower than Cluster 3. The highest DALYs rate within this cluster was in Vanuatu (507.433), while the lowest one belonged to the Democratic Republic of the Congo (299.823). Countries in this cluster included Pakistan, Bangladesh, Bhutan, China, Cambodia, Somalia, Madagascar, Rwanda, and Timor-Leste. Though these nations are from South Asia, Southeast Asia, Oceania, East Asia, and Sub-Saharan Africa, they were clustered in this group due to the similarities of their disease burden levels.

Cluster 1 was formed by countries with moderate DALYs rates, with the values ranging from Denmark (148.181) to Singapore (30.927). The countries involved in the cluster were from Western Europe, North Africa and Middle East, Latin America and Caribbean, Central Europe, East Asia, Southeast Asia, and High income. It is evident that the clustering of both high income and middle-income countries into one cluster signifies the ability to have similar disease burden irrespective of the economic condition. High income countries enjoy improved health care systems, low smoking rates and environmental laws while middle-income countries have their disease burden reduced through improvements in health care access and reduction of tobacco use. As a result, they converge at moderate DALYs rates (GBD Chronic Respiratory Disease Collaborators, 2020; WHO, 2023).

Countries with medium to high DALYs rates were in Cluster 0 and the country with the highest DALYs rate was Zimbabwe (291.629) and the country with the lowest DALYs rate was the Syrian Arab Republic (155.551). The countries of the cluster were mainly countries of Sub-Saharan Africa, Southeast Asia, Oceania, Central Asia and North Africa and Middle East regions. This further supports that the clustering algorithm is primarily clustered based on similarities in disease burden rather than geographical regions. Several countries in the cluster remain highly exposed to pollution, burning of biomass fuels, respiratory infections and health care disparities (GBD 2021 Diseases and Injuries Collaborators, 2024; Burney et al., 2015). In conclusion, comparing the clustering of data-driven methods versus the GBD classification suggests that countries belonging to one particular GBD region may not have a similar level of chronic respiratory diseases. On the contrary, countries belonging to different regions were grouped together due to similar Cardiovascular diseases DALYs rates. This study shows that data-driven clustering offers complementary data to the conventional GBD classification since it groups countries having similar epidemiological characteristics irrespective of their geographical locations. This method might be useful to policymakers for intervention design based on epidemiological and risk factor similarity, rather than on regional classification.

4.4 Geospatial Hotspot and Cold spot Analysis

Geospatial hotspot and cold spot analysis were undertaken to assess the spatial distribution and clustering of the burden of major NCDs in countries and regions. This analysis is useful for understanding geographical disparities in disease burden, such as hotspots and cold spots, thus gaining insight into geographical inequalities. Three main diseases, namely diabetes mellitus, Cardiovascular diseases (CVDs) and chronic respiratory diseases (CRDs), were used in this study for hotspot and cold spot analyses. The results identify areas with a disease burden that is consistently high and areas that have a lower burden, which will inform evidence-informed public health planning and priority-setting for interventions. The identification of spatial clusters of disease burden will enable policymakers to focus resources and apply region-specific disease prevention and control to decrease the global impact of these diseases.

4.4.1 Geospatial Hot and Cold spot Analysis for CVDs

The global geospatial distribution of clusters of Cardiovascular disease (CVD) is shown by figure 4.25, which is categorized as hot spot, moderate or cold. The analysis categorizes countries into hotspot, moderate and cold spot groups using spatial clustering of the burden of CVDs. Hot spots refer to countries having a disease burden that is much higher compared to other countries. Cold spots refer to countries having relatively low burden. The data indicate significant geographical variations in cardiovascular diseases across the globe.

Figure 4.25: Geospatial Hot and Cold spots for Cardiovascular disease

According to the map, the Egypt country was the highest hotspot (Cluster 3), which meant that this country had the highest burden of Cardiovascular diseases among all the countries considered. Kazakhstan, on the other hand, was determined to have the lowest cold spot (Cluster 0) suggesting a relatively lower spatial concentration of the burden of CVD. A small number of countries in cold spot clusters were in North Africa, the Middle East and parts of Sub-Saharan Africa; and a significant share of countries in North America, South America, Western Europe, East Asia and Oceania were in moderate clusters. The spatial distribution shows that the burden of Cardiovascular diseases is not uniform and there are distinct regional clustering patterns. This high hotspot in Egypt might be associated with the presence of the major risk factors for the development of Cardiovascular diseases including hypertension, obesity, diabetes mellitus, physical inactivity, dietary habits and smoking. Rapid urbanization, demographic transition, and exposure to metabolic risk factors have been suggested as factors behind the heavy burden of Cardiovascular disease in the countries of the MENA region (Roth et al., 2020).

These factors result in an increase in cardiovascular morbidity and mortality, leading to hotspots. The clustering pattern found in various African and Asian countries can be associated with the diversity of disease management programs and healthcare access. Cardiovascular burden of disease is a worldwide phenomenon that still persists due to its late detection, limited access to healthcare facilities, and poor management of cardiovascular risk factors in LMICs (World Health Organization [WHO], 2023). Notably, the presence of inequalities and differences in healthcare structures and accessibility may be behind the geographic clustering of the burden of CVD in certain hotspots. The selection of Kazakhstan as the lowest cold spot can mean that this country has comparatively lower geographic concentration of the burden of Cardiovascular diseases than the neighbouring countries. This may be explained by the difference between the countries regarding population characteristics, health care interventions or epidemiology.

The fact that the region is a cold spot does not imply that Cardiovascular disease is absent in the region but that the burden of the disease is low as compared to other geographical clusters under consideration. Clusters that exhibited moderate levels of prevalence in several developed nations may be attributed to the success of public health measures employed to reduce cardiovascular risk factors. Favorable trends in health care services, screening, measures against smoking, as well as improved management of hypertension and dyslipidaemia have contributed to the reduction of the burden of Cardiovascular disease in high income nations (Mensah et al., 2019). Nevertheless, such measures have not been sufficient to stop Cardiovascular diseases from being a major cause of death and disability worldwide. In general, hotspot/cold spot analysis reveals that there are significant spatial inequalities in the burden of Cardiovascular diseases in the global scenario. Hotspot identification is important information for governments and public health authorities that would help them in using their resources effectively and implement preventive measures for these hotspot areas. Improvements in health care access, cardiovascular risk factor control, lifestyle promotion and early diagnosis will be essential in reducing the burden of Cardiovascular reduction in Cardiovascular health inequities between regions.

4.4.2 Diabetes Mellitus

Figure 4.26 shows the burden of diabetes mellitus by country in the world and its distribution at a geospatial level. Countries are classified as hotspot, moderate and cold spot clusters based on the spatial distribution of diabetes burden. Hotspots are countries where the diabetes burden is much higher than in neighbouring regions, and cold spots are areas with comparatively low diabetes burden. The geographical distribution shows a distinct geographical pattern indicating that there is a lot of regional variation in the burden of diabetes mellitus worldwide.

Figure 4.26: Geospatial Hot and Cold spots for Diabetes Mellitus

Results show that Fiji was ranked as the highest hotspot (Cluster 3), the country most concentrated on diabetes burden across space among countries analysed. Somalia, by contrast, was the lowest cold spot (Cluster 0), meaning that there was relatively less spatial clustering of diabetes burden in the region. The majority of the countries in North America, South America, Europe, Asia and Africa were identified as cold spots, indicating a generally low level of diabetes burden. There was a localized burden of high diabetes with only a limited number of countries classified as moderate or hotspot clusters. This finding that Fiji is the most important hotspot is in line with earlier studies on the high prevalence and mortality rates of diabetes in many Pacific Island countries. Rapid nutrition transition, the rise of obesity, lack of physical activity, urbanization and genetic predisposition of the Pacific populations have been associated with the burden of diabetes in these countries (International Diabetes Federation [IDF], 2021).

The general trend to eat energy-dense food products and the reduction of physical activity has also increased the diabetes epidemic in the region and helped to create spatial hotspots. Diabetes burden hotspot areas might also have their burden associated with a combination of several metabolic risk factors including high body mass index (BMI), high fasting plasma glucose, high blood pressure, and poor eating habits. GBD Risk Factors Collaborators (2024) note that high BMI is one of the greatest risk factors for morbidity and mortality from diabetes in the world through the  analysis. Risk factors are not independent of each other and the countries having hot spot features would certainly have experienced all of these risk factors. Cold spot clusters are found in many areas in the world indicating that there are relatively low concentrations of diabetes burden in most areas. This categorisation of a cold spot should however be interpreted with caution as it does not necessarily mean there is low prevalence of diabetes. Instead, it means that the burden is reduced in comparison with the other regions analyzed in the spatial analysis. Despite the prevalence of metabolic risk factors, significant diabetes-related morbidity and mortality can still be found in many considered “cold spot” countries. The identification of Somalia as the lowest cold spot could be due to differing epidemiological patterns, disease reporting systems, lifestyle and/or demographic structure. Underdiagnosis and limited disease surveillance could also impact the reported burden of diabetes in several low-income countries (World Health Organization [WHO], 2023).

Thus, the spatial clustering should be taken in context with other epidemiological data and characteristics of the healthcare systems. Geographic variation observed underscores the significance of the environmental, socioeconomic and behavioral factors in the diabetes burden. The unequal distribution of diabetes across countries can be attributed to a number of factors, such as differences in diet, health care access, obesity rates, urbanisation and the state of public health infrastructure (IDF, 2021). For regions that are identified as hotspots, there will be a need for increased public health activities such as obesity prevention initiatives, health promotion activities, enhanced screening, and enhanced diabetes management. Overall, the hotspot and cold spot analysis shed light on the significant spatial variations in the global burden of diabetes mellitus. Fiji is the most prominent hotspot, highlighting ongoing susceptibility of  Pacific Island countries to diabetes related health outcomes, and the prevalence of cold spot clusters indicates high regional variation in diabetes burden. The results of this study help guide policy and public health decisions and can guide planning of geographically targeted interventions that will help to reduce the burden of diabetes mellitus globally.

4.4.3 Chronic Respiratory Diseases

Figure 4.27 shows the distribution of chronic respiratory diseases (CRDs) globally as a geospatial hotspot and cold spot. Countries are clustered into hotspot countries, moderate countries, and cold spot countries by the spatial distribution of CRD burden. A hotspot region is a region of countries where the burden of CRD is significantly higher than neighbouring regions and a cold spot region a region of countries where the burden of CRD is comparatively lower. The findings show significant geographical variation in the prevalence of chronic respiratory diseases around the world, which could be attributed to different exposures of the environment, socioeconomic status, access to health services and population health attributes, among other factors.

Figure 4.27: Geospatial Hot and Cold spots for Chronic Respiratory Disease

The results show Fiji was the highest hotspot (Cluster 3) with highest spatial concentration of CRD burden among all countries in the analysis. In contrast, Indonesia became the lowest cold spot (Cluster 0), indicating a relatively low spatial clustering of CRD burden. There were hotspot clusters in several countries in Sub-Saharan Africa, Middle East, Central Asia and some parts of the Pacific region, and cold spot clusters for most of the countries in North America, South America, Europe, East Asia and Oceania. Fewer countries mostly in South Asia and the Pacific – were regarded as moderate clusters. A previous study identified Fiji as the most prominent hotspot as had many other studies that have previously reported the high burden of chronic respiratory diseases in many Pacific Island countries. This is because these areas are burdened with many respiratory diseases because of the increasing prevalence of risk factors for NCDs, inadequate healthcare, pollution in air both indoors and outdoors, and cigarette smoking (World Health Organization [WHO], 2023).

Geographical isolation of many PICs may also restrict access to specialist respiratory health care, leading to increased disability and early death from CRDs. The clustering of hotspots in several areas of Sub-Saharan Africa and Middle East could be attributed to chronic exposure to high risk factors for respiratory conditions. The factors lead to the development of important geographical clusters with high CRD burden. The distribution of cold spot clusters in North America, Europe, East Asia and other areas could be due to better public health practices and health care systems. In the past few decades, numerous countries have adopted tobacco control policies, environmental regulations, air quality standards, and enhanced respiratory health care services, leading to a decline in the burden of chronic respiratory diseases (Soriano et al., 2020). In addition to good monitoring and early detection and treatment of chronic respiratory diseases, they have also been key aspects in the reduction of the burden of morbidity and mortality associated with CRDs. Amongst the countries, it was discovered that Indonesia had the lowest cold spots, indicating that its spatial burden of CRDs is low when compared to the other countries in the region. This should not be interpreted to mean that there is no burden of respiratory disease, but rather reduced relative clustering in relation to the spatial distribution of the area being analyzed. The detected patterns might be influenced by differences in demography, health care systems, disease reporting systems, and exposure environments.

Some of the major factors that are identified by the  studies as causes of the burden of disability and death related to chronic respiratory diseases include smoking, air pollution due to particulates, occupational hazards and indoor air pollution (GBD 2021 Risk Factors Collaborators, 2024). “Hotspots” could be more vulnerable to these risk factors, resulting in higher burden of disease with spatial clustering. The hotspot and cold spot analyses, in general, indicate a number of differences between the regions concerning the burden of chronic respiratory diseases. Fiji ranks first as a hotspot whereas Indonesia ranks last as a cold spot, indicating the geographical variations in CRDs around the world. The findings indicate that the efforts made geographically in public health to reduce risk factors, improve health systems and increase access to respiratory healthcare facilities are essential. To decrease the burden of chronic respiratory diseases and level the outcomes of the disease among regions, these aspects need consideration.

4.5 Data-Driven SDI Thresholds Development

To comprehend the link between socio-demographic development and the NCD burden, SDI thresholds based on data and disease burden measurements at the country level were constructed. In contrast to the traditional SDI classification applied in the GBD study, such thresholds have been established through observations of the distribution of deaths and DALYs among countries. The suggested method can facilitate the process of clustering countries with comparable NCD burdens. The objective was to assess if current categories of SDI were adequate for explaining heterogeneity in NCD burden and to find other possible threshold values which may do a better job at highlighting differences in NCD outcome. Comparison was made between newly generated SDI categories and standard SDI categories used in the GBD study to gauge the usefulness of the former for health care planning purposes. This paper helps understand the transition process between different categories of socio-economic development and its effect on NCD mortality and DALYs burden.

4.5.1 Data Driven Thresholds for SDI using Deaths from Cardiovascular disease (CVD)

Figure 4.28 shows the Socio-demographic Index (SDI) classification of countries based on the decision tree and the association with age-standardised Cardiovascular disease (CVD) mortality rates. The analysis was conducted to determine data-based SDI thresholds which could classify countries by mortality burden. Through the classification tree approach, four SDI categories were produced, which are more objective classifications based upon the observed mortality patterns rather than on predefined SDI categories.

Figure 4.28: Data-driven SDI thresholds for Death Burden

From the results, SDI is one of the factors that played a vital role in differentiating the variation in Cardiovascular disease mortality between the countries. SDI values for decision trees include SDI = 0.345, 0.414, and 0.627; there were four SDI classes. Countries with the SDI values in the range of 0-0.345 belong to category 1, and there were 12 countries. Category 2 was also represented by 12 countries, with SDI values in the range of 0.346 – 0.414. In addition, category 3 contained 56 countries with SDI values in the range of 0.414-0.627. Finally, category 4 contained 124 countries, most of them, and their SDI values were > 0.627. From the distribution of the countries into the four SDI categories, it is evident that the larger percentage of the countries belonged to higher SDI categories. High-level socioeconomic development was found in most of the countries in Category 4. However, Category 1 and 2 were relatively smaller compared to the other categories, and this could be because there were fewer countries with very low SDI levels. From the analysis of the decision tree, it was clear that the countries whose SDI values were greater than 0.627 were relatively homogenous with respect to the type of mortality pattern.

From the above information, one can observe that the economic development of the nations has an influence on the outcomes of Cardiovascular diseases. SDI is made up of variables such as per capita income, education level, and fertility rates that have been found to affect the health and the delivery of health services (GBD 2021 Demographics Collaborators, 2024). The higher the SDI of the nation, the better the quality of the healthcare system, the effectiveness of the disease prevention programs, the accessibility of medical care, and public awareness about the risk factors for Cardiovascular diseases. The difference between Categories 1 and 2 when the SDI is 0.345 indicates that there are potential Cardiovascular deaths that may arise as a result of improvements in socioeconomic development, no matter how slight they may be. Some of the issues that can face the least developed countries regarding SDI include the poor healthcare infrastructure, lack of preventive programs, insufficient healthcare professionals, and Cardiovascular risk factors.

In the presence of such parameters, one would most probably experience premature deaths as a result of Cardiovascular diseases. The SDI 0.627 that has been recognized to exist has much importance because it defines the distinction between the countries that have moderate socio-economic development and those that have relatively higher levels of socio-economic development. Countries with higher thresholds were more similar regarding mortality, owing to the advantage of being well-equipped with health care facilities, health care providers that have a good ability to control hypertension and dyslipidaemia, screening and emergency Cardiovascular health care. It has always been observed in previous research that the more socially developed the country, the less deaths there are because of Cardiovascular diseases (Roth et al., 2020). The findings imply that there exists no direct relationship between SDI and mortality of Cardiovascular diseases. But, some SDI cut points are indicative of higher socioeconomic development levels. Effects have previously been observed for threshold effects in past studies of the s, whereby health outcomes in so-called “low and lower-middle” income countries increased at a much quicker pace as the income of such countries increased and became “lower-middle” and “high” income countries. (GBD 2021 Demographics Collaborators, 2024).

In general, from the decision tree analysis perspective, it can be concluded that SDI serves as an effective predictor of variation in Cardiovascular disease mortality rates across various countries. By analyzing differences in the burden of CVD in the four categories of SDI based on a data-driven approach, countries can be identified, which share similar socioeconomic and health conditions. The obtained findings underline the importance of socioeconomic development, investment in health care, and health policies aimed at reducing CVD deaths.

4.5.2 Data Driven Thresholds for SDI using DALYs From CVDs

Figure 4.29 displays the classification of countries by the Socio-demographic Index (SDI) based on the decision tree and the relationship between SDI and the burden of Cardiovascular diseases, based on Disability-Adjusted Life Years (DALYs). The analysis was carried out to find objective SDI thresholds which differentiate countries with different levels of Cardiovascular disease burden. The decision tree used statistically derived threshold values to classify the SDI into four categories, which created a data-driven approach and framework for understanding the relationship between socioeconomic development and DALYs attributed to CVDs.

Figure 4.29: Data-driven SDI thresholds for DALYs Burden

From the findings, it is evident that SDI is one of the factors used to differentiate countries based on their Cardiovascular disease burden. Four categories of SDI were established using three cut-off values of SDI, which include 0.345, 0.414, and 0.627. Countries in category 1 have SDI values between 0.000 and 0.345, and there were twelve such countries. Twelve countries fell into category 2, and their SDI values ranged between 0.346 and 0.414. There were 12 countries falling in category 2 (0.346 to 0.414). Fifty-six countries had SDI values between 0.415 and 0.627; they belonged to Category 3. Countries in Category 4 had SDI values between 0.628 and 1.000, and they form the biggest group (124 countries). From the findings above, it is clear that many countries fall within the higher ranges of SDI, more than three-fifth of the countries belong to category 4, which indicates relatively high socioeconomic development. On the other hand, very few countries (category 1 and 2) formed a smaller share.

The decision tree model revealed that countries whose SDI index was above 0.627 formed a homogeneous category with similar DALYs trends, meaning that socioeconomic development played a significant role in reducing the burden of Cardiovascular disease. This observed association between SDI and DALYs burden indicates the significant contribution that socioeconomic factors make in defining health outcomes in the population. DALYs take into account both deaths occurring before the expected life expectancy age and years spent in a state of disability, and it is therefore a comprehensive measure of disease burden (Murray & Lopez, 1996). Countries with high SDIs generally have better health systems, have more access to preventive health care, have more disease control measures, and health literacy which are all important in reducing the burden of Cardiovascular disease. The lower SDIs indicated by the decision tree might refer to countries with several Cardiovascular health issues. The value of 0.627 seems to be an important threshold, indicating that it divides the countries into those of moderate socioeconomic development and those of relatively advanced development level. Health outcomes tend to be better in the countries above this threshold, which may reflect better emergency cardiac care, better healthcare systems, and more effective care for chronic Cardiovascular disease.

The association between better socioeconomic development and decreased Cardiovascular disease burden has also been found in previous  estimates (GBD 2021 Demographics Collaborators, 2024). Further, the results indicated a non-linear relationship between SDI and Cardiovascular DALYs. Rather, some thresholds for SDI seem to represent tipping points at which the burden of disease is reduced more significantly when there are improvements in socioeconomic conditions.On the other hand, some SDI thresholds look like thresholds where the improvement of socioeconomic status is accompanied by a more significant decrease in disease burden. The pattern suggests that the investments in education, income growth, healthcare infrastructure and public health programs can provide significant health gains at certain levels of development. The overall results of the decision tree analysis show the significant contribution of socioeconomic development to the burden of Cardiovascular disease, regarding DALYs. The SDI thresholds identified are useful to categorize countries based on their level of development and their associated disease burden. The results highlight the need for better health systems, increased access to preventive care, and socio-economic inequities as critical measures to help decrease the burden of Cardiovascular disease in the world.

4.5.3 Data Driven Thresholds for SDI using Deaths from Diabetes

Figure 4.30 The classification of countries by the Socio-demographic Index (SDI) and its association with diabetes mellitus Cardiovascular diseases death rates (ASDR) is presented based on a decision tree analysis.A decision tree analysis is used to present the classification of countries by the Socio-Demographic Index (SDI) and its association with diabetes mellitus Cardiovascular diseases death rates (ASDR). The analysis was conducted to find out SDI thresholds based on data that can be used to classify countries by diabetes-related mortality burden. The decision tree, unlike SDI classifications, produced threshold values directly from the observed data, which enabled countries to be classified into groups that are alike regarding mortality patterns.

Figure 4.30:  Data-driven SDI thresholds for Death Burden

The results clearly indicate that SDI plays an important role in causing disparities in mortality rate due to diabetes among countries. The decision tree showed three critical points in SDI at SDI points 0.414, 0.782, and 0.847, thus making four categories of SDI. Category 1 included 24 countries with SDI values from 0.000 to 0.414. Category 2 comprised countries with SDI values from 0.415 to 0.782. It contained 126 countries out of all the countries. Category 3 consisted of countries with SDI values ranging from 0.783 to 0.847 and 29 countries. Category 4 contained countries with SDI values greater than 0.847 and had 25 countries. Countries plotted in these categories reveal that most countries were grouped in the low and middle range of SDI. Over half of the countries fell into Category 2, which indicates a significant portion of the world’s population lives in countries in a socioeconomic transition. Categories 3 and 4 had relatively few countries, which are considered to have a high socioeconomic level. The decision tree also showed that the countries in the highest SDI category were a relatively homogenous group, having similar mortality features. The thresholds identified indicate a complex socioeconomic development and diabetes mortality relationship.

Diabetes mellitus is strongly linked with economic development, urbanization, population aging and lifestyle changes, unlike many communicable diseases which tend to decrease as development increases (International Diabetes Federation [IDF], 2021). In many countries, as societies develop economically and socially, the diet gets worse, physical activity declines, obesity tends to rise and metabolic risk factors become more common, resulting in a greater incidence of diabetes and its complications. The SDI 0.414 seems to mark a distinction between those countries that are very low on the development scale and countries that are at a point of economic and demographic transition. Countries that fall below this threshold tend to have poorer life expectancy, more limited capacity to detect disease and less extensive health reporting systems. However, with the increasing recognition of diabetes as a major public health problem, as the SDI rises above this level, it is due to the improvement in diagnosis, population aging and increasing exposure to metabolic risk factors (World Health Organization [WHO], 2023).

SDI 0.782 is an important transitional point between the middle and upper middle development countries. Countries belonging to this group usually experience higher levels of urbanization and nutrition transition, including a higher consumption of processed food, physical inactivity and increasing incidence of obesity. This is considered one of the main reasons why the burden of diabetes is currently increasing around the globe (GBD Risk Factors Collaborators, 2024). This means that in countries belonging to this SDI level, the mortality rate caused by diabetes may be rather high despite the existence of an advanced health care system. SDI 0.847 is the upper level which divides highly developed countries from others. In countries above this level, advanced health care systems, screenings, glycemic control and complications management are usually present. However, in countries with high SDI, which will experience an increasing burden of diabetes due to demographic changes and presence of obesity and metabolic disorders, the burden of diabetes is still quite high (IDF, 2021).

Countries classified under this category normally have higher rates of urbanization and nutrition transition, involving greater consumption of processed foods, physical inactivity, and growing rates of obesity. It is regarded as one of the key reasons for the increase in the burden of diabetes in the world (GBD Risk Factors Collaborators, 2024). Thus, in countries classified under this SDI level, the mortality rate associated with diabetes can be rather high even if there is a well-developed health care system in the country. SDI 0.847 is the highest threshold that separates highly developed countries from other nations. Advanced health care systems, screenings, glycemic control and complications management are typical features of countries that have this SDI level. However, in countries that have high SDI and which will experience increased burden of diabetes due to demographic changes and obesity, the burden of diabetes is quite high (IDF, 2021).

4.5.4 Data Driven Thresholds for SDI using DALYS From Diabetes

Figure 4.31 displays the decision tree-based classification of countries by the Socio-Demographic Index (SDI) and its relationship with the burden of diabetes mellitus assessed regarding Disability-Adjusted Life Years (DALYs). This analysis was carried out to establish SDI thresholds determined by evidence and distinguishing between countries based on diabetes-associated burden of disease. The decision tree produced four categories of SDI defined by patterns of observed DALYs.  Three major SDI thresholds were established at the levels of 0.572, 0.782, and 0.847, defining four SDI categories in total. Countries in Category 1 had SDI levels ranging between 0.000 and 0.572 and included 62 countries. Countries in Category 2 had SDI levels ranging between 0.573 and 0.782 and formed the biggest group consisting of 88 countries. Countries in Category 3 had SDI levels ranging between 0.783 and 0.847 and consisted of 29 countries. Countries in Category 4 had SDI levels higher than 0.847 and consisted of 25 countries. Thus, it can be stated that most of the countries belonged to low- to middle-SDI and upper-middle-SDI categories are shown in Table.

Figure 4.31: Data-driven SDI thresholds for DALYs Burden

The distribution pattern of countries in the four categories shows that socioeconomic development is a major determinant of DALYs caused by diabetes. Category 2 had the highest proportion of countries since there are a considerable number of countries undergoing demographic, epidemiological, and economic transitions and experiencing an increase in the prevalence rate of diabetes because of urbanization, changes in dietary behavior, lack of physical activities, and obesity (International Diabetes Federation [IDF], 2021).The first threshold, where SDI value equals 0.572, divides countries into those with lower socioeconomic development and those with higher levels of development. Countries below this SDI value have few health care infrastructure, low health care expenditures, and lack of accessibility to health care infrastructure for diagnosis and treatment of diabetes patients. Even though the prevalence rate of diabetes in these countries is relatively low, unmanaged diabetes cases lead to disabilities and premature death of the concerned people (World Health Organization [WHO], 2023).

The second SDI threshold at 0.782 seems to mark a significant turning point in the burden of DM. The SDI range from 0.573 to 0.782 generally marks countries that are economically and socially developing rapidly, thus leading to increased exposure to metabolic risks. Rapid consumption of processed food, occupation changes, and obesity prevalence are among the major factors causing more instances of DM diseases (GBD Risk Factors Collaborators, 2024).Threshold three at SDI 0.847 marks the difference between highly developed countries and those at the upper-middle level of development. Countries under category four have an advanced healthcare system, good DM management program, and easy access to preventive healthcare facilities. All these attributes help to enhance glucose control and decrease complications associated with DM. Nevertheless, despite all these developments in healthcare, DM continues to be a major cause of DALYs due to aging of the population, prolonged period of disease, and the prevalence of obesity among other metabolic risk factors in developed countries (IDF, 2021).

According to the results obtained by using the decision tree algorithm, it is clear that there is a complex and nonlinear connection between SDI and DM DALYs. In contrast to infectious diseases which tend to decline steadily with increasing socioeconomic development, the burden of DM increases through early and middle stages of development before it finally decreases (Murray & Lopez, 1996). The identified thresholds can provide useful insights into global disparities in the burden of DM. Countries falling within Category 1 and Category 2 might be those that stand to gain most from preventive interventions such as prevention of obesity, lifestyle modification, and improving their primary health care systems. On the other hand, countries falling within Category 3 and Category 4 might be those that need intervention strategies aimed at managing the disease in the long run. The results obtained from the decision tree analysis confirm that SDI is an important factor associated with the DALYs due to DM and also helps in providing important cut-off values for classifying countries based on their degree of economic development. The four SDI groups identified using the above approach are indeed useful for making decisions regarding the management of DM across countries.

4.5.5 Data Driven Thresholds for SDI using Deaths from Chronic Respiratory

The decision tree-based classification of countries on the basis of their Socio-Demographic Index (SDI) and its relationship with Cardiovascular diseases death rates for chronic respiratory diseases (CRDs) is shown in Figure 4.32. This study was performed to find the data-driven SDI cut-off values, which could help to classify the countries on the basis of their death rates due to chronic respiratory diseases. The SDI cut-off values were obtained using a decision tree approach based on similarities in mortality patterns rather than predefined socioeconomic classifications.

Figure 4.32: Data-driven SDI thresholds for Death Burden

The findings yielded three main threshold levels of SDI that were 0.413, 0.629, and 0.722. This gave rise to four categories of SDI. The first category contained countries with SDI ranging from 0.000 to 0.413 and contained 24 countries. The second category contained countries with SDI that ranged from 0.414 to 0.629 and had 57 countries. The third category contained countries with SDI ranging from 0.630 to 0.722 and had 40 countries. The fourth category contained countries with SDI greater than 0.723 and had 83 countries.Distribution of countries into these categories shows that quite a significant number of countries belonged to the higher SDI categories. More than 40% of countries fell into Category 4, which means high socioeconomic development of these countries. However, Category 1 was the category which included the least number of countries; this means that these countries have lower socioeconomic development. It can also be concluded that the mortality rates in the countries with higher SDI become more uniform.

These thresholds reflect the significance of socio-economic development in affecting mortality due to chronic respiratory disorders. The Socio-Demographic Index includes income, education and fertility, all of which affect exposures to risk factors for respiratory problems and health care facilities (GBD 2021 Demographics Collaborators, 2024). The countries with high SDIs are also likely to have more developed healthcare infrastructure, effective monitoring of diseases, better availability of treatment for respiratory disorders and more preventive programs for public health.The cut-off point of SDI 0.413 is a point that distinguishes countries with poor socioeconomic development from countries with higher levels of socioeconomic development. Nations that are below this cut-off point may encounter problems with health care, exposure to the environment, and prevention of diseases. Air pollution due to the usage of biomass fuels, working conditions, smoking of tobacco, and underdeveloped health care systems are some of the causes that lead to respiratory disease mortality in many nations with low SDIs (World Health Organization [WHO], 2023).

The second threshold in SDI 0.629 represents the stage of transition from lower-middle to upper-middle stages of development. Nations at this level are prone to fast urbanization and industrialization that can result in increased exposure to air pollutants and other environmental risks. Although there is an improvement in healthcare provision during socioeconomic development, the health effects of such development can be somewhat reduced by the growing exposure to harmful substances (Soriano et al., 2020).SDI 0.722 is especially noteworthy in relation to this case study since it clearly differentiates those countries that are socially and economically more developed from the rest. Countries above this value usually have better health care systems, stricter tobacco control measures, better environmental legislation, and easier access to services related to treatment of chronic respiratory disorders. This results in decreased mortality rate and more efficient treatment of such diseases. As many studies show, decline in prevalence of smoking and improvement of air quality is connected with decreased mortality rate caused by chronic respiratory diseases (GBD Risk Factors Collaborators, 2024).

The results also indicate that the association between SDI and CRD mortality is not linear. There are specific stages of socioeconomic development, at which the further improvement of accessibility of medical care and environmental factors leads to greater decrease in mortality rate. Such results imply that certain investment in the sphere of public health, environment protection, and prevention of respiratory diseases can bring notable health benefits for some countries when they reach the specified level of development. The results of decision tree analysis indicate that SDI is a major determinant of mortality from chronic respiratory diseases and offers important cut-off points for classification of countries depending on their socioeconomic status. Four groups of SDI that have been obtained using the data-driven method provide a good foundation for investigating global differences in CRD mortality and developing policies related to CRD. Effective implementation of tobacco control policies, reduction of air pollution exposure, improved access to healthcare services, and elimination of socioeconomic barriers will play an important role in the reduction of CRD burden worldwide.

4.5.6 Data Driven Thresholds for SDI using Deaths from Diabetes Mellitus (DM)

The decision tree classification for countries based on the Socio-demographic Index (SDI) and its relationship with chronic respiratory diseases (CRD) burden regarding DALYs is shown in Figure 4.33. The decision tree analysis has been done to establish an objective threshold for SDI to classify countries based on their burden of CRDs. In doing so, countries have been classified using the decision tree analysis methodology, which is data-driven and based on similarities of DALYs pattern across SDI groups.

Figure 4.33: Data-driven SDI thresholds for DALYs Burden

The findings showed three significant threshold values at SDI levels of 0.378, 0.618 and 0.722 resulting in four different SDI groups. Group 1 consisted of countries with SDI values ranging between 0.000 and 0.378. The group consisted of 17 countries. Group 2 had countries with SDI values of 0.379 to 0.618. The group was made up of 59 countries. Group 3 included countries with SDI values from 0.619 to 0.722, including 45 countries in total. Group 4 consisted of 83 countries all having SDI values above 0.723.Based on the classification of countries into the respective groups, it can be concluded that the majority of countries were in the higher SDI groups. More than 40 percent of the countries were in Category 4, which indicates relatively high levels of socioeconomic development. On the other hand, the category with the least number of countries was Category 1 which implied countries with relatively low levels of socioeconomic development. From the decision tree model, it can be seen that countries with SDI > 0.722 had relatively similar DALYs, thus the effect of socioeconomic advancement. These cutoff points underscore the critical importance of socio-economic development for the disability and early deaths associated with CRDs. DALY is a complete measure of disease burden that takes into account the loss of life expectancy from early death along with disability adjusted life-years spent (Murray & Lopez, 1996). Therefore, the noted SDI cutoff points represent both mortality as well as the adverse health effects of chronic respiratory diseases.

SDI 0.378 is the first cut-off point for very poorly developed countries compared to moderately developed nations. The low-income nations that fall below this threshold usually have high levels of exposure to respiratory risk factors such as air pollution due to burning of solid fuel, poor living environment, occupational exposure, smoking, and poor access to health care services (World Health Organization [WHO], 2023). These factors cause significant levels of disability and death resulting from chronic respiratory illnesses.At SDI 0.618, there is another cut-off point after which any improvement in socioeconomic status could help reduce the level of diseases. Countries within this threshold have better access to health care facilities, education, and infrastructure. (GBD Risk Factors Collaborators, 2024).

The threshold level for SDI 0.722 defines nations having relatively developed socio-economic status compared to less developed nations. Countries which exceed this threshold level are usually characterized by more advanced health care, strong policies concerning tobacco control, air pollution regulation, and better provision of preventive as well as curative services associated with respiratory health care. These factors make it possible for such countries to have low mortality and disability rates associated with chronic respiratory diseases. Studies prove that reduced smoking rate and better environmental conditions are key factors behind the falling burden of CRDs in high SDI countries (Soriano et al., 2020).The analysis of decision trees also shows that there is a non-linear relationship between SDI and DALYs due to CRD. Contrary to expectations, as the countries become more developed, the DALYs decline not at an even pace but rather accelerate after passing particular socio-economic thresholds. This means that investments in health care systems, protection of the environment, education and prevention of diseases can bring considerable improvements in the health of people.

DALYs patterns seen in countries falling under Category 4 indicate that socio-economic development can assist in bridging the gaps related to respiratory disease results. However, it is important to emphasize that chronic respiratory diseases persist as being burdensome even in those countries which have high socio-economic development because of reasons like population aging, smoking, and environmental dangers. The decision tree analysis reveals that SDI is an important predictor of DALYs due to CRDs and presents useful cutoff values for classifying countries by their socioeconomic status. The four categories of SDI derived using this data-driven approach provide useful insights into inequalities in the burden of CRD on a global scale. Effective measures for prevention of respiratory diseases, protection from environmental factors, and healthcare should be taken to decrease the burden of chronic respiratory diseases globally.

CHAPTER-05

CONCLUSION

The first objective of this study was to estimate the trends of Cardiovascular diseases Deaths and DALYs for Cardiovascular diseases (CVDs), diabetes mellitus (DM), and chronic respiratory diseases (CRDs) globally, regionally and country-wise using the data from GBD 2021. The analysis revealed significant temporal trends in the period 1990-2021. In general, the rates of mortality and DALYs for CVDs and CRDs were decreasing, suggesting that improvements in disease prevention and healthcare services. However, the DALYs rate for DM was increasing but the mortality rate was relatively stable. There was also significant regional and country-level variation, with South Asia and some countries in the Pacific Islands having relatively high disease burden. In addition, results from the ARIMA forecasting indicated that the burden of CVDs and CRDs is likely to continue to decrease gradually, but DM is likely to be a public health challenge in the future by 2050. These results met the first research goal and constitute evidence to assist with trends in disease monitoring and public health planning and resource allocation in the future.

The second objective of this study was successfully achieved by developing data-driven clusters by Cardiovascular diseases Deaths and DALYs rates using K-means clustering. The data-driven perspective of disease distribution was offered by the clustering analysis, which grouped countries together according to disease patterns, not based on geographical location. Results showed that geographical proximity was not the only factor affecting disease burden, but countries from different GBD regions often fell into the same cluster, reflecting the fact that there are epidemiological and socio-demographic similarities between regions. When examining the similarities and differences in both the existing GBD regional classification and the comparison of the obtained classification, there was a noticeable difference in the burden of disease patterns in traditional regional classifications, suggesting that disease burden patterns do not always match existing regional classifications. Overall, the results showed that data-driven clustering offers a more objective and informative way of defining countries with similar health profiles and can help public health plans to be more targeted.

To achieve the third objective, the disease burden patterns observed across GBD regions were compared to those found through data-driven clustering. The comparison showed that the countries were not clustered by region but rather by burden, with many countries falling into different burden clusters than those to which they belonged in the different GBD regions. The results indicate that countries in different geographic regions may have similar epidemiological situations despite geographical isolation. The cluster-based approach thus offered further clues to understanding the disease burden and identified potential unknown similarities. Overall, this comparison illustrated the value of data-driven approaches in enhancing the understanding of the distribution of diseases globally and in aiding planning for regions on the basis of evidence.

This fourth objective was achieved using geospatial hotspot analysis of the data-driven clustering results. The countries with high, moderate and low disease burden were identified and the geographical distribution of Cardiovascular diseases, DM and chronic respiratory diseases was clearly illustrated. Various countries were repeatedly recognized as hotspot areas with a high disease burden, and others as cold spot areas with relatively low disease burden. Spatial visualization showed that the disease burden was not equally distributed geographically around the globe and was used to identify the priority areas to be targeted by public health interventions. Overall, the hotspot analysis proved to be an effective visual presentation of the global patterns of diseases and aided in evidence-based disease prevention and resource allocation planning.

The fifth of this study was accomplished by creating data-driven SDI thresholds with Decision Tree Classifier. The countries were first classified based on their disease burden level and then analysed to determine SDI cut-off values linked to an increased disease burden. The thresholds identified showed that socio-demographic development has a significant association with mortality and DALYs for the selected NCDs. The thresholds were different from those used in SDI classifications, as they were directly derived from the observed data and more representative of actual disease burden patterns. Overall, the results suggest that the suggested data-driven SDI thresholds may play a significant role in high-risk population identification, health policy making and better resource allocation within the healthcare system.

5.1 Recommendation

1. Governments should prioritize high-burden countries for early diagnosis and disease management.

2. Public health programs should focus on reducing modifiable risk factors including unhealthy diet, physical inactivity, smoking, obesity and high blood pressure.

3. Future studies should consider further indicators of the disease burden, including prevalence, incidence, YLLs and YLDs, to provide a comprehensive evaluation of NCDs.

4. Advanced clustering techniques should be explored and compared with K-means clustering, such as Hierarchical Clustering and DBSCAN.

5. Future forecasting studies should be conducted to compare the ARIMA with the advanced machine learning and deep learning models like Prophet and LSTM for enhancing the predictability.

6. Incorporation of further socioeconomic, environmental, healthcare accessibility variables into future SDI analyses would enhance the predictive capability of the disease burden. Traditional GBD regional classification should be used in conjunction with data driven clustering to inform better public health planning and resource allocation.

5.2 Limitation

The quality and completeness of the available data limited the work, as it relied on secondary data from the GBD 2021. The study was limited to 3 major NCDs, Cardiovascular diseases, DM, and chronic respiratory diseases. Other burden measures including prevalence, incidence, YLLs and YLDs were not analyzed. The clustering results may be affected by the number of clusters that is predefined in the K-means clustering algorithm.Cluster classifications were used as the basis for the hotspot analysis. The SDI analysis used only the Socio-demographic Index and did not take into account other factors like healthcare level, environment or lifestyle.

5.3 Significance of Study

The present study is a comprehensive data driven assessment of the burden of CVDs, DMs and CRDs using the data from GBD 2021. This study combines the trend analysis, ARIMA forecasting, K-means clustering, geospatial hotspot analysis, and Decision Tree-based SDI threshold analysis, providing a more comprehensive understanding of the disease burden at the global, regional, and country levels. The results can inform policy makers and public health officials regarding high-burden areas, resource allocation, and creating evidence-based strategies for intervention. Additionally, the study offers a repeatable analytical framework adhering to the GATHER guidelines and can be used in future studies to inform public health decision making and efforts to achieve Sustainable Development Goal 3 (SDG 3).

REFERENCES

Adeloye, D., Chua, S., Lee, C., Basquill, C., Papana, A., Theodoratou, E., … Rudan, I. (2015). Global and regional estimates of COPD prevalence: Systematic review and meta-analysis. Journal of Global Health, 5(2), 020415.

Ahmed, S. M., Hadi, A., Razzaque, A., et al. (2009). Clustering of chronic non-communicable disease risk factors among selected Asian populations: Levels and determinants. Global Health Action.

Allen, L., Williams, J., Townsend, N., et al. (2017). Socioeconomic status and non-communicable disease behavioural risk factors in low-income and lower-middle-income countries. The Lancet Global Health, 5(3), e277–e289.

American Diabetes Association. (2022). Classification and diagnosis of diabetes: Standards of medical care in diabetes. Diabetes Care, 45(Supplement_1), S17–S38.

Bai, J., Cui, J., Shi, F., & Yu, C. (2023). Global epidemiological patterns in the burden of main non-communicable diseases, 1990–2019: Relationships with socio-demographic index. International Journal of Public Health

Beaglehole, R., Bonita, R., Alleyne, G., et al. (2011). UN High-Level Meeting on Non-Communicable Diseases: addressing four questions. The Lancet, 378(9789), 449–455.

Bhutta, Z. A. (2025).  2023: Challenges and opportunities for a growing collaboration. PLOS Medicine, 22(11), e1004838.

Bloom, D. E., & Cadarette, D. (2019). Infectious disease threats in the twenty-first century: strengthening the global response. Frontiers in Immunology, 10, 549.

Centers for Disease Control and Prevention (CDC). (2022). About chronic diseases. Atlanta: CDC.

Forouhi, N. G., & Wareham, N. J. (2019). Epidemiology of diabetes. Medicine, 47(1), 22–27.

Freihat, O. (2025). Global burden and future projections of non-communicable diseases. PLOS ONE.

G. B. (2020). Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: A systematic analysis for the Global Burden of Disease 2019. Collaborators.

GBD 2019 Diseases and Injuries Collaborators. (2020). Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: A systematic analysis for the  Study 2019. The Lancet, 396(10258), 1204–1222.

GBD 2019 Risk Factors Collaborators. (2020). Global burden of 87 risk factors in 204 countries and territories, 1990–2019: A systematic analysis for the Global Burden of Disease 2019. The Lancet, 396(10258), 1223–1249.

GBD 2021 Cardiovascular diseases Collaborators. (2024). Global, regional, and national burden of Cardiovascular diseases and risk factors, 1990–2021. Journal of the American College of Cardiology.

GBD 2021 Chronic Respiratory Disease Collaborators. (2024). Global, regional, and national burden of chronic respiratory diseases and risk factors, 1990–2021. The Lancet Respiratory Medicine.

GBD 2021 Disease and Injury Collaborators. (2024). Global burden of 371 diseases and injuries in 204 countries and territories, 1990–2021: A systematic analysis for the  Study 2021. The Lancet.

GBD 2021 Risk Factors Collaborators. (2024). Global burden and strength of evidence for 88 risk factors in 204 countries and territories, 1990–2021: A systematic analysis for the  Study 2021. The Lancet, 403(10440), 2162–2203.

GBD 2023 Diseases and Injuries Collaborators. (2025). Burden of 375 diseases and injuries, risk-attributable burden of 88 risk factors, and healthy life expectancy in 204 countries and territories, including 660 subnational locations, 1990–2023: A systematic analysis for the  Study 2023.

GBD Chronic Respiratory Disease Collaborators. (2020). Global burden of chronic respiratory diseases and risk factors, 1990–2019. The Lancet Respiratory Medicine.

GBD Chronic Respiratory Disease Collaborators. (2020). Prevalence and attributable health burden of chronic respiratory diseases, 1990–2017: A systematic analysis for the  Study 2017. The Lancet Respiratory Medicine, 8(6), 585–596

GBD Chronic Respiratory Disease Collaborators. (2024). Global burden of chronic respiratory diseases and risk factors, 1990–2021. The Lancet Respiratory Medicine.

GBD Risk Factors Collaborators. (2024). Global burden and strength of evidence for 88 risk factors in 204 countries and territories, 1990–2021: A systematic analysis for the  Study 2021. The Lancet, 403(10440), 2162–2203.

GBD Risk Factors Collaborators. (2024). Global burden of 88 risk factors in 204 countries and territories, 1990–2021. The Lancet, 403(10440), 2162–2203.

 (GBD). (2020).  study 2020. Institute for Health Metrics and Evaluation (IHME).

Hay, S. I., et al. (2025). Burden of 375 diseases and injuries, risk-attributable burden of 88 risk factors…  Study 2023. The Lancet.

Heller, O., Somerville, C., Suggs, L. S., et al. (2019). The process of prioritization of non-communicable diseases in low- and middle-income countries. Health Policy and Planning, 34(5), 370–383.

Institute for Health Metrics and Evaluation (IHME). (2024).  (GBD). Seattle, WA: IHME.

International Diabetes Federation. (2021). IDF Diabetes Atlas (10th ed.). Brussels, Belgium: International Diabetes Federation.

Khunti, K., Del Prato, S., Mathieu, C., Kahn, S. E., Gabbay, R. A., & Buse, J. B. (2021). COVID-19, hyperglycemia, and new-onset diabetes. Diabetes Care, 44(12), 2645–2655.

Knutsson, S., & Munthe, C. (2020). Devils in the DALY: Prevailing evaluative assumptions. Public Health Ethics, 13(3), 259–274.

Li, J., Pandian, V., Davidson, P. M., Song, Y., Chen, N., & Fong, D. Y. T. (2025). Burden and attributable risk factors of non-communicable diseases and subtypes in 204 countries and territories, 1990–2021: A systematic analysis for the  Study 2021

Li, Z., Shi, J., Li, N., & Zheng, Z. (2022). Temporal trends in the burden of non-communicable diseases in countries with the highest malaria burden, 1990–2019. Globalization and Health.

M. A. Bhuiyan, N. G. (2024). A comparative systematic review of risk factors, prevalence, and challenges contributing to non-communicable diseases in South Asia, Africa, and Caribbeans. Journal of Health, Population and Nutrition, Vol. 43 Issue 1 Pages 140.

M. Lu, B. L. (2024). Temporal Decomposition Analysis of Noncommunicable Disease Burden: The Interplay of Population Aging, Population Growth, and Low Physical Activity, 2010–2019. Journal of Physical Activity and Health, Vol. 1 Issue aop Pages 1-10.

Mannino, D. M., & Buist, A. S. (2007). Global burden of COPD: Risk factors, prevalence, and future trends. The Lancet, 370(9589), 765–773.

Mensah, G. A., Roth, G. A., & Fuster, V. (2018). The global burden of Cardiovascular diseases and risk factors: 2020 and beyond. Journal of the American College of Cardiology, 74(20), 2529–2532.

Mensah, G. A., Roth, G. A., & Fuster, V. (2019). The global burden of Cardiovascular diseases and risk factors: 2020 and beyond. Journal of the American College of Cardiology, 74(20), 2529–2532.

Mishra, S. R., Ghimire, S., & Gyawali, B. (2021). Diabetes in Pacific Island countries: Challenges and opportunities. The Lancet Regional Health – Western Pacific, 14, Article 100231.

Murad, A., & Khashoggi, B. F. (2020). Using GIS for disease mapping and clustering in Jeddah, Saudi Arabia. ISPRS International Journal of Geo-Information.

Murray, C. J. L. (2022). The  Study at 30 years. Nature Medicine. World Health Organization (WHO). (2025). Noncommunicable diseases fact sheet. WHO.

Murray, C. J. L., Abbafati, C., Abbas, K. M., Abbasi-Kangevari, M., Abd-Allah, F., Abdelalim, A., et al. (2020). Five insights from the  Study 2019. The Lancet, 396(10258), 1135–1159.

N Gibbs, J. K. (2020). Operational research to support equitable non-communicable disease policy in low-income and middle-income countries in the sustainable development era: a …. BMJ Global health,

NCD Alliance. (2023). Non-communicable diseases in small island developing states.

Nishino, K., Gyeltshen, T., & Rahman, M. (2025). Non-communicable diseases among forcibly displaced people: A systematic mapping review. International Journal of Environmental Research and Public Health.

Nugent, R. B. (2018). Investing in non-communicable disease prevention and management to advance the Sustainable Development Goals. The Lancet, 391(10134), 2029-2035. .

Ong, K. L., et al. (2023). Global, regional, and national burden of diabetes from … s, Injuries, and Risk Factors Study (GBD). The Lancet.

World Health Organization. (2021). Noncommunicable diseases. Retrieved from https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases. World Health Organization (WHO).

Poblete, F., Dannewitz-Prossello, A., Latta, D., & Galaviz, K. I. (2022). Dietary patterns and type 2 diabetes risk in Pacific Island populations: A systematic review. Nutrients, 14(3), Article 512.

Qiao, J., Lin, X., Wu, Y., et al. (2022). Global burden of non-communicable diseases attributable to dietary risks in 1990–2019. Journal of Human Nutrition and Dietetics.

Roth, G. A., Mensah, G. A., Johnson, C. O., Addolorato, G., Ammirati, E., Baddour, L. M., et al. (2020). Global burden of Cardiovascular diseases and risk factors, 1990–2019: Update from the GBD Study 2019. Journal of the American College of Cardiology, 76(25), 2982–3021.

Saeedi, P., Petersohn, I., Salpea, P., Malanda, B., Karuranga, S., Unwin, N., et al. (2019). Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045. Diabetes Research and Clinical Practice, 157, 107843.

Soares Andrade, C. A., Mahrouseh, N., & Gabrani, J. (2023). Inequalities in the burden of non-communicable diseases across European countries: A systematic analysis of the  2019 study. International Journal for Equity in Health.

Soodejani, M. T. (2024). Non-communicable diseases in the world over the past century: a secondary data analysis. Frontiers in Public Health , Vol. 12 Pages 1436236.

Soriano, J. B., Kendrick, P. J., Paulson, K. R., Gupta, V., Abrams, E. M., Adedoyin, R. A., et al. (2020). Prevalence and attributable health burden of chronic respiratory diseases, 1990–2017. The Lancet Respiratory Medicine, 8(6), 585–596.

Stanaway, J. D., Afshin, A., Gakidou, E., Lim, S. S., Abate, D., Abate, K. H., & GBD 2017 Risk Factors Collaborators. (2018). Global, regional, and national comparative risk assessment of 84 behavioural, environmental and occupational, and metabolic risks or clusters of risks for 195 countries and territories, 1990–2017. The Lancet, 392(10159), 1923–1994. https://doi.org/10.1016/S0140-6736(18)32225-6

Vos, T., Lim, S. S., Abbafati, C., Abbas, K. M., Abbasi, M., Abbasifard, M., et al. (2020). Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: A systematic analysis for the  Study 2019. The Lancet, 396(10258), 1204–1222.

Wang, X., et al. (2024). Global trends in Cardiovascular disease burden and mortality from the GBD 2021 Study.

Wang, Y. W. (2020). Modelling and prediction of global non-communicable diseases. BMC public health , Vol. 20 Pages 1-13.

World Health Organization (WHO). (2021). Noncommunicable diseases. Geneva: World Health Organization.

World Health Organization (WHO). (2022). Diabetes. Geneva: World Health Organization.

World Health Organization (WHO). (2022). Noncommunicable diseases progress monitor 2022. Geneva: World Health Organization.

World Health Organization (WHO). (2023). Communicable diseases. Geneva: World Health Organization

World Health Organization. (2023). Cardiovascular diseases (CVDs). Geneva, Switzerland: World Health Organization. https://www.who.int/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds)

World Health Organization. (2023). Chronic respiratory diseases. Geneva, Switzerland: World Health Organization.

World Health Organization. (2023). Noncommunicable diseases. Geneva: World Health Organization.

World Health Organization. (2023). Noncommunicable diseases.

X. Pan, S. Q. (2021). Global and Regional Burden of Non-Communicable Diseases Among Older Aged 70 and Older (1990–2021): A Systematic Analysis from the  Study 2021.

Yusuf, S., Joseph, P., Rangarajan, S., et al. (2020). Modifiable risk factors, Cardiovascular disease, and mortality. The Lancet, 395(10226), 795–808.

Zhang, J., et al. (2025). Global, regional, and national Cardiovascular disease burden attributable to smoking from 1990 to 2021. Tobacco Induced Diseases.

Zheng, Y., Ley, S. H., & Hu, F. B. (2018). Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nature Reviews Endocrinology, 14(2), 88–98.

APPENDICES

APPENDIX-A

Table A.  1: Comparison of Data Driven Clustering Deaths Rate and GBD regions of Cardiovascular diseases (CVD).

Table A.2: Comparison of Data Driven Clustering DALYs Rate and GBD regions of Cardiovascular diseases (CVD)

Table A.  3: Comparison of Data Driven Clustering Deaths rate and GBD region of Diabetes Mellitus

Table A.  4: Comparison of Data Driven Clustering DALYs rate and GBD region of Diabetes Mellitus