Article Text
PDF
Abstract
Objective With socioeconomic development, improvement in preventing and curing infectious diseases, and increased exposure to non-communicable diseases (NCDs) risk factors (eg, overweight/obesity, sedentary lifestyle), the majority of adult deaths in Bangladesh in recent years are due to NCDs. This study examines trends in cause-specific mortality risks using data from the Matlab Health and Demographic Surveillance System (HDSS).
Design, settings and participants We conducted a follow-up study from 2003 to 2017 using data from Matlab HDSS, which covers a rural population of 0.24 million (in 2018) in Chandpur, Bangladesh. HDSS assessed the causes of all deaths using verbal autopsy and classified the causes using the 10th revision of the International Statistical Classification of Diseases. We examined 19 327 deaths involving 2 279 237 person-years.
Methods We calculated annual cause-specific mortality rates and estimated adjusted proportional HRs using a Cox proportional hazards model.
Results All-cause mortality risk declined over the study period among people aged 15 and older, but the risk from stroke increased, and from heart disease and cancers remained unchanged. These causes were more common among middle-aged and older people and thus bore the most burden. Mortality from causes other than NCDs—namely, infectious and respiratory diseases, injuries, endocrine disorders and others—declined yet still constituted over 30% of all deaths. Thus, the overall mortality decline was associated with the decline of causes other than NCDs. Mortality risk sharply increased with age. Men had higher mortality than women from heart disease, cancers and other causes, but not from stroke. Lower household wealth quintile people have higher mortality than higher household wealth quintile people, non-Muslims than Muslims.
Conclusion Deaths from stroke, heart disease and cancers were either on the rise or remained unchanged, but other causes declined continuously from 2003 to 2017. Immediate strengthening of the preventive and curative healthcare systems for NCDs management is a burning need.
- Stroke
- PUBLIC HEALTH
- Heart failure
- Hypertension
- Health informatics
Data availability statement
Data are available upon reasonable request. The HDSS data are sharable on request and only for the purposes related to this study.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
The study used data from the Matlab Health and Demographic Surveillance System (HDSS), which provides complete registration of household-level demographic events (eg, birth, death) and precise population at-risk data are collected through bi-monthly follow-up of a well-defined population. The high level of completeness for registration of events is a strength of the study.
Matlab HDSS provides a unique opportunity to analyse adult mortality data, including causes of death, which are rarely available in developing countries. The causes of deaths are assessed based on the WHO-comparable verbal autopsy (VA) questionnaire and strict adherence to the 10th revision of the International Statistical Classification of Diseases.
Using the VA-based cause of death instead of clinical assessment of the causes of death is a limitation of this study. However, it should be noted that close agreement was found between the Matlab HDSS cause of death classification and another Smart VA procedure tested in Matlab during 2011–2014.
Introduction
Adult mortality in developing countries has received little attention, until recently. This is especially so in Bangladesh, mainly due to prioritising child survival and more pressing social issues.1 With substantial improvements in child survival, an increasing proportion of deaths are now from the adult population—the majority of which are from non-communicable diseases (NCDs). NCDs are the world’s major killers, with cerebrovascular and cardiovascular diseases (CVDs), cancers, chronic respiratory diseases and diabetes among the major contributors. Worldwide, over 70% of deaths are from NCDs.2 The WHO estimated that NCDs would be responsible for 80% of the global burden of disease by 2020.3 According to the Sustainable Development Goal target 3.4, deaths from NCDs should be reduced by a third by 2030 relative to the 2015 levels.4
Bangladesh, a South Asian country, recently graduated to low-income and middle-income level and has achieved an unprecedent improvement in terms of the health of its population. For example, life expectancy has reached 71 years and 74 years for men and women, respectively; 20 years higher than it was 30 years ago.5 These gains in life expectancy were achieved through dramatic improvement of child survival, and the share of adult mortality (ages 15 years or older) almost doubled from 46% during the years 1983–1987 to 87% during 2013–2017, mainly due to NCDs.6 NCDs include CVDs (including diseases of the circulatory system, that is, hypertension, ischaemic heart disease, stroke and other CVDs) and malignant neoplasms or cancers. There are concerns that mortality from CVDs is rapidly increasing in Bangladesh, for example, the age-standardised mortality rates from CVDs among women increased from around 7 per 100 000 to about 330 per 100 000 during 1986–2006.7 The malignant neoplasm mortality rate increased rapidly during the period.7 However, positive changes are taking place in terms of increased access to healthcare, as well as the introduction and/or availability of health technologies to more efficiently manage NCDs and thus lower related fatalities.
Rapid socioeconomic improvements in Bangladesh have influenced higher caloric intake and sedentary lifestyle, etc, which have direct effects on the onset of CVDs and other conditions. The economy has shifted from being agriculturally based to one more focused on industry and services (currently, 55% and 27% of Bangladesh’s gross domestic product (GDP) is generated from service and industrial sectors, respectively), with accompanying increases in per capita income and caloric and other consumptions.8 Physical activities have declined due to work pattern changes in the country’s transition from agricultural to industrial. A third of men and two-thirds of women were observed to have low levels of daily physical activity. Due to Bangladesh’s extensive road networks, people do not walk as frequently, or as far, when moving from one place to another. The nutritional status of the adult population has been increasing over time—for example, 52% of women aged 15–49 years were underweight (with body mass index (BMI) below 18.5) in 1996–1997, which declined to below 15% in 2017–2018.9 The prevalence of those considered overweight/obese (BMI ≥25) increased from less than 3% in 1996–1997 to over 25% in 2017–2018.
Along with the increases in those considered overweight or obese, associated NCDs such as hypertension and diabetes are also growing in communities with demographic and socioeconomic variations. Between 2011 and 2017, nationwide, the prevalence of hypertension among women and men aged 35 years and older increased from 25% to 39% (32%–45% for women and 19%–34% for men).9 10 In 2017–2018, about half of hypertensive women and two-thirds of hypertensive men were unaware of their elevated blood pressure (BP) levels. Only 15% of hypertensive women and 9% of hypertensive men were aware, taking medication and had their BP under control.9 The prevalence of diabetes among women and men aged 35 years and older increased from 11% to 14% in the same period. Around 60% of diabetic women and men were unaware of their elevated blood glucose levels. Only 13% of diabetic women and men were aware, taking medication and had their blood glucose under control.9
Though all the above factors (ageing population, changes in dietary pattern, sedentary lifestyle and others) are likely to enhance the risk of NCDs, investigation regarding the risks of mortality from NCDs in Bangladesh is scant. The purpose of this study was to examine the individual risk of adult (aged 15 years and older) mortality from NCDs between 2003 and 2017 using high-quality longitudinal data from the Matlab Health and Demographic Surveillance System (HDSS). Mortality risk may have been reduced owing to the increased accessibility to improved medical technologies that can more effectively treat NCDs, but the incidence of NCDs may have increased due to the risk factors associated with economic and social improvements during the time of the study (ie, from 2003 to 2017), when the trends of resulting mortality risks might have been upward, downward or stabilised. There is a methodological issue in the intended trend analysis. The population overall is ageing (eg, 65+ years population increased from 5.2% in 2003 to 7.3% in 2017) and significant social and economic improvements such as education, industrialisation and road transportation have taken place in Bangladesh. Mortality rates among adult populations may increase in time as the proportion of older people increases. In contrast, mortality rates may decrease as a result of social and economic improvements. Thus, comparison of adult mortality over time can be confounded. In this study, we use regression analysis to control the effects of these two indicators—age and economic conditions—on mortality risks. We place emphasis on the cause-specific mortality risks from stroke, heart disease, cancers and other causes, including changes in these risks over time.
Materials and methods
Data
We used adult mortality data from Matlab HDSS maintained by International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b) since 1966.11 The Matlab HDSS site, in 142 villages, is located 55 km south-east of Dhaka. In 2017, around 239 000 people lived in the area.6 Vital events—namely births, deaths, migrations and marital events—are registered by local female community health research workers (CHRWs) through 2 monthly home visits. The completeness of registration of events is expected to be almost universal because of strong field implementation of HDSS and with support from the Matlab community. One reason for this community support is icddr,b’s community-based and facility-based interventions on the elimination of deaths from cholera and diarrhoeal diseases, including the promotion of child immunisation and other child health and maternal health services which have resulted in an unprecedented reduction in infant and child mortality and maternal mortality. The Matlab CHRWs have been reputable as trusted health agents in the community and thus have been able to reach very high levels of completion in terms of registering births, deaths and other vital events. The Matlab HDSS has been treated as the gold standard of vital registration, and global health researchers have undertaken studies to test survey data collection tools that collect data on births, deaths, pregnancy wastage, perinatal mortality and others.12–14 Currently, a number of studies on the validation of adult deaths completeness in surveys using recent household death module and sibling survival history, and accuracy of age and time of death are being undertaken in Matlab HDSS areas.15 16 Adult mortality findings based on Matlab HDSS data have been widely disseminated.1 17–21 One of these studies compared Matlab old-age mortality levels and patterns with those in Sri Lanka and Sweden.1
Matlab HDSS has developed a cause-of-death classification algorithm in which supervisory staff collects information on cause of death through a verbal autopsy (VA) procedure that aims to describe the conditions before death, signs and symptoms of morbidity associated with the death, treatments given, including the type of provider, and place of death. Information from the deceased’s healthcare documents such as prescriptions and diagnostic tests are reviewed, including a death certificate if the death occurred in a health centre. A medical assistant reviews the completed VA questionnaire and assigns a cause of death in accordance with the online 10th revision of the International Statistical Classification of Diseases and related health problems using standardised disease coding guidelines.22 Hazard and colleagues found close agreement of the Matlab HDSS cause of death classification with another Smart VA procedure tested in Matlab from 2011 to 2014.18 The cause of death classification was based on semistructured narrative questionnaires before 2003, thus the long-term comparison of causes becomes difficult. Therefore, we used the VA data for the period from 2003 to 2017.
We analysed a total of 19 327 deaths from the adult population aged 15 years and older. There were 1775 deaths among those aged 15–44 years, 4831 among those aged 45–64 years, 10 772 among those aged 65–84 years, and 1949 among those aged 85 years and older. These deaths constituted 9%, 25%, 56% and 10% of the above age groups, respectively. The number of combined person-years involved in the analysis was 2279 237.
Statistical analysis
Bivariate analysis
Age-specific and cause-specific mortality rates were compared in the bivariate analysis by dividing the number of deaths by the number of person-years. The mortality rates have been expressed as deaths per number of person-years instead of deaths per number of persons because the study includes deaths and population at risk for multiple years. The sum of mid-year population at risk for multiple years (as needed in the analysis) have been treated as person-years. The person-years sizes by year are presented in online supplemental table 1.
Supplemental material
Broad age groups considered were: 15–29 years, 30–44 years, 45–54 years, 55–64 years, 65–74 years, 75–84 years and 85+. The mortality rates within the above broad age groups were found to be similar. It should be noted that the size of the 85+ years population was very small—at about 0.4%—but we kept this group separate as members of this age range often require special attention in terms of care at home and health facilities.
The causes of death were categorised as (A) Stroke or cerebrovascular diseases, (B) Heart disease, (C) Cancers, (D) Endocrine disorder, (E) Other NCDs, (F) Communicable diseases, (G) Respiratory diseases, (H) Injuries (intentional and unintentional), and (I) Others. See online supplemental table 2 for detailed categories of deaths. The above categorisation allows for an examination of various causes of death.
The study period was divided into five subperiods—2003–2005, 2006–2008, 2009–2011, 2012–2014 and 2015–2017—for the comparison of mortality risks over time.
Regression analysis
Our aim was to examine if mortality risk of the adult population declined during the period from 2003 to 2017. As indicated above, there have been population ageing and economic improvements during the study period which might have affected mortality risk. Therefore, we needed to control for the effects of ageing and economic conditions in the analysis. Age of individuals and household wealth quintiles were used as indicators of ageing and economic conditions, respectively. The ‘household wealth quintiles’ is a measure of household durable assets, for example, land, television, refrigerator, based on principal component analysis. We used proportional hazards regression with study period as the primary explanatory variable, which was divided into five groups (2003–2005, 2006–2008, 2009–2011, 2012–2014 and 2015–17). The control variables used were: (1) Age of individual; (2) Household wealth quintile, as a measure of household economic condition; (3) Sex of individual; and (4) Religion.
Unadjusted hazard ratios (UHRs) and adjusted hazard ratios (AHRs) are presented for overall mortality (all causes) and separately for stroke, heart disease, cancers and other causes. We examine stroke, heart disease and cancer mortality separately because of the higher number of deaths due to the three causes in the Matlab HDSS area.6
Outcome variable
The binary outcome of interest was survival status. In survival analysis, participants who died within a 1 year follow-up period were considered as the event cases, with the respondents who were alive after the end of the follow-up period considered as the censored cases. For a cause-specific analysis, censoring happened due to the survival until the end of the follow-up or death due to causes other than the cause of death of interest. The censoring indicator was divided into four causes of death: stroke (event: died from stroke; censor: alive or died from other causes except for stroke), heart disease (event: died from heart disease; censor: alive or died from other causes except for heart disease), cancers (event: died from cancers; censor: alive or died from other causes except for cancers), and other causes (event: died from other than stroke/heart disease/cancers; censor: alive or died from stroke/heart disease/cancers). Cause of death classifications are presented in online supplemental table 2.
Explanatory variable
The explanatory variables were included in the regression model as dummy variables. As time trend, five subperiods—2003–2005, 2006–2008, 2009–2011, 2012–2014 and 2015–2017—were considered, with 2003–2005 as the reference category.
The ageing effect was considered in seven age categories—15–29 years (reference category), 30–44 years, 45–54 years, 45–64 years, 65–74 years, 75–84 years and 85+ years. The household wealth quintiles—a measure of household economic conditions—were based on the wealth score constructed via principal component analysis of housing characteristics and possession of durable assets. Housing characteristics considered were roof, wall and floor materials, livestock, land, sanitary toilet, and a number of durable assets (eg, television, refrigerator, electric fan). The range of asset scores was divided into five equal groups, and these quintiles were constructed in 2004 and 2010. The quintile categories were the lowest, second, middle, fourth and highest quintile, which are addressed as poorest, poor, average, above average and rich, respectively, in the following sections.
There was a marked variation of mortality by sex, and women experienced lower mortality than men, as shown in both national and Matlab data.5 6 9 11 21 22 Adult mortality in Matlab was negatively associated with socioeconomic conditions, measured by education and economic status, as observed by previous research.1 15 16 18 21 22
Similarly, there were mortality differences between religious groups.23 We included these two variables in our analysis. Women were treated as the reference category in the sex variable and Muslims were treated as the reference category in the religion variable.
We did not consider some important explanatory variables/factors that significantly affect mortality—one being education, which was not included in the analysis due to the unavailability of updated education status of individuals. The HDSS collects education data periodically. The remarkable improvements in education that occurred in Bangladesh during the study period were mainly among children in elementary and middle school and are unlikely to affect the adult population we studied.
Other behaviours and lifestyles of adults can affect survival in very significant ways; for example, smoking or drinking habits, physical activity, exercise, or being overweight or obese, are all good predictors of mortality, but our data set does not include indicators on such behaviours or lifestyles. Occupational hazards are also significant risk factors for adult populations, but we do not have data on updated occupation status for all because the Matlab HDSS collects occupation information periodically, not in routine visits. Our study of adult mortality suffers from the limitation of not being able to consider these indicators.
Patient and public involvement
No patients were involved in this study.
Results
Age-specific and cause-specific mortality rates and their trends
Cause-specific mortality rates, by broad age group, for the years 2003–2017 are shown in table 1 (Panel A shows mortality rates and Panel B shows the relative proportion of mortality rates). Over this 15-year period, the annual mortality rate among people aged 15 years and older was 848 per 100 000 (reflected in the cells for ‘all causes’ and ‘all ages’ in table 1), or 8.48 per 1000.
Cause-specific annual mortality rate, by broad age category, Matlab, 2003–2017
All-cause mortality rate was 122 per 100 000 in ages 15–44 years, 807 in ages 45–64 years, 5147 in ages 65–84 years and 18 162 in ages 85+ years (shown in the last row of Panel A in table 1, for all causes)—a sharp increase of rate by age. For all ages (15+ years), CVDs or stroke was the most common cause (25%), followed by heart disease (23%) and cancers (11%) (shown in the last column of Panel B in table 1). Each of the categories of respiratory diseases, communicable diseases, injuries, other NCDs and other causes constituted 6%–8% of deaths. All NCDs constituted about 70% of deaths, while 30% were associated with non-NCDs (see online supplemental tables 2 and 3 for more detailed cause of death).
Considering each of the age groups in table 1, the most common cause of death in ages 15–44 was injuries (25%), followed by heart disease (15%), other NCDs (14%) and cancers (14%). In ages 45–64, the most common cause was heart disease (25%), followed by cancers (19%), CVD/stroke (18%) and other NCDs (10%). In ages 65–84, stroke (30%) was the most common cause of death, followed by heart disease (23%) and respiratory diseases (9%). In ages 85+, stroke (34%) and heart disease (22%) were the most common causes of death, followed by other causes.
The relative contribution of stroke was highest among those aged 65–84 years and 85+ years, that of heart disease was highest in ages 45–84 years, and of cancers was highest in ages 15–45 years (table 1). Other NCDs were more common in ages 15–45 years than older ages. The relative contribution of NCDs increased until age 64 years, then declined, as can be seen from the third lowest row in the right-hand panel of table 1. Injury was the most common cause of death in younger ages, 15–45 years (table 1); communicable diseases were almost equally common in all ages; and respiratory diseases were more common in ages 45 years or older.
Table 2 shows the trends of cause-specific mortality rates separately for the four broad age groups—15–44 years, 45–64 years, 65–84 years and 85+ years—in Panels A, B, C and D. Panel E shows rates for all ages combined. In each panel, the percentage of change in cause-specific mortality rates between 2003–2005 and 2015–2017 are shown. A negative percentage indicates a decline in mortality rate, with a positive percentage indicating an increase in mortality.
Cause-specific mortality trends by broad age category, Matlab, 2003–2017: annual mortality rate per 100 000
In Panel E in table 2, where mortality rates are lumped together for all ages, there were 825 deaths per 100 000 people from 2003 to 2005, which gradually increased over time to 876, a 6% increase. The second to last columns show that rate from ‘all other causes’ consistently declined over time by 22%, from 285 in 2003–2005 to 221 in 2015–2017. The rate from all NCDs increased from 540 in 2003–2005 to 655 in 2015–2017, a 21% increase. The magnitude and direction of changes varied by cause (shown in the last column, Panel E, of table 2). The risk of dying from cancers, heart disease and stroke increased by 26%–68% between 2003–2005 and 2015–2017.
When considering the changes in different age groups, we find both declines and increases in cause-specific mortality risks (table 2). All-cause mortality rates declined from 2003–2005 and 2015–2017 for all age groups (as shown in panels A–E), and the declines were generally greater in lower age groups than higher age groups. The decline was 28% and 11% in ages 15–44 years and 45–64 years, respectively, while it was 16% in ages 65–84 years and 7% in ages 85 years or older. There was a 26% decline of NCD mortality in ages 15–44 years, but only minimal changes (between −1% and 6%) were observed in older ages. Non-NCD mortality declined substantially across all age groups—31% in ages 15–44 years, 44% in ages 45–64 years, 38% in ages 65–84 years and 27% in ages 85+ years.
Considering individual causes, mortality from stroke increased, and the magnitude of these increases, by and large, was positively associated with age. Heart disease slightly declined for the age group 15–44 years, but for older age groups it increased. Similarly, cancer rates declined for the age group 15–44 years, whereas for older age groups it increased. The rate of communicable diseases declined substantially, in the age groups 15–44 years, 45–64 years and 65–84 years. However, the rate increased by 31% in the 85+ years age group. The respiratory disease mortality rate substantially declined in different age groups, for example, by 30% in ages 65–84 years and 62% in ages 85 years or older. The risk of mortality from injuries also declined substantially across all age groups—with the 45–64 years age group achieving the highest decline (47%) and the 85+ years age group the lowest (9%).
Age-specific, sex-specific and cause-specific mortality trends
We considered sex in the analysis at this stage, with cause-specific mortality trends, by sex, examined in figure 1 separately by age group—15–44 years, 45–64 years, 65–84 years and 85+ years—across four panels. Each panel (figure 1) contains four cause groups: (1) CVD/stroke, (2) Heart disease, (3) Cancers, and (4) Others. There are three common characteristics: (1) Men experienced higher mortality than women (with the exception of stroke mortality in the 15–44 years age group); (2) Mortality from stroke, heart disease and cancers, with slight fluctuations, remained virtually unchanged for both sexes over the study period; and (3) There was a consistent decline of mortality from other causes (ie, other than stroke, heart disease and cancer) for both sexes, but the sex differential tended to decline with age.
Annual mortality rate per 100 000 for stroke, heart disease, cancers and other causes, by age group, sex and calendar-year group, Matlab (Health and Demographic Surveillance System) HDSS, 2003–2017.
Patterns emerged related to the levels of mortality within each age group. For example, in the 15–44 years age group, mortality from heart disease and cancers was higher than that from stroke, both for men and women. In the 45–64 years age group, mortality from stroke, heart disease and cancers was similar for women, but for men the risk from heart disease and cancers was higher than that from stroke. Mortality risk from other causes was notably lower than stroke, heart disease and cancers for both men and women. As noted above, excess mortality risk of men compared with women was more pronounced in the 45–64 years age group. In the 65–84 years age group, the highest level of mortality risk was from stroke, followed by heart disease and cancers.
Proportional hazards regression results
We observed an increase in the all-cause mortality rate over time, which may be associated with the ageing population. In table 3, we observed that the population average age increased from 27.2 years in 2003–2005 to 30.2 years in 2015–2017. Thus, the apparent increase in mortality may be, in part, due to the proportionate increase in people’s ages. In table 3, we show the UHR of mortality related to the study period (unadjusted as no other explanatory variables were considered in the hazards regression). The UHR of 1.07, for all causes, means that mortality risk increased significantly—(p<0.05) by 7% ((1.07–1.00)×100=7)—between 2003–2005 and 2015–2017 (table 2 shows a similar increase of 6% during these periods). The UHRs for other causes were comparable to the changes observed in table 2.
Unadjusted HRs (95% CI) of mortality over the study period with Matlab 2003–2017 (N*=2276 735)
In table 4, where we control for the effects of age and other explanatory variables (sex, age, household wealth and religion), we found startlingly different results. For example, AHRs from all causes (shown in the last column of table 4) were lower than 1.00 and significant, meaning that mortality risk decreased over time. AHR was 0.94 for the years 2006–2008, meaning that mortality risk was (1.00–0.94=0.06)×100, or 6% lower in 2006–2008 than in 2003–2005; in 2009–2011, AHR was 13% lower than in 2003–2005; in 2012–2014, it was 15% lower, and in 2015–2017, it was 18% lower than in 2003–2005. In other words, mortality risk from any cause consistently declined over the study period, and the decline from 2003 to 2017 was 18% and was significant, implying over 1% of mortality risk decline per year.
Adjusted HRs (AHRs) (95% CI) of mortalitytoby explanatory variablestoMatlab 2003–2017 (n=2276 735)
When considering an individual cause of death, for example, stroke (shown in the left-most column for cause of death in table 4), the risk of dying from stroke increased significantly over the years in the study period—an increase of between 16% and 41%. The next two columns of table 4 show that there were no significantly consistent increases or decreases of mortality risk from heart disease or cancers.
The mortality risk of dying from other causes sharply and significantly declined—by 23% between 2003–2005 and 2006–2008, by 32% between 2003–2005 and 2009–2011, by 38% between 2003–2005 and 2012–2014, and by 43% between 2003–2005 and 2015–2017.
It should be noted that other causes include other NCDs (other than stroke, heart disease and cancers). We ran a hazards regression with other NCDs and found that the risk of dying from other NCDs declined significantly (data not shown). The other causes consisted of 30% of all-cause mortality among all ages and other NCDs consisted of 8% among all ages (see table 1). The all-cause mortality risk decline was essentially associated with the decline in the risk of dying from non-NCD causes.
The following were the determinants of cause-specific and all-cause mortality. The all-cause mortality risk increased sharply with age, as observed in table 1, in the bivariate analysis. The last column of table 4 shows that AHR was 1.79 for ages 30–44 years compared with ages 15–44 years, and it sharply increased with advancement of age—to 5.33 in ages 45–54 years, to 14.66 in ages 55–64 years, to 41.17 in ages 65–74 years, to 103.49 in ages 75–84 years, and to 203.01 in ages 85 years or above. The age effect was different for different causes—with the sharpest increase observed for stroke, followed by heart disease. The age effect was weaker for cancers and other causes, as the increase in AHRs was lower in these causes than in stroke and heart disease.
Men had higher mortality risk than women for all causes, heart disease, cancers and other causes, but not for stroke (AHR=0.98; 0.93–1.04). For all causes, men had 26%–34% (AHR=1.30; 1.26–1.34) higher risk than women; for heart disease their risk was 31%–48% higher (AHR=1.39; 1.31–1.48); for cancers their risk was 105%–147% higher (AHR=2.25; 2.05–2.47), and for other causes male risk was 25%–36% (AHR=1.30; 1.25–1.36) higher than that of women.
There was a negative association between mortality risk and household wealth quintile for all causes, stroke and other causes. There was no association of mortality risk for heart disease and cancers (table 4). For all causes, persons from the ‘poor’ and higher quintiles had lower risk than the ‘poorest’ quintile; AHR gradually declined with quintile, from 0.92 in the ‘poor’ quintile, to 0.83 in the ‘average’ quintile, to 0.79 in the ‘above average’ quintile, and to 0.72 in the ‘rich’ quintile. There was a similar decline of AHRs for stroke and other causes.
Non-Muslims had 14%–24% (AHR=1.19; 1.14–1.24) higher mortality risk than Muslims from all causes. Such risk was higher for heart disease—by 42%–67% (AHR=1.54; 1.42–1.67)—and for stroke—by 12%–32% (AHR=1.21; 1.12–1.32). No religious differentials were observed for cancers and other causes.
Discussion
Overall mortality risk declined among people ages 15 years and older during the study period, from 2003 to 2017, in Matlab, Bangladesh, but mortality from CVD/stroke increased, while that from heart disease and cancers remained stable. Mortality from other causes, namely, communicable diseases, respiratory diseases, other NCDs, injuries, endocrine disorders, and other causes, declined. These other causes constituted about 30% of all deaths. Thus, the overall mortality decline was associated with declines in causes other than NCDs.
The finding that the risks of mortality from causes other than CVD/stroke, heart disease and cancers are declining likely means that the country’s healthcare systems have achieved reasonable management of these causes. Bangladesh has a large health infrastructure with tertiary-level facilities spread over 64 districts and primary healthcare facilities spread over approximately 492 subdistricts, supplemented by a large number of private-sector facilities which are equipped for dealing with these diseases and/or conditions.
However, mortality risks from NCDs were either increasing or unchanged, which implies, by and large, two things. First, the NCDs (eg, hypertension, diabetes) and the risk factors associated with NCDs (eg, overweight or obesity)—are on the rise. The management of stroke, heart disease and cancers require high tech infrastructure and advanced provider skills which remain inadequate in Bangladesh; although, these are becoming increasingly available in Dhaka, the country’s capital, though mostly at the private sector. These services are expensive—with only relatively affluent people able to afford them—and, as our findings suggest, people in the higher wealth quintiles experienced lower mortality risk than those in the ‘poorest and poor’ quintiles. Public-sector infrastructure and facilities are open to all socioeconomic groups, but with inadequate access, as most are located in Dhaka and select mid-sized cities. Moreover, these facilities have poor quality of care.24 25 Thus, public-sector facilities are yet to achieve efficiency in handling NCD cases. Second, the fact that increased mortality risk from NCDs is more common among middle-aged to older people, and that the mortality risk from NCDs is not declining, likely means that middle-aged to older people’s suffering from NCDs is increasing. The more vulnerable are likely to be those from the lower two wealth quintiles (the poorest and the poor), with limited financial capacity, and thus less access to healthcare; and they remain the group most vulnerable to the NCD burden.
As always, the public sector should enhance its efforts—including improved access, range of required services and quality of care—to deal with NCDs, namely CVD/stroke, heart disease and cancers, in facilities located in the capital city of Dhaka, and at least at the district-level facilities in the country’s 64 districts. The low-income groups largely depend on public-sector services in the capital, as well as at the district level.
On the preventive side, the government health system has not yet implemented any large-scale preventive interventions that can yield high impact. There are several initiatives (eg, BP measurement for all people age 40 years and above) of the government and non-governmental organisations to implement appropriate programmes, but many areas remain where their efforts could be enhanced or strengthened (eg, mass communication to increase people awareness).26 The government has begun development of NCD preventive messages (eg, all people age 40 years and above should get their BP measured in a regular basis) for dissemination through mass media (electronic and print media), community health workers and facility-based healthcare providers, but this is still in the planning stage. The government also has plans to screen for hypertension and diabetes, and development of management protocols and small-scale interventions are at the pilot level, but scaling up these interventions may take some time.27 28
We found that female mortality was remarkably and significantly lower than that for men. The observed lower female mortality is consistent with the expectation of female survival superiority over men. Nationally, in 2017, women had nearly 3 years higher life expectancy than men (73.5 years vs 70.6 years, respectively) according to the Sample Vital Registration System report.5 In 2013, the Matlab HDSS, in a rural area, with high levels of completeness of vital events registration, showed female life expectancy at 74 years, versus 70 years for men.29 Women possess a biological superiority for survival, as indicated by lower neonatal mortality among women than men across almost all populations. As we examine adult mortality, we recognise that adult men are more involved than women in heavy work—for example, in factories, construction and agricultural activities—that exposes them to injuries, thus leading to higher mortality risk. Fatal road accidents are phenomenally high in Bangladesh. The workforce is composed mainly of male workers who are frequently involved in work-related travel, resulting in a large number of men involved in accidents with high levels of fatalities. Men are exposed to higher levels of risk factors associated with NCDs, such as smoking, drinking (although drinking is not socially acceptable) or other risky lifestyles.30 Women, in contrast, appear to be more likely to seek healthcare, more likely to adhere to preventive interventions and thus more likely to achieve the associated survival benefits.9 However, women are also more likely to be hypertensive or overweight than men,28 31 and thus are more likely to carry higher risk of dying from NCDs.
Women’s survival superiority over men has been a relatively recent phenomenon in Bangladesh. As recently as around two decades ago, the country’s women had lower life expectancy than men.32 Over time, women continued to have lower life expectancy compared with men, but the male-female gap gradually diminished before reversing in the late 1980s. By 2014, women had achieved 4 years of higher life expectancy compared with men (women: 74 years; men: 70 years). Bangladesh being a patriarchal and male-dominant society with a subsistence farming economy, women too often are relegated to roles of subordination. Women have been discriminated against in terms of food and healthcare allocation, resulting in higher rates of malnutrition and mortality among the country’s female population. In lean seasons, and during times of resource scarcity, women and girls used to receive less than their required food allocation, and healthcare for women was delayed, deferred or ignored. Although neonatal mortality was lower among female neonates, beginning in the postneonatal period, women had higher rates of mortality over the remaining course of their lifetimes. Moreover, women were exposed to high rates of fertility (bearing six or more children) and high levels of maternal mortality, with many women dying as a result of the adverse health conditions experiences during reproductive age. All these factors have impacted women’s survival. Now, with improved social and economic conditions, and with more households being relatively affluent, families need not optimise allocation of resources based on gender preference. Aided by this, women in Bangladesh have achieved greater survival potential and thus higher life expectancy than men.
The observed higher mortality of non-Muslims, most of which are Hindus, is consistent with the findings of Moinuddin and colleagues.23 Over the course of this study, we found that higher mortality among non-Muslims was due to higher mortality from stroke and heart disease, but not from other causes. A recent study31 showed that non-Muslims have a significantly higher risk of elevated BP or hypertension than Muslims. And hypertension is a strong risk factor for stroke and heart disease, thus explaining the higher non-Muslim mortality when compared with Muslims.
Conclusion
Overall, adult (ages 15 years and older) mortality is declining in Bangladesh. However, the decline is associated with declines in non-NCD causes, and the risks of dying from major causes like CVD or stroke, heart disease, and cancers are either on the rise or unchanged from 2003 to 2017. Older people, men, the poor and non-Muslims are more likely to die from NCDs. The government’s preventive and curative healthcare systems in relation to NCDs should be strengthened, including the effective management of NCDs at all district hospitals, at the minimum.
Data availability statement
Data are available upon reasonable request. The HDSS data are sharable on request and only for the purposes related to this study.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
Acknowledgments
The authors thank icddr,b for sharing the Matlab HDSS data. Wayne Hoover from Data for Impact, University of North Carolina at Chapel Hill, USA, did a great job in English editing of the manuscript. icddr,b acknowledges the Government of the People’s Republic of Bangladesh and the Government of Canada for unrestricted support in its operation and research.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors AA and MR conceptualised the study, with AA completing the analysis with support from MMH and MR. AA, MAN, MMR and MR drafted the manuscript. MMH, NA, QN and PKS provided comments. PKS performed the English language editing. All authors reviewed the final draft and agreed with the findings and interpretations. AA takes responsibility for this study as guarantor.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.