Summary
Background
Evidence on non-communicable disease (NCD) prevalence among adults in extreme poverty is sparse. We aimed to determine the national and subnational prevalence of NCD risk factors and chronic conditions among middle-aged and older adults in extreme poverty in India, where a large share of the global population in extreme poverty in this age group lives.
Methods
In this cross-sectional study, we analysed baseline data (2017–19) from the Longitudinal Ageing Study in India using the World Bank’s international poverty line set at $1·90 in 2011 purchasing power parity dollars to define extreme poverty. Outcomes included metabolic (eg, hypertension) and behavioural (eg, tobacco use) risk factors, and chronic conditions (eg, depression). We conducted weighted descriptive analyses to obtain prevalence estimates and multivariable Poisson regression to estimate differences in the relative risk of all outcomes by sex and residence among participants in extreme poverty (n=11 243). We also estimated the relative risk of all outcomes by poverty levels by categorising all participants (n=66 617) into mutually exclusive groups based on the three World Bank poverty line cutoffs (≤$1·90, $1·91–3·20, and $3·21–5·50 versus >$5·50).
Findings
In this nationally representative sample of 66 617 adults aged 45 years and older, 53·4% were male, 46·6% were female, and 39·2% were aged 45–54 years. Among participants in extreme poverty (11 243 [18·1%]), the most prevalent risk factors and chronic conditions were tobacco use (40·3% [95% CI 39·2–41·5]), frailty (39·0% [37·9–40·2]), and hypertension (28·7% [27·7–29·8]). The prevalence of specific outcomes varied across states and territories, and the relative risk differed by residence, sex, and poverty levels.
Interpretation
Middle-aged and older adults in extreme poverty in India have a high prevalence of some NCD risk factors and chronic conditions than those not in extreme poverty, but prevalence varies substantially across states and territories and subgroups, calling for more targeted public health policies for specific geographical areas and subgroups.
Funding
US National Institute on Aging, National Institutes of Health.
Introduction
Improving the health of people in extreme poverty is essential to achieve several of the UN Sustainable Development Goals.1 This population has been thought to have a relatively low prevalence of non-communicable diseases (NCDs);2 however, multicountry analyses reveal that this assumption might no longer hold. A pooled analysis of 105 nationally representative surveys across 78 countries found that key cardiovascular risk factors such as hypertension, diabetes, smoking, and obesity were prevalent even among adults living below the $1·90 purchasing power parity dollars per day extreme poverty line.3 Middle-aged and older adults aged 45 years and older are the population with the highest prevalence of NCDs and, in 2022, India accounted for 5·1% of the global population in extreme poverty in this age group.4 Thus, data on the prevalence of NCD risk factors among this population in India can provide important insight into the global burden of NCDs among middle-aged and older adults in extreme poverty.
In India, NCDs kill approximately 6 million people and account for roughly 63% of all deaths annually,5 but there is ongoing debate about the socioeconomic patterning of NCDs.6,7 Disparate findings across studies stem from the paucity of reliable data. Some studies have attempted to remedy this gap;3,8,9 however, these studies have several methodological limitations, particularly inconsistent measurement of poverty and sampling.
In addition to NCDs, chronic conditions such as mental disorders, frailty, impaired vision, and chronic pain are substantial sources of morbidity and are associated with an increased mortality risk.10 However, the prevalence of these conditions among people in extreme poverty is unknown. Moreover, existing global evidence highlights that the burden of NCD risk among individuals in poverty is not uniform across settings; a 2024 systematic review comparing south Asia, sub-Saharan Africa, and the Caribbean concluded that urbanisation, poor dietary habits, poverty, and fragmented health systems collectively drive the burden of cardiovascular diseases in these regions, but the intensity and relative importance of these drivers vary by geography.11 These findings reinforce the value of examining geographical and sociodemographic variations.
To fill these gaps, we leveraged baseline data from the Longitudinal Ageing Study in India (LASI), a nationally representative study of adults aged 45 years and older in India.12 The aims of this study were to estimate the prevalence of NCD behavioural and metabolic risk factors, and chronic conditions among adults aged 45 years and older in extreme poverty in India; examine variations in the prevalence of these conditions among adults in extreme poverty by state or territory, sex, and residence (urban or rural); and compare the burden of these conditions among adults at varying poverty levels relative to adults not in poverty.
Methods
Data
In this nationally representative cross-sectional study, we used baseline (wave 1) data from LASI, which were collected between 2017 and 2019 from all 36 Indian states and union territories, except for Sikkim, where data were collected in 2021. LASI used a multistage stratified cluster sample design in which rural and urban areas were sampled in three (rural) and four (urban) selection stages. Households were eligible if they had at least one member aged 45 years or older. From eligible households, individuals aged 45 years or older and their spouses (regardless of age) could participate. Adults aged 65 years and older from all selection stages and major cities were oversampled. All surveys and biomarker collection were administered by trained field investigators. More details on the LASI method and biomarker collection protocol are described elsewhere.12,13
LASI was approved by the institutional review boards of the University of Southern California, Harvard University, and the International Institute for Population Sciences. Written informed consent was obtained from all participants. Separate consent was obtained for biomarker collection.
Sample
We excluded participants younger than 45 years because participants from this age group were not representative of the population. Poverty was measured at the household level via self-reported consumption of food and non-food items and expressed in terms of the World Bank’s international poverty lines (appendix pp 1–2). The World Bank has defined three poverty lines set at ≤$1·90, ≤$3·20, and ≤$5·50 per person per day in 2011 purchasing power parity dollars.14 The $1·90 line was used to define extreme poverty and participants at or below this threshold were included in analyses for aims one and two (n=11 243).
For analyses comparing outcomes between participants at different poverty levels, all participants aged 45 years and older were included. Of the 73 408 individuals who completed the baseline survey, 6791 were excluded (n=6789 younger than 45 years; n=2 missing poverty data), leaving 66 617 individuals in the aim three analytic sample.
Of these 66 617 individuals, 60 589 (91·0%) completed the biomarker module and 58 384 (87·6%) provided dried blood specimens (DBS). HbA1c values were missing for 33 participants; therefore, 58 351 participants were included in analyses that used DBS. The appendix (p 3) shows differences in sociodemographic characteristics between participants with and without biomarker and DBS data.
Outcomes
Outcomes included metabolic and behavioural risk factors as these are strongly associated with several NCDs.15 Additionally, we examined select chronic conditions. A detailed description of outcome measurements and definitions is presented in the appendix (pp 4–6). More details on biomarker and physical measurement protocols are available in the LASI biomarker data documentation.13
Hypertension was defined as systolic blood pressure ≥140 mm Hg or diastolic blood presure ≥90 mm Hg. Blood sugar was assessed using a DBS of capillary blood collected via a fingerstick. From blood spots, punches of diameter 3·2 mm were used to measure HbA1c using the Cobas Integra 400 Plus Biochemistry analyser (Roche Diagnostics, Switzerland). Diabetes was defined as HbA1c ≥6·5%. General obesity was defined as BMI ≥25 kg/m2 based on recommended cutoffs for the diagnosis of obesity in Asian Indians.16 Alcohol use was defined as any alcohol consumption within the past 3 months. Tobacco use was defined as current tobacco product smoking or current use of smokeless tobacco products. Depression was measured using the Composite International Diagnostic Interview-Short Form;17 the presence of three or more symptoms indicated a probable major depressive episode (appendix p 7). Frailty was assessed using the Vulnerable Elders Survey assessment.18 A score of three or more was used to define frailty.18 Visual acuity was measured using a Tumbling E Logmar Chart. Distance vision was measured at a 4 m distance and near vision was assessed at 40 cm. Participants with visual acuity of 20/201 or worse for near or distance vision in the better seeing eye were considered to have vision impairment. Chronic pain was defined as self-reportedly having pain frequently (ie, on 5 or more days per week).
Among participants in extreme poverty, we examined variations in the prevalence and risk of all outcomes by sex (male or female) and residence (urban or rural). The $1·90 line is used by the World Bank to define extreme poverty. The $3·20 and $5·50 lines were subsequently introduced to reflect national poverty lines in lower-middle-income and upper-middle-income countries.14 Individuals at these consumption levels remain vulnerable to extreme poverty and its associated health consequences. Therefore, we used all three poverty lines to classify participants into mutually exclusive groups (ie, ≤$1·90; $1·91–$3·20; $3·21–$5·50; and >$5·50). Participants above the $5·50 line were considered to not be in poverty.
Statistical analysis
First, we conducted weighted descriptive analyses to examine sociodemographic characteristics and the prevalence of NCD risk factors and chronic conditions for the total sample and by poverty levels (n=66 617). We then estimated the prevalence of all outcomes by state or territory among those in extreme poverty (n=11 243). We did not estimate prevalence for states or territories with fewer than ten cases for an outcome, due to low reliability. To obtain crude subnational estimates of the absolute number of people in extreme poverty with each of the outcomes, we multiplied the weighted subnational prevalence of each outcome with the total number of adults aged 45 years and older living in extreme poverty in each state or territory based on the 2011 census.
Next, we conducted survey-weighted multivariable Poisson regression with robust standard errors to estimate differences in the relative risk of each outcome by sex and residence among participants in extreme poverty (n=11 243). Models with sex as the primary predictor were adjusted for age and state or territory. Models with residence as the primary predictor were adjusted for sex, age, and state or territory. To account for non-linearities, we modelled age as a continuous variable using restricted cubic splines with knots placed at the following percentiles: 5·0, 27·5, 50·0, 72·5, and 95·0. We used a binary indicator for each state or territory.
Lastly, to obtain differences in the relative risk of each outcome by poverty levels, we ran a survey-weighted multivariable Poisson regression model with robust standard errors using the whole sample (n=66 617) with poverty level as the primary predictor. Participants not in poverty (ie, consuming >$5·50 per day) were the reference group. We adjusted for sex, age, and state or territory as described above.
Analyses for all outcomes (except diabetes) used individual level post-stratification weights to account for differential coverage and response rates, ensuring the LASI sample to be representative of the overall population of adults aged 45 years and older. More details on the construction of post-stratification weights are presented in the LASI documentation.19 Analyses for diabetes used the DBS post-stratification weight, which makes the DBS sample representative of the population of adults aged 45 years and older in terms of age, sex, education and residence. Analyses were conducted using Stata version 14.
Sensitivity analyses
We used alternative cutoffs to examine general obesity (BMI ≥30) and alcohol use (binge drinking; see appendix pp 8–9) and examined variations in the prevalence and risk of all outcomes among participants in extreme poverty by additional sociodemographic characteristics. We also conducted analyses using the World Bank’s multidimensional poverty measure (MPM) to define poverty status.20 Whereas the World Bank’s international poverty line only captures monetary poverty, the MPM captures poverty through the weighted sum of six indicators from three dimensions: monetary (international poverty line), education (school attendance and educational attainment), and access to basic infrastructure (drinking water, sanitation, and electricity). An individual was considered to be in extreme poverty if they were deprived in one or more dimension or the sum of the weighted indicators was ≥0·333 (appendix pp 10–11). We selected the World Bank MPM over the global Multidimensional Poverty Index, as the Multidimensional Poverty Index includes health indicators, which could introduce confounding.
Role of the funding source
The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.
Results
The majority of participants were male (53·4%, vs 46·6% female), had completed no schooling (52·4%), were illiterate (53·0%), currently working (55·5%), married or partnered (77·0%), rural residents (67·8%), and did not have health insurance (78·7%); 39·2% of participants were between 45 and 54 years and 29·2% were ≥65 years; 44·4% of participants belonged to other backward classes and 28·2% were from scheduled caste or tribe (table; appendix p 12). These classifications are the official categories used by the Government of India.
Table:
Sample characteristics and prevalence of metabolic and behavioural risk factors, and chronic conditions by poverty levels: Longitudinal Ageing Study in India (2017–19)
| Total (n=66 617) | ≤$1·90 (n=11 243) | $1·91-$3·20 (n=18 209) | $3·21-$5·50 (n=20 142) | >$5·50 (n=17 023) | |
|---|---|---|---|---|---|
|
| |||||
| Sociodemographic factors | |||||
| Age | |||||
| 45–54 years | 39·2% (38·7–39·7) | 36·1% (34·9–37·3) | 38·0% (37·1–39·0) | 39·4% (38·6–40·3) | 42·8% (41·8–43·8) |
| 55–64 years | 31·6% (31·1–32·0) | 32·2% (31·1–33·3) | 32·1% (31·2–32·9) | 31·1% (30·3–32·0) | 31·0% (30·1–31·9) |
| ≥65 years | 29·2% (28·8–29·6) | 31·7% (30·7–32·8) | 29·9% (29·1–30·7) | 29·4% (28·6–30·2) | 26·2% (25·3–27·0) |
| Sex | |||||
| Female | 46·6% (46·2–72·7) | 47·9% (46·7–49·1) | 47·3% (46·4–48·2) | 46·2% (45·3–47·0) | 45·4% (44·5–46·4) |
| Male | 53·4% (52·9–53·8) | 52·1% (50·9–53·3) | 52·7% (51·8–53·6) | 53·8% (53·0–54·7) | 54·6% (53·6–55·5) |
| Highest level of education | |||||
| No schooling | 52·4% (51·9–52·8) | 71·1% (70·0–72·1) | 59·1% (58·2–60·0) | 48·4% (47·5–49·3) | 34·4% (33·4–35·3) |
| Less than secondary school | 28·9% (28·4–29·3) | 22·7% (21·8–23·7) | 29·2% (28·4–30·0) | 30·9% (30·1–31·7) | 30·6% (29·7–31·5) |
| Secondary school or higher | 18·8% (18·4–19·2) | 61·2% (5·6–6·8) | 11·7% (11·1–12·3) | 20·7% (20·0–21·4) | 35·1% (34·1–36·0) |
| Literate | 47·0% (46·5–47·5) | 28·4% (27·3–29·4) | 40·2% (39·3–41·1) | 50·9% (50·0–51·8) | 65·0% (64·0–65·9) |
| Currently works | 55·5% (55·0–56·0) | 56·3% (55·2–57·5) | 56·0% (55·1–56·9) | 55·2% (54·3–56·1) | 54·5% (53·5–55·5) |
| Relationship status | |||||
| Separated, divorced, or never married | 2·5% (2·3–2·6) | 2·5% (2·1–2·9) | 2·3% (2·0–2·6) | 2·4% (2·1–2·7) | 2·8% (2·5–3·2) |
| Married or partnered | 77·0% (76·6–77·4) | 74·1% (73·1–75·1) | 76·0% (75·3–76·7) | 77·7% (76·9–78·4) | 79·7% (79·0–80·5) |
| Widowed | 20·5% (20·1–20·9) | 23·4% (22·5–24·4) | 21·7% (21·0–22·4) | 19·9% (19·3–20·6) | 17·4% (16·8–18·2) |
| Urban residence Caste* |
32·2% (31·7–32·6) | 16·3% (15·5–17·2) | 26·2% (25·4–27·0) | 36·3% (35·4–37·1) | 46·8% (45·8–47·8) |
| Scheduled caste or scheduled tribe | 28·2% (27·8–28·7) | 42·4% (41·2–43·6) | 31·9% (31·1–32·8) | 24·0% (23·3–24·8) | 17·8% (17·0–18·6) |
| Other backward class | 44·4% (43·9–44·9) | 41·7% (40·5–42·9) | 44·3% (43·4–45·2) | 45·9% (45·0–46·8) | 44·9% (43·9–45·9) |
| No caste or other caste | 27·0% (26·6–27·4) | 15·5% (14·6–16·3) | 23·4% (22·6–24·1) | 29·8% (29·0–30·6) | 37·0% (36·1–38·0) |
| Do not know | 0·4% (0·3–0·4) | 0·5% (0·3–0·7) | 0·4% (0·3–0·5) | 0·3% (0·2–0·4) | 0·3% (0·2–0·4) |
| Has health insurance | 21·3% (20·9–21·7) | 19·9% (19·1–20·8) | 19·9% (19·2–20·6) | 22·0% (21·3–22·7) | 23·2% (22·3–24·0) |
| Metabolic risk factors | |||||
| Hypertension | 31·3% (30·8–31·8) | 28·7% (27·7–29·8) | 30·4% (29·5–31·2) | 32·5% (31·6–33·3) | 33·0% (32·0–34·0) |
| Diabetes | 15·5% (15·1–15·8) | 10·4% (9·7–11·2) | 13·4% (12·8–14·1) | 16·8% (16·1–17·6) | 20·2% (19·3–21·0) |
| General obesity (BMI ≥25) | 26·8% (26·4–27·3) | 14·3% (13·5–15·2) | 21·0% (20·2–21·8) | 30·2% (29·3–31·0) | 40·0% (39·0–41·0) |
| Behavioural risk factors | |||||
| Alcohol use | 10·3% (10·0–10·6) | 10·5% (9·8–11·3) | 10·2% (9·6–10·7) | 10·3% (9·8–10·9) | 10·4% (9·8–11·1) |
| Tobacco use | 35·0% (34·5–35·4) | 40·3% (39·2–41·5) | 38·1% (37·2–39·0) | 33·8% (33·0–34·7) | 28·2% (27·2–29·1) |
| Smoking | 15·6% (15·3–16·0) | 15·7% (14·8–16·6) | 16·7% (16·0–17·5) | 16·0% (15·3–16·7) | 13·7% (13·0–14·5) |
| Smokeless tobacco use | 21·3% (20·9–21·7) | 26·7% (25·6–27·7) | 23·8% (23·0–24·6) | 19·8% (19·0–20·5) | 15·9% (15·1–16·7) |
| Chronic conditions | |||||
| Depression | 8·6% (8·3–8·8) | 8·8% (8·1–9·5) | 8·4% (7·9–8·9) | 8·1% (7·6–8·6) | 9·2% (8·6–9·8) |
| Frailty | 36·4% (35·9–36·8) | 39·0% (37·9–40·2) | 37·5% (36·6–38·4) | 35·8% (35·0–36·7) | 33·6% (32·6–34·5) |
| Vision impairment | 3·5% (3·4–3·7) | 4·9% (4·4–5·4) | 3·9% (3·5–4·2) | 3·3% (3·0–3·6) | 2·4% (2·2–2·8) |
| Chronic pain | 12·1% (11·8–12·4) | 12·3% (11·5–13·0) | 12·0% (11·4–12·6) | 12·0% (11·5–12·6) | 12·0% (11·4–12·7) |
Data are weighted % (95% CI). Poverty levels were defined based on World Bank international poverty lines set at $1·90, $3·20, and $5·50 per person per day in 2011 purchasing power parity dollars.
Classifications are the official categories used by the Government of India.
The prevalence of extreme poverty (≤$1·90) in India was 18·1% (95% CI 17·7–18·5; approximately 46 million people; appendix p 13). The prevalence of poverty was 29·1% (28·7–29·6; approximately 74 million people) at a consumption level of $1·91–3·20 and 29·6% (29·2–30·1; approximately 75 million people) at a consumption level of $3·21–5·50. A higher proportion of participants at all poverty levels (vs not in poverty) were aged 65 years or older, illiterate, widowed, rural residents, had completed no schooling, and belonged to a scheduled caste or tribe (table). In general, central, eastern, and northeastern states (eg, Chhattisgarh [39·5%], Odisha [29·6%], and Meghalaya [31·8%]) had a high prevalence of extreme poverty. Uttar Pradesh had the largest population in extreme poverty (approximately 9·3 million people; appendix p 13).
Among those in extreme poverty, the prevalence of metabolic risk factors was 28·7% (95% CI 27·7–29·8; approximately 13 million people) for hypertension; 10·4% (9·7–11·2; approximately 4·8 million people) for diabetes; and 14·3% (13·5–15·2; approximately 6·6 million people) for general obesity. The prevalence of behavioural risk factors was 10·5% (9·8–11·3; approximately 4·8 million people) for alcohol use, and 40·3% (39·2–41·5; approximately 18·6 million people) for tobacco use. The prevalence of chronic conditions was 8·8% (8·1–9·5; approximately 4 million people) for depression, 39·0% (37·9–40·2; approximately 18 million people) for frailty, 4·9% (4·4–5·4; approximately 2·3 million people) for vision impairment, and 12·3% (11·5–13·0; approximately 5·7 million people) for chronic pain (table; appendix pp 14–19).
Among those in extreme poverty, hypertension prevalence exceeded 25% in most states and territories and was highest in Sikkim (64·2%; figure 1; appendix pp 14–15). The prevalence of diabetes and general obesity was generally higher in economically developed states and territories in the south and north. Andhra Pradesh had the highest prevalence of diabetes (28·6%) and New Delhi had the highest prevalence of general obesity (54·6%; figure 1; appendix pp 14–15). For behavioural risk factors, alcohol use was highest in the western union territories of Dadra and Nagar Haveli (41·7%), and Daman and Diu (35·9%). States in the eastern, central, and southern region (eg, Jharkhand [28·1%], Chhattisgarh [23·8%], Telangana [28·7%]) also had a high prevalence of alcohol use (figure 1; appendix pp 16–17). Tobacco use prevalence exceeded 25% in 26 states and territories. A high prevalence of tobacco use was documented in several northeastern states, such as Mizoram (76·3%) and Tripura (69·2%; figure 1; appendix pp 16–17). The prevalence of chronic conditions also varied across states and territories (appendix pp 18–20).
Figure 1: Prevalence of metabolic and behavioural risk factors by state or territory among middle-aged and older adults in extreme poverty: Longitudinal Ageing Study in India (2017–19).
Grey indicates that prevalence was not estimated because <10 cases occurred. Note that scales vary across figures. Extreme poverty was defined as consumption ≤$1·90 per person per day in 2011 purchasing parity dollars. See appendix (p 33) for state and territory codes.
Compared with males, females had a higher risk of hypertension (relative risk [RR]=1·09 [95% CI 1·01–1·17]) and general obesity (1·52 [1·34–1·72]), and a lower risk of alcohol (0·19 [0·17–0·23]) and tobacco use (0·36 [0·33–0·38]). Females also had a higher risk for all chronic conditions. Diabetes risk did not significantly differ by sex. Rural residents had a lower relative risk of around 18–53% for metabolic risk factors and a higher risk of alcohol use (1·37 [1·07–1·74]), tobacco use (1·20 [1·09–1·31]), depression (1·34 [1·05–1·70]), and frailty (1·24 [1·14–1·34]). The risk of vision impairment and chronic pain did not significantly differ by residence (figure 2; appendix p 21).
Figure 2: Differences in the risk of metabolic and behavioural risk factors, and chronic conditions by sex and residence among middle-aged and older adults in extreme poverty: Longitudinal Ageing Study in India (2017–19).
Models with sex as the primary predictor adjusted for age and state or territory. Models with residence as the primary predictor adjusted for sex, age, and state or territory. Age was modeled as a continuous variable using restricted cubic splines with 5 knots placed at the following percentiles: 5·0, 27·5, 50·0, 72·5, and 95·0. State or territory was modeled using a binary indicator for each state. Extreme poverty was defined as consumption ≤$1·90 per person per day in 2011 purchasing parity dollars.
For metabolic risk factors, participants at all poverty levels (vs >$5·50) had a significantly lower risk of diabetes and general obesity, and the relative risks decreased as poverty levels increased (figure 3). However, only participants consuming ≤$1·90 (vs >$5·50) had a significantly lower risk of hypertension (RR=0·93 [95% CI 0·88–0·97]). For behavioural risk factors, participants at all poverty levels (vs >$5·50) had a higher risk of tobacco use. Additionally, participants at the two highest poverty levels (≤$1·90 and $1·91–3·20; vs >$5·50) had a higher risk of alcohol use. Of chronic conditions examined, participants at all poverty levels (vs >$5·50) had a higher risk of vision impairment and a lower risk of depression. Participants at the two highest poverty levels also had a higher risk of frailty and a lower risk of chronic pain.
Figure 3: Differences in the risk of metabolic and behavioural risk factors, and chronic conditions by poverty levels: Longitudinal Ageing Study in India (2017–19).
All models adjusted for sex, age, and state or territory. Age was modeled as a continuous variable using restricted cubic splines with 5 knots placed at the following percentiles: 5·0, 27·5, 50·0, 72·5, and 95·0. State or territory was modeled using a binary indicator for each state. Poverty levels were defined based on World Bank international poverty lines set at $1·90, $3·20, and $5·50 per person per day in 2011 purchasing power parity dollars. Participants consuming >$5·50 per day were the reference group.
In sensitivity analyses, among those in extreme poverty, the prevalence of general obesity (BMI ≥30) was 2·7% (95% CI 2·3–3·1) and the prevalence of binge drinking was 5·5% (95% CI 5·0–6·1; appendix pp 8–9). For both outcomes, risk estimates by residence and sex differed slightly from primary analyses but trended in the same direction; however, the risk of binge drinking did not significantly differ by poverty level (appendix pp 8–9). Differences in prevalence and risk of several outcomes also varied by education, marital status, and caste (appendix pp 22–23). Using the World Bank MPM to measure poverty, the prevalence of extreme poverty was 21·3% (95% CI 20·9–21·7; approximately 54 million people; appendix p 24). Results of descriptive analyses using the MPM and results from regression models comparing the relative risk of outcomes across sociodemographic subgroups are presented in the appendix (pp 25–32). There were some differences in point estimates, but findings from sensitivity analyses were largely consistent with primary analyses.
Discussion
We found that people in extreme poverty (ie, consuming ≤$1·90 per day) in India have a high prevalence of hypertension (28·7%), tobacco use (40·3%), and frailty (39·0%). The prevalence of most outcomes in this population varied considerably across states and territories and by sex and residence. Relative to those not in poverty, middle-aged and older adults at all poverty levels had a lower risk of diabetes, general obesity, and depression, and a higher risk of tobacco use and vision impairment. As one of the first studies to examine the prevalence of a wide range of NCD risk factors and chronic conditions among adults in extreme poverty, these findings have implications for policy makers.
The high prevalence of tobacco use in this population highlights an urgent need to strengthen tobacco control policies. Although India has introduced tobacco control schemes such as the National Tobacco Control Program, the lack of substantial reductions in tobacco use21 and variable performance of tobacco control efforts at the subnational level22 suggest room for improvement in the implementation of these measures. Tobacco taxation is a powerful policy tool but has thus far been underutilised.23 Raising taxes on tobacco products could be especially effective in curbing use among those in poverty as this population is highly price sensitive.24
There were substantial geographical variations in the prevalence of NCD risk factors and chronic conditions, which highlights the need for state-specific policies. For instance, eastern and central states, such as Jharkhand and Chhattisgarh, could consider prioritising alcohol prevention efforts, whereas more socioeconomically developed states in the south and north need to address the high prevalence of diabetes and obesity. We also documented variations in risk across sociodemographic subgroups of this population, which can further inform targeting of policies. For example, the higher risk of chronic conditions among females could be addressed by integrating screening for conditions such as depression and eye diseases into women’s health services. Urban residents in extreme poverty had a higher risk of metabolic risk factors than rural residents, whereas rural residents had a higher risk of behavioural risk factors. These findings can guide priority setting and resource allocation for the National Urban and Rural Health Missions.
International comparisons further affirm the importance of tailoring polices to local context; a review of regional cardiovascular disease drivers found that in south Asia, urbanisation and shifting dietary patterns are crucial targets for NCD interventions, whereas in African settings, the lack of health care infrastructure is more salient.11 India’s health policy framework must reflect these intracountry and intercountry variations to maximise the efficiency of interventions.
We found a lower risk of diabetes and general obesity for participants at all poverty levels compared with those not in poverty, with the risk decreasing as poverty levels increased. However, for hypertension, the risk was lower only for those at the highest poverty level. These findings suggest that as people move out of extreme poverty, their risk of metabolic risk factors approaches the risk in those not in poverty, highlighting an urgent need to expand NCD prevention and management efforts. The Health and Wellness Centres (now transformed into Ayushman Arogya Mandir25), that are part of the Ayushman Bharat scheme, include NCD management in their package of services.26 However, some evidence shows barriers (eg, shortage of testing supplies, lack of knowledge among health workers, and poor referral mechanisms) to the delivery of NCD care in Health and Wellness Centres, which need to be addressed.27
The national NCD policy priorities can also draw on evidence from other global contexts. For instance, the global study that documented high hypertension prevalence in extreme poverty also found extremely low levels of diagnosis and treatment, suggesting major gaps in health-care access and affordability for this group.3,28 This finding is consistent with our findings and strengthens the case for investing in community-based screening and integrated primary care services for NCDs in underserved populations.
The higher burden of vision impairment and frailty in this population also merits action. Community-based outreach programmes have been successful in reducing inequities in access to eye care in Tamil Nadu and could serve as a model for other states.29 Similar programmes to screen for frailty could help identify vulnerable adults and facilitate linkages to services to address their needs. Adequate implementation and scaling up of existing policies, such as the National Policy for Older Persons, are needed to meet the needs of older adults in poverty.30
This study had some limitations. For states and territories with small sample sizes of participants in extreme poverty, prevalence could not be reliably estimated or estimates had lower precision, reflected in wider CIs. Alcohol use, tobacco use, depression (scale using self-reported items), frailty (scale derived from self-reported items) and chronic pain were all based on self-reported data. Alcohol and tobacco use could be subject to under-reporting due to social desirability bias. Participants who completed the physical measures and biomarker modules differed on some sociodemographic characteristics (eg, rural vs urban residence) to those who did not complete them, which might result in selection bias. However, differences for most characteristics were small and the survey and DBS weights mitigate the effect of selection bias. The outcomes examined here are not exhaustive; data on some important risk factors were not available (eg, cholesterol, diet, and infections) or were not measured in sufficient detail (eg, physical activity and hearing). We also did not examine NCDs directly, as they were measured via self-report, and self-report substantially underestimates disease prevalence among impoverished populations.31
This study also had several strengths. To our knowledge, this is one of the first studies to estimate the prevalence of a wide range of NCD risk factors and chronic conditions in a nationally representative sample of middle-aged and older adults in extreme poverty. We defined poverty using an absolute measure, which overcomes methodological limitations of previous studies and enables comparisons over time and across countries. To the extent possible, we used physical or scale-based measures to measure outcomes. Additionally, results from our primary analyses were largely consistent with our sensitivity analyses, which strengthens confidence in our findings.
The prevalence of NCD risk factors and chronic conditions among middle-aged and older adults in extreme poverty varied substantially across states and territories and sociodemographic subgroups. Our subnational estimates can guide state-level planning and resource allocation decisions and our findings documenting differences in the burden across sociodemographic subgroups can inform policies to address these disparities.
Supplementary Material
Research in context.
Evidence before this study
We conducted a review of PubMed and Google Scholar, using the search terms “socioeconomic status”, “poverty”, “noncommunicable diseases”, “cardiovascular disease”, “hypertension”, “diabetes”, “obesity”, “depression”, “frailty”, “vision impairment”, “pain”, and “India” on July 30, 2025, with no restrictions on date or language. We found several studies that were conducted in single sites and states in India. A few studies reported the prevalence of some non-communicable disease (NCD) risk factors by socioeconomic status, using the nationally representative National Family Health Survey (NFHS). However, the NFHS sample excludes older adults, who have the highest prevalence of NCDs and other chronic conditions. Studies used heterogeneous relative measures of poverty and socioeconomic status. We did not find any nationally representative study in India that reported the prevalence of NCD risk factors and other chronic conditions among middle-aged and older adults in extreme poverty, using an absolute measure of poverty.
Added value of this study
To our knowledge, this is one of the first studies to provide nationally representative prevalence estimates for a wide range of NCD risk factors and chronic conditions among middle-aged and older adults in extreme poverty in India. We used physical measures or validated scales to measure most outcomes. We defined poverty levels using the World Bank’s international poverty lines, an absolute measure of poverty, which enables comparisons of the prevalence of these diseases in this population over time and across settings. We also conducted sensitivity analyses using the World Bank’s multidimensional poverty measure, an alternative absolute measure of poverty.
Implications of all the available evidence
Middle-aged and older adults in extreme poverty in India have a notable prevalence of some key NCD risk factors and chronic conditions, and prevalence varied substantially across states and territories and sociodemographic subgroups. These findings can guide national-level and state-level policies and programmes to improve the health and quality of life of people in extreme poverty and address sociodemographic disparities in the burden of these diseases. As India is home to the largest share of the global population in extreme poverty in this age group, these findings have implications for global health and sustainable development goals.
Acknowledgments
This analysis uses data or information from the Harmonized LASI dataset and Codebook, version A.3, developed by the Gateway to Global Aging Data (https://doi.org/10.25549/h-lasi). The development of the Harmonized LASI was funded by the National Institute on Aging (R01 AG042778, 2R01 AG030153, and 2R01 AG051125). LASI is funded by the Ministry of Health and Family Welfare, Government of India, the National Institute on Aging (R01 AG042778 and R01 AG030153), and UN Population Fund, India. DF is supported by funding from the National Heart, Lung, And Blood Institute of the National Institutes of Health under award number K23HL161271. KS is supported by the Fogarty International Center, National Institutes of Health under award number 1K43TW011164. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. PG is a Chan Zuckerberg Biohub investigator.
Footnotes
Declaration of interests
We declare no competing interests.
Contributor Information
Nandita Krishnan, Department of Medicine, Division of Primary Care and Population Health, Stanford University School of Medicine, Stanford, CA, USA.
Hunter Green, Center for Economic and Social Research, University of Southern California, Los Angeles, CA, USA.
Jinkook Lee, Center for Economic and Social Research, University of Southern California, Los Angeles, CA, USA; Department of Economics, University of Southern California, Los Angeles, CA, USA.
David Flood, Department of Internal Medicine, University of Michigan, Ann Arbor, MI, USA; Institute for Health Policy and Innovation, University of Michigan, Ann Arbor, MI, USA.
Kavita Singh, Heidelberg Institute of Global Health, Medical Faculty, Heidelberg University, Germany; Centre for Chronic Disease Control, New Delhi, India.
T V Sekher, International Institute for Population Sciences, Mumbai, India.
David E Bloom, Department of Global Health and Population, Harvard T H Chan School of Public Health, Boston, MA, USA.
Pascal Geldsetzer, Department of Medicine, Division of Primary Care and Population Health, Stanford University School of Medicine, Stanford, CA, USA; Chan Zuckerberg Biohub, San Francisco, CA, USA.
Data sharing
Harmonized LASI wave 1—version A.3 data can be accessed at https://doi.org/10.25549/h-lasi. LASI survey data is also available with IIPS (https://www.iipsindia.ac.in/content/LASI-data). Shapefiles for creating maps were obtained from DataMeet, accessible at http://projects.datameet.org/maps/states/. Code used for analyses will be uploaded to a public repository upon article acceptance.
References
- 1.United Nations. Transforming our world: the 2030 Agenda for Sustainable Development. 2015. https://sdgs.un.org/2030agenda (accessed Oct 7, 2022).
- 2.Gwatkin DR, Guillot M, Heuveline P. The burden of disease among the global poor. Lancet 1999; 354: 586–89. [DOI] [PubMed] [Google Scholar]
- 3.Geldsetzer P, Tisdale RL, Stehr L, et al. The prevalence of cardiovascular disease risk factors among adults living in extreme poverty. Nat Hum Behav 2024; 8: 903–16. [DOI] [PubMed] [Google Scholar]
- 4.Clock WP. World Data Lab. https://worldpoverty.io/map (accessed Aug 1, 2025).
- 5.WHO. Noncommunicable diseases India 2018 country profile. Geneva: World Health Organization, 2018. [Google Scholar]
- 6.Subramanian SV, Corsi DJ, Subramanyam MA, Smith GD. Jumping the gun: the problematic discourse on socioeconomic status and cardiovascular health in India. Int J Epidemiol 2013; 42: 1410–26. [DOI] [PubMed] [Google Scholar]
- 7.Prabhakaran D, Jeemon P, Reddy KS. Poverty and cardiovascular disease in India: do we need more evidence for action? Int J Epidemiol 2013; 42: 1431–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Corsi DJ, Subramanian SV. Socioeconomic gradients and distribution of diabetes, hypertension, and obesity in India. JAMA Netw Open 2019; 2: e190411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Luhar S, Mallinson PAC, Clarke L, Kinra S. Trends in the socioeconomic patterning of overweight/obesity in India: a repeated cross-sectional study using nationally representative data. BMJ Open 2018; 8: e023935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Prince MJ, Wu F, Guo Y, et al. The burden of disease in older people and implications for health policy and practice. Lancet 2015; 385: 549–62. [DOI] [PubMed] [Google Scholar]
- 11.Bhuiyan MA, Galdes N, Cuschieri S, Hu P. A comparative systematic review of risk factors, prevalence, and challenges contributing to non-communicable diseases in South Asia, Africa, and Caribbeans. J Health Popul Nutr 2024; 43: 140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Perianayagam A, Bloom D, Lee J, et al. Cohort profile: the longitudinal ageing study in India (LASI). Int J Epidemiol 2022; 51: e167–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Bloom D, Perry Hu P, Arokiasamy P, et al. Longitudinal aging study in India: biomarker data documentation. 2014. https://www.rand.org/pubs/working_papers/WR1043.html (accessed Aug 29, 2022).
- 14.World Bank. Poverty and shared prosperity 2018: piecing together the poverty puzzle. World Bank, 2019. [Google Scholar]
- 15.Ezzati M, Riboli E. Behavioral and dietary risk factors for noncommunicable diseases. N Engl J Med 2013; 369: 954–64. [DOI] [PubMed] [Google Scholar]
- 16.Misra A, Chowbey P, Makkar BM, et al. Consensus statement for diagnosis of obesity, abdominal obesity and the metabolic syndrome for Asian Indians and recommendations for physical activity, medical and surgical management. J Assoc Physicians India 2009; 57: 163–70. [PubMed] [Google Scholar]
- 17.Kessler RC, Andrews G, Mroczek D, Ustun B, Wittchen HU. The World Health Organization composite international diagnostic interview short-form (CIDI-SF). Int J Methods Psychiatr Res 1998; 7: 171–85. [Google Scholar]
- 18.Saliba D, Elliott M, Rubenstein LZ, et al. The Vulnerable Elders Survey: a tool for identifying vulnerable older people in the community. J Am Geriatr Soc 2001; 49: 1691–99. [DOI] [PubMed] [Google Scholar]
- 19.Chien S, Young C, Meijer E, et al. Harmonized LASI, Version A.3. The Gateway to Global Aging Data. 2023. https://lasi-dad.org/codebooks/Harmonized%20LASI-DAD%20A.3.pdf (accessed Aug 1, 2025). [Google Scholar]
- 20.Diaz-Bonilla C, Sabatino C. April 2022 update to the multidimensional poverty measure: what’s new. Global Poverty Monitoring Technical Note. World Bank. 2022. [Google Scholar]
- 21.Nazar GP, Chang KC, Srivastava S, Pearce N, Karan A, Millett C. Impact of India’s National Tobacco Control Programme on bidi and cigarette consumption: a difference-in-differences analysis. Tob Control 2020; 29: 103–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kaur J, Jain DC. Tobacco control policies in India: implementation and challenges. Indian J Public Health 2011; 55: 220–27. [DOI] [PubMed] [Google Scholar]
- 23.Kaur J, Rinkoo AV, Gouda HN, Prasad V, Pendse RN. Implementation of MPOWER package in the South-East Asia region: evidence from the WHO report on the global tobacco epidemic. Asian Pac J Cancer Prev 2021; 22: 71–80. [DOI] [PubMed] [Google Scholar]
- 24.Global Tobacco Economics Consortium. The health, poverty, and financial consequences of a cigarette price increase among 500 million male smokers in 13 middle income countries: compartmental model study. BMJ 2018; 361: k1162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mandir AA. National portal of India. https://www.india.gov.in/website-ayushman-arogya-mandir (accessed Aug 1, 2025).
- 26.Health Lahariya C. & wellness centers to strengthen primary health care in India: concept, progress and ways forward. Indian J Pediatr 2020; 87: 916–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Brar S, Purohit N, Singh G, Prinja S, Kaur M, Lakshmi PVM. Health system readiness for roll out of the Ayushman Bharat health and wellnes centres—early experiences from Punjab state. J Family Med Prim Care 2022; 11: 1354–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sekher TV, Flood D, Green H, et al. Prevalence, awareness, treatment, and control of diabetes in India: a nationally representative survey of adults aged 45 years and older. Lancet Glob Health 2025; 13: E1543–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Namperumalsamy P Maintaining quality in community eye care—the Aravind model. Indian J Ophthalmol 2020; 68: 285–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Paul NS, Asirvatham M. Geriatric health policy in India: the need for scaling-up implementation. J Family Med Prim Care 2016; 5: 242–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Vellakkal S, Subramanian SV, Millett C, Basu S, Stuckler D, Ebrahim S. Socioeconomic inequalities in non-communicable diseases prevalence in India: disparities between self-reported diagnoses and standardized measures. PLoS One 2013; 8: e68219. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Harmonized LASI wave 1—version A.3 data can be accessed at https://doi.org/10.25549/h-lasi. LASI survey data is also available with IIPS (https://www.iipsindia.ac.in/content/LASI-data). Shapefiles for creating maps were obtained from DataMeet, accessible at http://projects.datameet.org/maps/states/. Code used for analyses will be uploaded to a public repository upon article acceptance.