Non-communicable diseases as a major contributor to deaths in 12 tribal districts in India

Non-communicable diseases (NCDs) accounted for 61.8 per cent of the deaths in India in 20161. The proportion of deaths due to NCDs was 55.1 per cent in the relatively less developed States (Empowered Action Group) compared to 68.5 per cent in the other States in 20161. There were variations across States in terms of the burden of NCDs; nevertheless, all States transitioned from infectious diseases to NCDs as the leading cause of death (COD), including States with a high proportion of the tribal population by 20161. As per the Government of India, there is a list of indigenous groups classified as Scheduled tribes (STs)2. India had 104 million (8.6%) population of STs as per the census of 2011, and nearly 45.3 per cent lived below the poverty line in rural areas2-4. The health system is not fully equipped in the tribal regions to address the population’s healthcare needs. The health facilities in tribal areas had many vacancies for doctors, nurses and health workers, according to the rural health statistics in 20174. Poor socio-economic development and lack of adequate medical infrastructure and human resources make the tribal population vulnerable to poor health outcomes.

Good quality mortality data are essential for determining priorities for populations. The mortality data is unreliable due to a lack of complete registration and low coverage of medical certification of cause of death (COD). The completeness of registration in the Indian Civil Registration system improved between 2005 and 2015 and reached 76.6 per cent by 20155. However, reporting of age and sex was incomplete. In addition, coverage was variable across States, with <50 per cent registration in five States in 20155. Although medical certification of COD has also been implemented in healthcare institutions, coverage was highly variable, ranging from 100 per cent in Goa to five per cent in Bihar in 20196. In addition, a survey of nearly a million households was proposed between 2004 and 2014 to collect detailed information regarding COD7. The study used verbal autopsy (VA) tools, including ascertaining the COD based on an interview with the family members of the deceased. The VA method involves a detailed narrative and a structured questionnaire to collect the required information to assign a probable underlying COD.

In the previous decades, the tribal population had a high burden of infectious diseases such as tuberculosis and malaria8. Several recent studies from various tribal regions in India suggest a high burden of NCD risk factors, potentially leading to a high proportion of deaths due to NCD9-14. Given the poor coverage of death registration and medical certification in tribal areas, there were limited data regarding causes of death that could inform policymakers about changing mortality patterns. We need to understand the disease burden in the tribal districts for realigning resources and planning various public health programmes. Therefore, a study was designed in a sample of predominantly tribal districts to document the mortality pattern. We estimated the proportion of deaths due to NCD among 15 yr and older in 12 predominantly tribal districts in India. We also described the distribution of causes of death among adults above 15 yr of age by the district.

Material & Methods

Study design and population: A community-based cross-sectional survey of deaths occurring 12 months prior to study was conducted in the division of Non-communicable Diseases, ICMR-National Institute of Epidemiology, Chennai, Tamil Nadu. Data was collected in a phased manner during 2015-2018 for deaths among adults more than 15 years of age. This COD survey was conducted in 12 States of India. One district per State was purposively selected for the survey with at least 50 per cent tribal population. The study sites in the northeast (NE) region were Sikkim (all districts), Lunglei (Mizoram), Dhalai (Tripura), East Garo Hills (Meghalaya), Mokokchung (Nagaland), Dhemaji (Assam), Senapati (Manipur) and East Kemang (Arunachal Pradesh). Four other sites representing various geographic regions and diverse tribes were also selected. The sites were Koraput (Odisha), Nicobar (Andaman and Nicobar Islands), Kinnaur (Himachal Pradesh) and Mandla (Madhya Pradesh). The survey was carried out after approval from the Institutional Ethics Committee of ICMR-RMRC, Dibrugarh, ICMR-RMRC, Port Blair and ICMR-NIRTH, Jabalpur. A written informed consent was obtained from the participants, and a thumb impression was taken from participants who could not read or write.

Sample size and sampling procedure: A 30 per cent proportion of deaths due to NCDs was assumed with 95 per cent confidence limits, absolute precision of five per cent and a design effect of 1.4. The estimated sample size was 452 deaths for each study district. A cluster sampling method was used with the village as the primary sampling unit. Panchayats (rural) and wards (urban) were selected as clusters using probability proportional to size using systematic linear sampling from 2011 census data. 45 clusters from each study site were chosen and obtained a list of deaths in the previous year. Although ten deaths per cluster was initially planned, it was not feasible to do the fixed sample size per cluster as several villages had a small population. Therefore, all the deaths were surveyed in the selected clusters. Furthermore, the cluster size was increased by adding adjacent villages if the sample size could not be achieved by covering all deaths in 45 PSUs. Due to the small population size, all the previous year’s deaths were surveyed in Kinnaur, East Kameng and Nicobar districts.

Data collection: The Registrar General of India/Centre for Global Health Research VA tool was used for ascertaining the CODs in the study7.

The list of deaths for the reference period of one year were obtained from the concerned health worker for each cluster. In addition, the list was updated based on additional deaths reported by community members after visiting the area. Other sources were also used to ensure complete coverage, such as data from the local church or funeral sites.

Staff who knew the local language were recruited and trained for data collection. After taking informed consent, the family of the deceased were approached. The semi-structured VA questionnaire was administered in the local language to the family member closest to the deceased before the terminal illness. The trained staff collected information about sociodemographic parameters, place of death, medical history of preexisting diseases, treatment history (medication, hospitalization) and risk behaviours in the five years before death (e.g. smoking and alcohol use). The narrative history in the local language included illness or events leading to the death of the individuals in the chronological order of occurrence. The complete history of symptoms, signs, events, investigations and treatment was obtained so that the medical reviewer gets sufficient information to assign a probable underlying COD.

Data analysis: Two medical doctors assigned the COD as per the International Classification of Diseases (ICDs) 10 code independently. If there was a disagreement, then data were reviewed by the two doctors together to discuss and agree based on the consensus. Data were analyzed and summarized into major categories for the COD based on the ICD, tenth revision15. We estimated the proportions for COD and 95 per cent confidence interval for the broad categories: NCDs, infectious diseases, injuries and others. Proportions of known risk factors, pre-existing conditions and place of treatment before death were also computed. Epi info software version 7 (Centers for Disease Control and Prevention (CDC), Atlanta, USA) for data analysis. We also estimated the age-standardized death rate (ASRD) for NCDs, infectious diseases and others16.

The ICD-10 codes included in the NCD category were circulatory system (I00-I99), neoplasm (C00-D49), chronic respiratory diseases (J00-J99), digestive system disease (K00-K93), kidney disease (N00-N99), diabetes mellitus (E00-E90), nervous system diseases (G0-G99) and mental and behavioural disorder (F00-F99)15. The infectious diseases category included the A00-B99 codes. The major conditions were malaria (B54), tuberculosis (A15.0), infectious gastroenteritis (A09) and acute viral hepatitis (B19.9). The injuries (including external causes) category included S00-Y98 codes15. The conditions which could not be included in the three categories were grouped under not elsewhere classifiable (R00-R99). The conditions included age-related senility (R54), abdominal pain unspecified (R10.9), unspecified fever (R50.9) and unknown aetiology (R99).

Results

Verbal autopsy was conducted for 5292 deaths, of which 62 per cent were male. One-fifth of the deceased were 15-39 years old, and 44 per cent were between 40 and 69 years of age (Table I). Overall, 49 per cent of the deceased had never attended school, and 26 per cent were landlords/farmers. The primary religion was hinduism (50%), followed by christianity (48%). Overall, 77 per cent of the deceased belonged to ST, and 56 per cent were BPL (below poverty line) cardholders. Home was the most common place of death (70%). District hospital (25%), private hospital (20%), PHC/CHC/Rural hospital (19%) and medical college/cancer hospital (9%) were the places where treatment was sought before death. However, one-fourth of the deceased did not seek any treatment for the illness, leading to death. Nearly one-fourth sought care in the private sector in the NE sites compared to only 11 per cent in the other sites (Table I).

Hypertension (29%) was the leading pre-existing disease, followed by chronic respiratory disease/asthma (17%), stroke (12%), heart diseases (11%), cancer (10%) and diabetes (9%) in all States. In NE sites, hypertension was the leading pre-existing disease compared to chronic respiratory diseases/asthma (21%) in the other districts. Overall, 42 per cent of the deceased had smoked five years before death, and 37 per cent of the deceased had used smokeless tobacco. Alcohol consumption was prevalent among 47 per cent of the deceased in the previous five years.

Overall, NCDs accounted for most of the deaths (66%), followed by infectious diseases (15%) and injuries (11%) (Table II). In NE districts, NCDs accounted for 69 per cent of deaths compared to 59 per cent in other districts. Deaths due to infectious and parasitic diseases were higher in the other four sites (18%) than in NE sites (13%). The proportion of NCD deaths ranged from 77 per cent in Mokokchung (Nagaland) to 55 per cent in Kinnaur (Himachal Pradesh) and Koraput (Odisha) sites. The proportion of deaths from injuries was above 10 per cent in four NE sites and the highest in Kinnaur (24%). The two districts with the highest proportion of infectious disease deaths were Koraput (35%) and East Garo hills (18%) (Table II). The Kinnaur (7%), Senapati (9%) and Mokokchung (9%) were the three districts with the lowest infectious disease deaths.

ASRD due to NCDs ranged from 426 in Kinnaur to 756 per 100,000 in the East Garo Hills district (Table III). ASDR due to infectious diseases ranged from 52 in Kinnaur to 288 per 100,000 population in the Koraput district.

Overall, cardiovascular diseases (CVDs) (24%) were the leading cause of death (Table IV). Neoplasms (13%) were the second leading cause of death, and chronic respiratory diseases were the third leading cause of death (10%) (Table IV). HIV/AIDS, tuberculosis (7%) and infective gastroenteritis (5%) were the leading causes of the infectious diseases group. Malaria caused two per cent of deaths. Transport injuries, suicide and interpersonal violence caused 3 per cent of deaths. CVDs was the leading COD in 10 of the 12 sites (Table IV). In East Garo Hills (18%) and Lunglei (26%), neoplasms were the leading COD. The proportion of deaths due to CVDs ranged from 37 per cent in Mokokchung to 16 per cent in East Garo Hills and Kinnaur. Chronic respiratory diseases accounted for 23 per cent of deaths in Dhalai and 18 per cent in the Nicobar district. Digestive system diseases, including liver diseases, caused 17 per cent of the deaths. The proportion of deaths due to kidney disorders was the highest in Mokokchung (7%). Tuberculosis and HIV/AIDS accounted for 14 per cent of the deaths in Koraput, 11 per cent in Nicobar and 10 per cent in East Kemang. Koraput had 15 per cent of the deaths due to infective gastroenteritis and other common infectious diseases. Transport-related injuries caused 11 per cent in Kinnaur and five per cent in East Kemang. Suicides and interpersonal violence accounted for 7 per cent of Kinnaur deaths, five per cent in Dhemaji and four per cent in East Kemang/Sikkim (Table IV).

Discussion

The present study documents NCD as the leading COD across all the study districts in 12 States of India. Our findings were consistent with the State-wise disease burden study, which reported NCDs as the leading cause of death across all Indian States1. The age-standardized NCD mortality rate for various States was within the WHO estimate of 563.3 (406.5-756.4) per 100,000 for India in 201817. The data were collected from diverse tribal populations, confirming the changing disease burden irrespective of geography and cultural differences. Our study provides evidence for the need for investment in the health system to cater to the rising burden of NCDs and the involvement of communities to increase awareness regarding risk factors.

The study districts and States have a variable coverage of death registration and medical certification of COD at health facilities5,6. The survey provided a one-time estimate of the COD. However, it is critical to strengthen the death registration and medical certification of the COD to generate reliable estimates on an ongoing basis. One of the significant findings was that home was the most common death, and one-fourth of the deceased did not seek care before death. Therefore, even if death was registered, the COD was most likely not documented in the health facility. In this context, the VA method adds value as it captures the narrative from close family members regarding symptoms and other pre-existing conditions. The million deaths study developed and used a VA tool in a large sample of deaths in India. The tool was validated in large-scale surveys before deploying in the study7. We used the same tool and collected the data involving local investigators who understood and spoke local languages. Although this approach has several limitations, it allowed us to assign broad categories of COD and understand the patterns in mortality in remote and underdeveloped districts in India with a predominantly tribal population.

CVD, including stroke and heart attacks, led to many deaths and hypertension was the most commonly reported pre-existing condition. Several studies have documented the high burden of CVD and hypertension in the tribal populations. A systematic review of studies including over 64,000 subjects estimated a 16 per cent prevalence of hypertension in tribal populations10. A community survey of 3582 adults 20-70 yr from Kinnaur (Himachal Pradesh) reported a 19.7 per cent prevalence of hypertension, and only 23 per cent of the hypertensives had controlled BP12. Community-based studies from the NE region, such as Mokokchung (Nagaland), also reported a high prevalence of hypertension and low treatment coverage9. A survey from tribal areas in Maharashtra reported stroke as the leading COD18. Our study reiterated that the pattern was consistent across the country’s regions and diverse tribal populations. Improving early detection of hypertension and treatment with low-cost antihypertensives can reduce the risk of stroke and heart attacks. In addition, increasing awareness regarding symptoms, timely referral and well-equipped district hospitals can reduce CVD-associated deaths.

Cancers were the second leading COD in several sites, especially in the NE region. Our findings were similar to the data from cancer registries which reported a high incidence of cancers in the region19. In this study, Mizoram had the highest proportion of deaths due to cancers, consistent with data from the Aizawl cancer registry, which had the highest incidence of the age-adjusted cancer rate of 269.4 per 100,00019. Although the cancer burden is high, the cancer treatment facilities are inadequate in this region. The high burden of cancers and other NCDs in tribal areas is possibly influenced by a high burden of risk factors such as tobacco and alcohol use. In this study, nearly half of the deceased had used tobacco or alcohol in the previous five years. Previous studies have documented tobacco use ranging from 23 to 84 per cent and alcohol use ranging from 17 to 73 per cent in various tribal communities12-14,20,21. A multisectoral approach is hence needed to address the underlying risk factors and improve the screening and treatment of cancers in the tribal districts.

Tuberculosis was the leading COD in the category of infectious diseases. The proportion was relatively higher in Koraput, Nicobar and East Kemang districts. The data from several studies across tribal populations in India documented a high burden of tuberculosis8. Our study noted that deaths also occurred among patients already diagnosed with tuberculosis. Therefore, the interventions in tribal areas need to focus on increasing detection and improving retention in care. In addition to tuberculosis, infectious gastroenteritis and malaria were other important causes of infectious disease related deaths which also need to be addressed. Infectious diseases are on the decline, but focused interventions may be required to accelerate the decline in the associated mortality.

The strength of the study was COD data from predominantly tribal districts from diverse geographic areas collected involving well-trained staff who could understand the local language. The data were collected in a large sample for every district using standardized tools which allowed estimates of the COD at the district level. One of the significant limitations, however, was that the hospital records for all deceased who had sought care before death could not be reviewed. Another limitation was the accuracy of translation of some of the narratives from tribal languages to English, as the descriptions were not as per the classical symptom list. We tried to overcome this challenge by involving local staff who understood local languages.

NCDs were the leading COD among the tribal population in India in the background of the unfinished agenda of control of infectious diseases. Based on the findings of this study, we recommend reorientation of the policies to address the changing scenario, investments and resources to reduce morbidity and mortality due to NCD in the tribal population.