Can the rural poor in India afford to treat non-communicable diseases


Corresponding Author David M. Dror, Micro Insurance Academy, 52B Okhla Industrial Estate Phase III, New Delhi 110020, India. Tel.: +91 99 5820 6633; Fax: +91 11 4379 9117; E-mail:


Objective  Non-communicable diseases (NCD) are on the increase in low-income countries, where healthcare costs are paid mostly out-of-pocket. We investigate the financial burden of NCD vs. communicable diseases (CD) among rural poor in India and assess whether they can afford to treat NCD.

Methods  We used data from two household surveys undertaken in 2009–2010 among 7389 rural poor households (39 205 individuals) in Odisha and Bihar. All persons from the sampled households, irrespective of age and gender, were included in the analysis. We classify self-reported illnesses as NCD, CD or ‘other morbidities’ following the WHO classification.

Results  Non-communicable diseases accounted for around 20% of the diseases in the month preceding the survey in Odisha and 30% in Bihar. The most prevalent NCD, representing the highest share in outpatient costs, were musculoskeletal, digestive and cardiovascular diseases. Cardiovascular and digestive problems also generated the highest inpatient costs. Women, older persons and less-poor households reported higher prevalence of NCD. Outpatient costs (consultations, medicines, laboratory tests and imaging) represented a bigger share of income for NCD than for CD. Patients with NCD were more likely to report a hospitalisation.

Conclusion  Patients with NCD in rural poor settings in India pay considerably more than patients with CD. For NCD cases that are chronic, with recurring costs, this would be aggravated. The cost of NCD care consumes a big part of the per person share of household income, obliging patients with NCD to rely on informal intra-family cross-subsidisation. An alternative solution to finance NCD care for rural poor patients is needed.


Objectif:  Les maladies non transmissibles (MNT) sont en augmentation dans les pays à faibles revenus, où les coûts des soins de santé sont payés la plupart du temps de la poche. Nous étudions la charge financière des MNT contre celle des maladies transmissibles (MT) entre zones rurales pauvres en Inde et évaluons si elles peuvent se permettre de traiter les MNT.

Méthodes:  Nous avons utilisé les données de deux enquêtes auprès des ménages menées en 2009-2010 chez 7.389 ménages ruraux pauvres (39,205 personnes) à Odisha et dans le Bihar. Toutes les personnes dans les ménages échantillonnés, quelque soit l’âge et le sexe, ont été incluses dans l’analyse. Nous classons les maladies auto déclarées comme MNT, MT ou ‘autres morbidités’ selon la classification de l’OMS.

Résultats:  les MNT représentaient environ 20% des maladies dans le mois précédant l’enquête à Odisha et 30% dans le Bihar. Les MNT les plus répandues, représentant la plus grande part des coûts en ambulatoire, étaient des maladies musculo-squelettiques, digestives et cardiovasculaires. Les problèmes cardiovasculaires et digestifs généraient également des coûts d’hospitalisation plus élevés. Les femmes, les personnes âgées et les membres des ménages moins pauvres rapportaient une prévalence plus élevée des MNT. Les coûts en ambulatoire (consultations, médicaments, examens de laboratoire et d’imagerie) représentaient une part plus importante des revenus pour les MNT que pour les MT. Les patients avec des MNT étaient plus susceptibles de déclarer une hospitalisation.

Conclusion:  Les patients avec des MNT dans les zones rurales pauvres en Inde payent beaucoup plus que les personnes ayant des MT. Pour les cas des MNT chroniques, avec des coûts récurrents, cela serait aggravé. Le coût des soins des MNT consomme une grande partie des parts des revenus du ménage par personne, obligeant les patients avec des MNT à compter sur diverses subventions intrafamiliales informelles. Une solution alternative pour financer les soins des MNT pour les patients pauvres en milieu rural est nécessaire.


Objetivo:  Las enfermedades no transmisibles (ENTs) están aumentando en los países de renta baja, en donde los cuidados sanitarios se pagan principalmente del bolsillo del paciente. Hemos investigado la carga financiera de las ENTs versus las enfermedades transmisibles (ETs) entre la población rural de la India, y evaluado si pueden permitirse el tratamiento para las ENTs.

Métodos:  Hemos utilizado datos recogidos durante dos estudios realizados entre el 2009-2010 en 7,389 hogares rurales pobres (39,205 individuos) en Odisha y Bihar. E n el análisis se incluyeron todas las personas que vivían en los hogares muestreados, independientemente de la edad o del género. Clasificamos las enfermedades autoreportadas como ENTs, ETs u “otras morbilidades” según la clasificación de la OMS.

Resultados:  Las ENTs eran responsables de alrededor del 20% de las enfermedades ocurridas en el mes anterior al estudio en Odisha y del 30% en Bihar. Las ENTs más prevalentes, representando la mayor parte en los costes para los pacientes, eran las enfermedades musculoesqueléticas, digestivas y cardiovasculares. Los problemas cardiovasculares y los digestivos también generaban los costes de ingresos más altos. Las mujeres, las personas mayores y los hogares menos pobres reportaban la mayor prevalencia de ENTs. Los costes de consultas externas (consultas, medicamentos, pruebas de laboratorio y pruebas de imagen) representaban una mayor parte de gastos para ENTs que para ETs. Los pacientes con ENTs tenían una mayor probabilidad de reportar una hospitalización.

Conclusión:  Los pacientes con ENTs en asentamientos rurales pobres en la India pagan mucho más que las personas con ETs. En los casos en los que las ENTs son crónicas, con costes recurrentes, esta situación es aún más grave. Los costes de cuidados para ENTs consumen una gran parte de los ingresos por persona de un hogar, obligando a los pacientes con ENTs a depender del subsidio intra –familiar informal. Se requiere una solución alternativa para financiar los cuidados de ENTs de pacientes pobres viviendo en áreas rurales.


Non-communicable diseases (NCD) are leading causes of death globally, but nearly 80% of NCD deaths occurred in low- and middle-income countries (World Health Organization, 2011). In 2004, nearly 60% of deaths worldwide were because of NCD, and around 23% because of communicable diseases (CD) (World Health Organization, 2008). In India, these percentages were 50% and 28%, respectively (World Health Organization, 2004). NCD are also responsible for the biggest part of disease burden as assessed by disability-adjusted life-years (DALYs),1 both worldwide and in the south-east Asian region (48% and 44%, respectively, in 2004) (World Health Organization, 2008). NCD deaths are projected to increase by 15% globally between 2010 and 2020; in south-east Asia, this increase could even be more than 20% and is a major barrier to the development (World Health Organization, 2011).

In India, around 70% of total health expenditure is paid out-of-pocket (World Health Organization, 2011) and caused some 7% of rural households to drop below poverty line in 1 year (Berman et al. 2010). Any intervention aiming to reduce financial shocks because of illness needs information on prevalence and cost. As a higher burden of NCD is unavoidable, better insights are needed on the morbidity of NCD and the changing implications for out-of-pocket healthcare expenditures.

Most of the literature on NCD deals with reducing the impact of NCD by addressing a number of risk factors underlying most high-burden NCD (e.g. tobacco use, physical inactivity, unhealthy diet, alcohol abuse). The aim of this literature is to have countries address these risk factors in prevention programs and reduce the burden of these NCD. The burden is investigated mainly through mortality (premature death) and disability (DALY) because of NCD (Murray & Lopez 1997; Boutayeb & Boutayeb 2005; Lopez et al. 2006; Kinra et al. 2010; Beaglehole et al. 2011; Dans et al. 2011; World Health Organization, 2011). However, the information about the number of patients afflicted with NCD and the financial consequence of NCD is sorely rare. Some information is available about the prevalence of certain NCD but not all. The rare studies that deal with the financial burden, also deal with specific NCD. We found one study looking at cost implications of NCD for all India (Engelgau et al. 2012), based on data from 1995 to 1996 and 2004 of the National Sample Survey Organization that provides important insights. For example, in 2004, almost 50% of out-of-pocket health expenses were on NCD and medicines represented 46% of total out-of-pocket health spending (Engelgau et al. 2012). To the best of our knowledge, no study examined the financial implications of NCD amongst the most vulnerable segment of society in India: the rural poor.

The purpose of this article is to examine the financial implications of NCD relative to CD among rural poor communities in two states of India (Odisha and Bihar), taking into account the prevalence of illness categories, and age and income of the ill persons.

Data and methods

Setting and sampling

We used data from two household surveys conducted in rural areas of Odisha (Kalahandi, Khorda and Malkangiri districts, January 2009) and Bihar (Gaya district, May–June 2010). Average monthly per capita consumer expenditure for rural Odisha (INR559, PPP$35.5) is lowest of all states in India; average expenditure for rural Bihar (INR598, PPP$38.0) is fourth lowest (after Odisha, Chhattisgarh and Jharkhand) (National Sample Survey Organization, 2010). These studies formed part of a baseline study to initiate rural micro-health insurance programmes, and the locations were selected in agreement with implementing local non-government organizations (NGOs).2

We followed a two-stage sampling procedure in each state. In stage 1, 130 villages (27 in Kalahandi, 22 in Khorda, 31 in Malkangiri and 50 in Gaya) were selected randomly from lists that local NGOs provided where they organised self-help groups (SHGs)3; in Gaya, 50 additional villages were selected randomly from Census 2001 registry. In stage 2, households in each village were sampled randomly by applying the ‘four winds’ (or ‘line sampling’) technique (Som 1996). Only households where an adult was available at the time of the survey were sampled.4 Households were sampled in two cohorts of equal size: member (if at least one person participated in an SHG linked to partner-NGOs) and non-member households. Sample size was 5383 households in Odisha (25 885 individuals) and 2006 households in Bihar (13 320 individuals). Hundred percentage of the sample was rural.

The survey questionnaire was translated from English into local languages (Oriya and Hindi), back translated for verification, and pre-tested in each state among 80 households. Surveyors fluent in local dialects conducted the interviews. In principle, one household member reported for all household members, but usually more persons were present to help. We obtained informed consent of the respondents and kept confidential participants’ names in data recording and analysis. This research project met all the requirements of the funding agency (NWO-WOTRO) on ethical issues arising in social science research.


The data collected included general demographics of household members (age, gender, education, economic activity), distance to their habitual primary care practitioner (min) and information on household expenditures. We obtained an income-proxy through questions on many items of household expenditure, following the method of the Indian National Sample Survey Organization (2008). Our ‘income-proxy’ is the monthly per capita consumer expenditure excluding healthcare costs, because we seek to identify patterns of financing healthcare (Flores et al. 2008; Wagstaff 2008).

The household survey also included questions on illness episodes and illness-related outpatient expenditures of household members during 1 month preceding the survey, and hospitalisations (exceeding 24 h) with related costs during 1 year preceding the survey. Information about the episodes was queried through open questions on the illness or symptoms that caused the episode. Illnesses that were clearly different or occurred at different periods were treated as separate episodes. Information about per-episode outpatient expenditures was collected, including consultation fees of qualified allopathic practitioners, and expenditures for medicines, laboratory and imaging tests. The self-reported illnesses or symptoms were categorised, following the WHO classifications (World Health Organization, 2008), as specific categories of NCD, CD and ‘other morbidities’ (injuries, maternal, perinatal or nutritional conditions). Ailments that could not be well defined based on the reported symptoms were categorised as ‘missing’. This classification of illnesses was carried out by two of the authors with some medical background, in consultation with a clinician who practiced in rural India.


All persons from the sampled households, irrespective of age and gender, were included in the analysis. The significance of difference between means was tested by a two-tailed t-test and the significance of difference of frequencies was established by chi-square tests. We aggregated the member and non-member sub-cohorts in each state for the purpose of the analysis reported here as we found no significant difference in frequency of reported illness episodes.

We measured the inequality of reported illness episodes by income proxy, using concentration indices. An index of zero indicates there is no inequality, a negative (positive) index a disproportionate concentration of the health indicator among the poor (rich) (O’Donnell et al. 2008). We show two versions of the index: the index as described by O’Donnell et al. (2008) and a corrected index as suggested by Erreygers (2009). We used STATA version 11.1 (StataCorp, Texas, USA) for all the analyses.

All amounts, reported in Indian rupee (INR), were converted into international dollars (purchasing power parity, PPP$) using the exchange rate of PPP$1 = INR 16.692 for 2009 and 18.073 for 2010. The amounts from the National Sample Survey Organization report of 2009/2010 were converted using the average exchange rate for the 2 years (17.383). For the report of 2007/2008, this rate was 15.727 (15.323 + 16.13/2) (International Monetary Fund, 2012).


Socio-economic status of studied populations.  In both populations, the income proxy is on average around the poverty line: about PPP$1 per person per day in Odisha and PPP$1.5 in Bihar. Considering the relevant data from the Indian National Sample Survey Organization (NNSO), we found that our sampled populations were comparable to the rural averages of their states. Most people depend on hard physical labour for their income as daily wage labourers (e.g. in agriculture, load carrying, construction) or self-employment in agriculture (39.5% in Odisha and 45.3% in Bihar). A minority (12.8% in Odisha and 6.5% in Bihar) is employed in what can be considered sedentary work (salaried employee or self-employed in business/trade, mainly shopkeepers, teachers). In both locations, 46% of the adult population did not have any education. In Bihar, 41% of the adult population have finished class 6 and higher compared to 36% in Odisha. This information is shown in Table 1.

Table 1.   Socio-economic and demographic status of studied populations
 Odisha (N = 25 885)NSSO*
Rural Odisha
Bihar (N = 13 320)NSSO*
Rural Bihar
  1. PPP, purchasing power parity.

  2. *National statistics data from National Sample Survey Organization of India (NSSO). Information shown for rural Odisha is from 2007/2008 (reports 530 and 531) and for rural Bihar from 2009/2010 (reports 537 and 538), periods that correspond best to the time when our data were collected.

  3. †Adult population: N = 17 716 in Odisha; N = 7954 in Bihar.

  4. ‡Other can mean going to school, being unemployed, disabled or pensioned.

Gender% of sampled population
 Male50.6 51.0 
 Female49.4 49.0 
 Missing0.05 0.0 
Age distribution
 Infant (under 5)8.99.311.811.7
 Child (5–14 years old)22.721.928.528.8
 Young adult (15–29 years old)26.025.825.323.1
 Adult (30–44 years old)22.021.317.520.4
 Midlife (45–59 years old)
 Old age (over 60)
Education of adult population (15 years and older)†
 No education45.7 45.8 
 Class 1–517.0 12.2 
 Class 6–1029.3 32.6 
 Class 11 and higher6.5 8.3 
 Missing1.5 1.1 
Economic activity of adult population (15 years and older)†
 Daily wage labourer20.7 32.7 
 Self-employed in agriculture18.2 12.2 
 Self-employed in business/trade9.4 3.6 
 Regular salaried employee3.2 2.8 
 Domestic duties32.2 28.1 
 Other‡14.6 19.7 
 Missing1.8 1.0 
Illness episode in month preceding survey17.0 31.0 
Hospitalised (for more than 24 h) in year preceding survey5.6 2.5 
 Mean (±SE)
Distance to preferred primary care practitioner (in minutes)29.95 (±0.18)
(1.2% missing)
 18.21 (±0.16)
(0.0% missing)
Income proxy per person per month (PPP$)30.09 (±0.13)
(4.7% missing)
35.5443.84 (±0.28)
(0.5% missing)

Around 17% of the surveyed individuals in Odisha and 31% in Bihar reported an illness episode in the month preceding the survey. Nearly 5.5% of the sampled population in Odisha and 2.5% in Bihar incurred a hospitalisation in the year preceding the survey. Respondents in Bihar reported that they needed 18 min on average to reach their preferred primary practitioner and in Odisha, 30 min.

Prevalence and classification of self-reported illness episodes.  In rural Odisha or Bihar, there are no official records about illness or treatment in the population. Our source of information on morbidity to classify illnesses as ‘NCD’, ‘CD’ or ‘other morbidity’ (Table 2) has therefore been the self-reported symptoms from our surveys.

Table 2.   Classification of self-reported illness episodes
 % of illness episodes% of NCD
Odisha (N = 4443)Bihar (N = 4217)Odisha (N = 917)Bihar (N = 1253)
  1. NCD, non-communicable diseases; CD, communicable diseases.

  2. *The group ‘other morbidities’ comprised of injuries, maternal, perinatal or nutritional conditions.

  3. †Ailments that could not be well defined based on the reported symptoms were categorised as ‘missing’.

(Malignant) Neoplasms  0.871.92
Diabetes mellitus  2.731.44
Endocrine disorders  0.220.32
Neuropsychiatric conditions  6.223.27
Sense organ diseases  0.111.2
Cardiovascular diseases  20.7210.45
Respiratory diseases  5.674.39
Digestive diseases  23.4535.12
Genitourinary diseases  2.072.47
Skin diseases  5.024.31
Musculoskeletal diseases  28.1432.16
Congenital anomalies  0.440.00
Oral conditions  2.732.47
Rest  1.640.48
Other morbidities*5.2518.36  

Non-communicable diseases accounted for around 20% of all diseases in Odisha and 30% in Bihar. CD were responsible for most of the illness episodes. The three most prevalent categories of NCD in both locations were musculoskeletal, digestive and cardiovascular problems.

The population reporting NCD was significantly older than the population experiencing CD (P < 0.001, t-test, both locations). The average age of patients with NCD was 42.1 (±0.6) and 37.9 (±0.6) years old, while the average age of patients with CD was 27.2 (±0.4) and 21.1 (±0.4) years in Odisha and Bihar, respectively.

Women were more afflicted with an illness (χ2, P < 0.05 in Odisha, P < 0.001 in Bihar). However, this effect was more pronounced for the prevalence of NCD than CD (P < 0.001, χ2, both locations): in Odisha, of those who reported NCD last month, 59.0% were women and of those who reported CD last month 51.1% were women. In Bihar, respectively, 65.3% and 53.9% were women.

Outpatient health-seeking and expenses by illness types.  We explored health-seeking behaviour and costs with three types of healthcare providers: government health facility, private general practitioner (GP) and private specialist (Table 3).

Table 3.   Frequency of access and costs of consultations
Treatment sought with% of ill cases where treatment was sought†Average cost per visit (±SE) (PPP$)
  1. NCD, non-communicable diseases; CD, communicable diseases; PPP, purchasing power parity.

  2. N = 917 NCD cases and 3276 CD cases in Odisha; N = 1253 NCD cases and 2190 CD cases in Bihar. Information on treatment was missing for 23 of the CD cases in Odisha (0.7%); no information was missing for the NCD cases in Odisha or CD and NCD cases in Bihar.

  3. ††Significance of difference in percentage of people that sought treatment with one of the listed providers out of the persons who reported either NCD or CD last month P < 0.01 (chi-square).

  4. †††Significance of difference in percentage of people that sought treatment with one of the listed providers out of the persons who reported either NCD or CD last month P < 0.001 (chi-square).

  5. *Significance of difference in cost between NCD and CD P < 0.05 (t-test).

  6. **Significance of difference in cost between NCD and CD P < 0.01 (t-test).

Government health facility
 NCD68.910.0†††1.43 (±0.40)
(3.7% missing)
1.22 (±0.41)
(0.0% missing)
 CD69.36.01.41 (±0.11)
(4.8% missing)
0.86 (±0.23)
(0.0% missing)
Private general practitioner
 NCD15.9††21.6†††5.09 (±0.67)
(1.5% missing)
6.19 (±0.51)
(0.4% missing)
 CD11.613.65.19 (±0.50)
(3.9% missing)
5.81 (±0.65)
(0.0% missing)
Private specialist
 NCD20.8†††9.6†††8.00 (±1.06)**
(4.5% missing)
11.08 (±1.50)*
(0.0% missing)
 CD6.83.84.66 (±0.56)
(1.0% missing)
6.52 (±0.78)
(0.0% missing)

Patients with NCD accessed healthcare providers more frequently (except government facility in Odisha). This more frequent health seeking is most manifested in consultations with private specialists. Both in government facilities and at private GPs, patients with NCD paid about the same per visit as patients with CD, but private specialists were costlier per visit for patients with NCD.

Outpatient costs often include medicines, laboratory and imaging tests. The average costs for these services are shown in Table 4. Two main findings emerge clearly: outpatient costs were lower in Bihar than in Odisha and patients with NCD had higher costs than patients with CD for all treatment categories. Aggregating all outpatient costs, patients with NCD reported about double the costs of patients with CD, in both locations. Medicines were the costliest item in all cases.

Table 4.   Outpatient expenditures
 Average cost last month (±SE) (PPP$)
  1. NCD, non-communicable diseases; CD, communicable diseases; PPP, purchasing power parity.

  2. †Significance of difference in cost between NCD and CD P < 0.05 (t-test).

  3. ††Significance of difference in cost between NCD and CD P < 0.01 (t-test).

  4. †††Significance of difference in cost between NCD and CD P < 0.001 (t-test).

Consultations with qualified practitioners
 NCD4.86 (±0.52)†††3.30 (±0.30)†††
 CD2.31 (±0.13)1.55 (±0.15)
 NCD45.24 (±2.52)†††12.71 (±1.10)†††
 CD26.31 (±0.87)6.52 (±0.68)
Lab tests
 NCD2.56 (±0.38)††0.76 (±0.19)†
 CD1.92 (±0.09)0.34 (±0.05)
 NCD1.64 (±0.26)†††0.94 (±0.16)†††
 CD0.37 (±0.06)0.14 (±0.04)
Total outpatient expenditures
 NCD53.49 (±2.86)†††
(2.18% missing)
17.27 (±1.28)†††
(0.9% missing)
 CD30.51 (±0.94)
(2.0% missing)
8.45 (±0.75)
(0.4% missing)

Income and the frequency of disease and its cost.  We examined the association between income proxy and reported prevalence of NCD and CD (Table 5). The positive concentration indices describing NCD show a significant ‘pro-rich’ distribution: NCD were reported more frequently by higher income segments of the study populations, more so than CD. This trend was consistent when calculated according to O’Donnell et al. (2008) and according to Erreygers (2009).

Table 5.   Association between income and prevalence of illness
 Concentration Index as described by O’Donnell et al. (2008) (±SE)Concentration Index as described by Erreygers (2009) (±SE)
  1. NCD, non-communicable disease; CDs, communicable diseases.

NCD0.157 (±0.019)0.115 (±0.016)0.022 (±0.003)0.044 (±0.007)
CD0.022 (±0.010)0.027 (±0.011)0.011 (±0.005)0.018 (±0.007)

Having found that the prevalence of NCD was higher amongst the higher income segments of our sampled rural poor and that NCD were costlier than CD, we examined whether the two phenomena cancelled each other out. We therefore calculated outpatient costs (reported for last month) as percentage of income proxy (per person per month), both for NCD and CD cases. The percentage spent by NCD was significantly and materially higher than by CD, (P < 0.001, t-test, both locations). In Odisha, patients with CD spent on average 125.6% (±4.8%) of income proxy while patients with NCD spent on average 180.6% (±10.6%). In Bihar, these percentages were 21.1% (±1.5%) and 41.4% (±3.0%), respectively. Even though NCD were more prevalent among higher income segments, the treatment costs for NCD represented a bigger share of their income. It can be expected that in case of chronic NCD, these expenses will be incurred again in other months during the year.

The percentages above also indicate that patients in Bihar (both CD and NCD) spent a much lower share of income on outpatient healthcare than in Odisha. The difference across locations seems to be because of two reasons: in Bihar (i) average income was higher and (ii) mean outpatient costs were lower.

Inpatient treatment and expenditure.  We also examined the risk and cost of hospitalisations. Persons who reported an NCD (last month) were more likely to be hospitalised (last year) than the sampled population without NCD: in Odisha 15.5%vs. 5.2%, in Bihar 5.0%vs. 2.2% (P < 0.001, chi-square test). The average cost of hospitalisation was comparable: in Odisha PPP$312 (±53) for patients with NCD and PPP$197 (±9) for the sampled population without NCD; in Bihar PPP$425 (±81) and PPP$380 (±40), respectively.

Share of NCD categories in outpatient and inpatient expenditures.  Finally, we analysed which category of NCD was the costliest with respect to overall outpatient and inpatient expenditures (Figure 1). The relative share of costs is determined by the frequency of an NCD category and its severity in terms of aggregated cost.

Figure 1.

 Share of non-communicable disease (NCD) categories in outpatient and inpatient expenditures.

The two NCD categories that contributed most to overall outpatient expenditures (both locations) were digestive and musculoskeletal problems, followed by cardiovascular and neuropsychiatric conditions. With respect to overall inpatient expenditures, digestive and cardiovascular diseases were amongst the three costliest. Neuropsychiatric, genitourinary and musculoskeletal problems were also important contributors to the overall costs in both locations. The cost of malignancies in Odisha seems to be much higher than in Bihar but is probably due to a small number of outliers as only a modest percentage of hospitalisations was because of cancers (9% in Odisha, 6% in Bihar).


The analysis of the financial burden of NCD on rural poor households in two states in India presented here differs from much of the literature on NCD in that it considers the information on prevalence and treatment cost of NCD, in general, and by the category of NCD, rather than deaths or DALYs associated with NCD. We found that, while CD represent the major share of self-reported illnesses, the prevalence of NCD is considerable, with about 20% of illness episodes in Odisha and 30% in Bihar. With increasing age (Lee & Mason 2011) and with a higher prevalence of NCD amongst the elderly, this percentage is likely to increase in the future.

Amongst the NCD, the three most prevalent in our study populations were musculoskeletal, digestive and cardiovascular diseases. These categories also represented the highest share of outpatient expenditures. Cardiovascular and digestive diseases generated the highest inpatient costs. The morbidity profile usually associated with NCD (cardiovascular problems, diabetes, cancers and chronic long conditions) (World Health Organization, 2011) is not replicated in our study. Instead, we see higher prevalence of NCD related to lifestyle, occupational and nutritional conditions of the rural poor which are exposed simultaneously to (i) very hard physical labour as daily wage labourers (e.g. in agriculture, load carrying, construction) or self-employment in agriculture (Table 1), (ii) unstable and irregular nutrition because of poverty that can cause cardiovascular diseases (Vorster & Kruger 2007; van Abeelen et al. 2012) and (iii) not very hygienic living conditions (Nath 2003; Khurana & Sen 2008; Asian Development Bank, 2009).

It is commonly claimed that poorer population segments are in worse health, also in low- and middle-income countries (Hosseinpoor et al. 2012). Our results showed, however, that self-reported illnesses, particularly NCD, are more frequent among the wealthier segments of the studied populations. This finding is in accordance with a previous report of positive association between household income and self-reported incidence among poor in India (Dror et al. 2009). Engelgau et al. (2012) also found that in India spending on NCD as share of income increases by income. As we investigate the lowest income segments only, it is possible that members of the poorest households cannot afford to stop daily activities for illnesses perceived as minor, and therefore they do not report them.

Analysis of the expenditure patterns of NCD revealed several important insights. First, patients with NCD sought care with private practitioners (GP or specialist) more frequently than patients with CD, even though these practitioners charge more than government facilities. In fact, patients with NCD even reported higher fees per visit with specialists than patients with CD. The public sector in India, especially the more peripheral facilities, commonly suffers from shortages in qualified staff and supply. The population therefore is sceptical about quality of public sector services (Berman 1998; Peters et al. 2002; Reddy et al. 2011). Viewed in combination with the possibility that patients with NCD consider their illness as requiring more serious diagnosis and treatment than CD, they might decide, despite their poverty, to prefer costlier private care.

Second, patients with NCD reported paying more for all types of outpatient care (consultations, medicines, tests) and medicines came out to be the costliest component of outpatient care. These findings are in line with those of Dror et al. (2008) that medicines were the largest component of healthcare costs for chronic care and of Engelgau et al. (2012) that patients with NCD had to bear higher costs for consultations and medicines than other patients.

Third, despite the higher prevalence of NCD among the less poor of our sampled rural poor, the cost for NCD represented a bigger portion of the patient’s share in household income than CD. In this analysis, we looked at outpatient costs paid out-of-pocket during 1 month only. For chronic NCD [many of the illnesses categorised as NCD can be considered chronic (Beaglehole et al. 2011; Dans et al. 2011; World Health Organization, 2011)], clearly these costs can accumulate to a significant portion of the patient’s share of income over the year. Such high costs we believe can only be met when considerable cross-subsidisation occurs within the household (other household members’ share of income has to be used to pay the patient’s treatment). When such informal cross-subsidisation is unavailable, access to care becomes unaffordable. This would presumably be more likely in small households composed of older persons only, or in poorer households. Indeed the higher reporting of NCD by households with more financial resources strengthens the impression that the diagnosis of and treatment for NCD strongly depends on intra-household cross-subsidisation.

Finally, patients with NCD were more likely to be hospitalised than the population without an NCD. And, even though the cost of a hospitalisation was comparable for NCD and non-NCD cases, this higher incidence of hospitalisations places an additional financial burden on patients with NCD.

At the time our surveys were conducted, the surveyed populations in Odisha and Bihar had practically no access to health insurance. The Government of India introduced hospitalisation insurance [Rashtriya Swasthya Bima Yojana (RSBY)] in 2008, which, however, was not implemented in the studied areas until much later (Ministry of Labour and Employment, India, 2012a). The information on enrolment, renewals and payment of claims is for the time being limited. Moreover, RSBY usually covers only the cost of care in an empanelled hospital. It has been shown here that most of the financial problems of patients with NCD arise from financing outpatient care. As this paper goes to print, the Government of India is piloting the expansion of RSBY to include outpatient care; however, it is unclear whether this pilot will be generalised, or which services would be included under RSBY in the future (Ministry of Labour and Employment, India, 2011, 2012b).

The big differences across the two locations included in this study do not change the general conclusion regarding the unaffordability of NCD care, or regarding the reliance of chronic NCD cases on intra-household subsidisation of their care costs. However, these big differences in reporting illnesses deserve some considerations. The higher income, better education and shorter travel time to primary care practitioners in Bihar, compared to Odisha (Table 1), may have contributed to the big difference in number of reported illness episodes; in Bihar, the sampled population had almost twice as many episodes compared to Odisha. On the other hand, the average treatment costs were higher in Odisha than in Bihar (Table 4) and several reasons could explain this: supply aspects (e.g. fewer providers), clinical aspects (e.g. morbidity that is more expensive to cure, or differences in seasonal morbidity).

This study offers a first estimation of the size of the problem of financing access to NCD-related care. On the basis of the results of this study, more qualitative questions could arise, such as, why is treatment sought with certain providers? Is there non-compliance of treatment and why (not)? These and similar questions could be addressed in a qualitative follow-up study.

A limitation of this study is that the data used for the analyses are self-reported. Reporting of the type, frequency, severity and cost can obviously suffer from recall bias and other biases, for example, underreporting of illnesses perceived as minor especially among the poorer households, or inaccurate reports by one household member of illnesses incurred by others.


Patients with NCD in rural poor settings in India pay considerably more for the treatment than patients with CD. This trend might be aggravated in the case of chronic NCD because of the expected recurrence of outpatient expenditures, which in this study were analysed only for 1 month. It is self-explanatory that chronicity is less likely with CD. This trend is likely to worsen over time because incidence, prevalence and chronicity are expected to increase with changes in lifestyle and extension of life-expectancy (Lopez et al. 2006; World Health Organization, 2008, 2011).

In the absence of any systemic solution (like insurance) to deal with the financing of access to care, we believe patients suffering from NCD rely on intra-household cross-subsidisation to pay for their care. This cross-subsidisation can work for as long as the proportion of NCD cases in the overall household population remains relatively small; but this proportion is likely to increase because of expected decrease in household size (Census of India, 2001) and increase in age. Consequently, the ability of households to continue funding family members requiring NCD care will decrease and could lead to lower drug and treatment compliance. This conclusion clearly points to the urgent need to develop suitable policy solutions which would allow rural poor populations to afford care for NCD even as NCD are expected to increase. Such policy choices would have to cover outpatient care in view of its major contribution to the financial burden of patients with NCD.


  • 1

     Quantification of the burden of disease which simultaneously considers both premature death (number of years of life lost due to premature death) as well as the non-fatal health consequences of disease and injury (number of years lived with a disability) (Murray 1994).

  • 2

     In Odisha the Madhyam Foundation [comprising 11 grassroots NGOs: Parivartan, PUSPAC, SOMKS, SDS, ODC (Malkangiri); Mahashakti Foundation, DAPTA, Lok Yojana, Sanginee (Kalahandi); MVPS, DSS (Khorda)] and in Bihar BASIX.

  • 3

     SHGs represent a unique approach to financial intermediation in communities. The approach combines access to low-cost financial services with a process of self-management and development for the SHG members. SHGs are seen to confer many benefits, both economic and social.

  • 4

     In rural India most of the time an adult is in the house or can be found near the house, and is willing to participate in an interview.


The authors gratefully acknowledge funding support from the Netherlands Organization for Scientific Research (NWO), under WOTRO Integrated Programme W01.65.309.00. Additional funding for the Odisha household survey was obtained from the German Federal Ministry for Economic Cooperation and Development. The authors gratefully acknowledge the logistical and research support from the Micro Insurance Academy (New Delhi) during data collection and field work. Thanks are extended to the NGOs that facilitated contacts with villages. The authors gratefully acknowledge useful suggestions on an early draft given by Dr. Richard Smith (formerly chief editor of BMJ) and Dr. Naomi Levitt (University of Cape Town). The authors affirm that their ability to complete the research as planned was not limited by any agreement with funding organizations and confirm that they had full control of all primary data. Partial and initial analysis of results was presented at ICDDR-B 13th Annual Scientific Conference-ASCON XIII (Dhaka, Bangladesh, 16 March 2011) and One Health Summit (Davos, Switzerland, 21 February 2012).