Abstract

This paper attempts to provide a wider understanding of the differentials in reported health status in Kerala, while comparing morbidity in the state with other regions in the Indian subcontinent. Reported morbidity and the duration of life lived with a disease is higher in Kerala. Economic inequalities were found only in late-working ages and the elderly, primarily due to higher prevalence of life style-associated chronic conditions in these two age groups. Significant caste-wise differences among adolescents and prime working ages indicated potential for health problems induced by income deprivation in socially disadvantaged subgroups. Self-reported morbidity was 65% higher than proxy-reported morbidity. Regional differences were significant across all age groups, with high morbidity in the most developed region in the state. Results also suggested the need to factor for self- and proxy-reported status in any analysis of morbidity using similar survey data.

Introduction

Kerala is the most advanced state in India in terms of demographic transition, with mortality levels close to those of developed nations for the last two decades. However, morbidity levels are comparatively higher than elsewhere in India in this highly literate, densely populated and rapidly aging population. The picture continues, with the two recent national level surveys showing reported morbidity in Kerala well above that in the other Indian States, in both rural and urban areas and across all age groups (National Sample Survey Organization [NSSO] 2006; Registrar General of India [RGI] 2007). Here, morbidity information is based on reported morbidity data, which is often questioned for its reliability and accuracy (Murray and Chen 1992). Reported morbidity is associated with biases from various sources, depending on the individual reporting as well as on perceptions regarding good health. Despite the well-known discrepancy in subjective and objective measures of health status/morbidity status (Gumber and Berman 1997), the reported information on health and morbidity obtained in large-scale surveys at the community level continues to dominate the analysis of disparity and inequality in health status in developing counties.

Over the last three decades, a number of researchers have attempted to explain why morbidity is higher in Kerala than elsewhere in India. On one hand, many of the researchers attribute the phenomenon to a higher perception of illness and good health in Kerala society (Kannan et al. 1991; Gumber and Berman 1997; Michael and Singh 2003). Others argue that Kerala has the potential for higher real morbidity (Kumar 1993; Panikar and Soman 1984; Panikar 1999). However, most researchers agree that to some extent exposure to healthcare services has induced health consciousness and has affected perceived morbidity, leading to a higher reporting of illness (Kumar 1993; Dilip 2002).

Morbidity differences similar to those we observed while comparing Kerala with other Indian states are also found within the state. Systematic variations in perception about "good health" are found across different economic subgroups and geographic regions in this study area (Dilip 2002). Morbidity levels are relatively lower among the population residing in poorer households and in less-developed regions than among their counterparts in wealthier households and those residing in developed regions, respectively. This study also documents the widespread seasonal variation in morbidity. Mohindra et al. (2006) showed that the association between socio-economic status and health varied across social groups. They argued that the burden of low socio-economic position coupled with low caste could create a double deficit in reported health status.

However, the nature of the relationship between the risk of reporting as ill and an individual's economic status is still inconclusive. While national-level surveys (NSSO 1998; NSSO 2006) indicate a positive relationship between reported morbidity and economic status, the reverse is observed in state-specific health surveys by the Kerala Shasthra Sahithya Parishad (KSSP) (Kannan et al. 1991; Kunhikannan and Aravindan 2000; KSSP 2006). One probable reason is the methodological variation in measurement of economic status adopted by these two survey agencies. The national-level surveys use consumption-expenditure patterns or household monthly per-capita consumer expenditure to classify whether a household is rich or poor, while the state-specific surveys use (1) per-capita income, (2) per-capita expenditure, (3) housing conditions and (4) ownership of selected durable appliances and goods to classify and identify the household's socio-economic status. Further, researchers comparing or separately analyzing these data sets should note that neither agency accounts for differences in household size while classifying households as rich or poor. The sensitivity of household size to per-capita consumer expenditure and per-capita monthly income is well documented (Deaton and Paxson 1998) but is yet to be addressed while examining the rich-poor divide in morbidity in the state.

There is a clear age pattern in morbidity levels, with disease composition revealing the relatively higher prevalence of acute ailments in younger age groups and of chronic ailments in older ones (NSSO 1998; Dilip 2002; Navaneetham and Kabir 2006). In fact this age differential in morbidity is much greater than some of the most commonly studied gender/class inequalities in reported morbidity. Consequently, an age-group disaggregated analysis is essential for a closer understanding of differential morbidity, but this has not yet been attempted in Kerala.

Methods

As a prologue, we examined how far the age pattern in reported morbidity in Kerala deviates from the national scenario. We contrasted the morbidity pattern in Kerala with that in all India and in Bihar, a state characterized by a low level of literacy and the lowest self-reported morbidity among the major states in India (NSSO 2006). Accounting for the magnitude of variation between self- and proxy-reported morbidity data is another unique feature in this analysis. Self-reporting by all individuals is impossible in surveys of this scale and coverage, and it is important to recognize and factor the deviation between self-reported and proxy-reported information.

In this study, we used an age-wise analysis of differentials in reported morbidity in Kerala based on the National Sample Survey Organisations' (NSSO) 60th round survey Morbidity and Health Care, carried out in 2004. The survey took place between January and June, and covered 2829 households in Kerala. Information on whether an individual was ailing during the last 15 days is available for 13,333 persons residing in these households. For the purpose of the survey, people self-reported a feeling of being ill, or of ailing. As already known, morbidity data obtained through similar health interview surveys are often incomplete due to proxy reporting (Murray and Chen 199; Gumber and Berman 1997). To factor this aspect, our analysis included a variable on whether information on ailment status was self- or proxy-reported.

Reported morbidity levels are assessed on the basis of proportion of persons reported as ailing during the 15-day recall period used in the survey. Hence co-morbidity reported during the survey is not accounted for in this analysis. The study population has been categorized into five age groups to examine the relative effect of certain background variables capable of influencing morbidity levels across these ages. The age groups are: (1) children (0-9 years), (2) adolescents (10-19 years), (3) prime working ages (20-49 years), (4) late working ages (50-59 years) and (5) elderly (60 years and above). Unlike the international criteria of 65 years, a person aged 60 years and above is included in the aged/elderly category at the public policy level in the Indian context (Rajan et al 2003).

First, morbidity in Kerala is compared with the all-India and Bihar scenarios in terms of (1) proportion of population ailing in the reference period, (2) disease composition in terms of the broad acute and chronic ailment classification and (3) median duration of ailment. The acute and chronic ailment classification is based on the general duration of the ailment. Ailments of less than 30 days' duration are treated as acute and those of more than 30 days' duration as chronic (NSSO 1998). Here, duration refers to the period from the commencement of the ailment - whether it started before or during the reference period - to its termination, or to the end of the reference period if the ailment continued. Then, a bivariate analysis was performed for Kerala to examine the relationship between the various independent variables and risk of reporting as ill in the reference period, across these broad age groups. Since the dependent variable - whether ill during the last 15 days (1 - Yes, 2 - No) - is dichotomous, logistic regression analysis was performed for predicting the likelihood of reporting as ill in the reference period.

Due to limitations in collecting reliable income data through household surveys, the NSSO collects data on consumption expenditure in its surveys. The monthly per-capita consumer expenditure (MPCE) information thus available for each sample household is used as a proxy for the household's income level. Monthly per-capita consumer expenditure for a household is the total consumer expenditure over all items (in 30 days) divided by the number of members in the household. A person's MPCE is understood as the economic status of the household to which he or she belongs. Here, in the multivariate analysis, the variable "household size" has been included as a covariate to factor for the impact of household size in measuring the household's monthly per-capita consumer expenditure.


[Table 1]


Results

Comparing Reported Morbidity in Kerala and India

The proportion reported as ailing in the 15-day reference period in the survey was about 25% in Kerala, 9% at the all-India level and a low of 5% in Bihar (Table 1). More disaggregated data presented in Figure 1 showed that the risk of ailing in each of the 10-year age groups was several times higher in Kerala than in India or Bihar. The expected J-shaped relation between age and self-reported morbidity was visible only in Kerala. This could be due to differences in composition of ailments between these three populations.


[Figure 1]


Further details of the above observation can be inferred from the part of Table 1 on composition of ailments and median duration of the ailment. Data show that Kerala was at a relatively higher risk of chronic ailments, accounting for 43% of total reported ailments, than India at 35% and Bihar at 21%.

Compared with those in Bihar and all India, chronic ailments were much higher in Kerala in age groups starting from prime working ages. It should be noted that chronic diseases can remain undiagnosed, but the diagnosed ailment is less likely to be under-reported. Although the chance of early diagnosis is higher in Kerala, the gap between share of chronic and acute ailments shows that the state's population was still at higher risk of chronic ailments. Ailment composition shows that share of acute ailments was higher in Bihar than in Kerala for all age groups. But when we consider the magnitude of variation in proportion ailing between Bihar and Kerala, it is evident that the reported vulnerability toward acute ailments was also on the higher side in Kerala.

Median duration of an ailment was also reported to be higher in Kerala. Median duration of acute ailments was 7 days in Kerala, while only 5 days and 6 days in Bihar and India, respectively. Similarly, median duration of chronic ailments varied from 730 days in Kerala to 365 days in India and 90 days in Bihar. In addition, the table shows that a large proportion of the elderly in Kerala had been suffering from at least one chronic disease for more than 3 years. Hence, duration lived with a disease was very high in the state compared with elsewhere in India.

With the exception of adolescents, median duration of acute illness was highest in Kerala for all age groups. This finding corroborates the earlier explanation that high morbidity is not due to healthcare consciousness alone. Better healthcare consciousness would be expected to pave the way for wider reporting of minor ailments of shorter duration. In this case, average duration would be low, which is not observed here. Hence there is reason to believe that along with real morbidity, the duration of illness/treatment is also high in Kerala. Further research is necessary to verify reasons for the existence of longer duration of illness/treatment in the highly medicalized Kerala society, where chance of early diagnosis and detection of disease is higher than elsewhere in India.


[Table 2]


Reported Morbidity Differentials in Kerala

Having compared the reported morbidity in Kerala with the overall national scenario, the next step is to gain an in-depth understanding of the intra-state variation. Proportion ailing in the 15-day reference period varied from 16% in adolescents to 58% in the elderly. The risk of reporting as ill is was only marginally higher in females (26%) than in males (24%; Table 2). Age-wise analysis showed clearly that a larger proportion of male than female children was falling sick. Greater vulnerability to illness in women began with the early-working-age group, peaked in the late working ages and diminished in older ages. It was evident that health in females deteriorated earlier than in males. A rural-urban differential in reported morbidity was marginal. The only notable rural-urban differential across age groups was the relatively low level of illness among people of prime working age in urban areas compared with those in rural areas.

Caste-wise differences in the proportion reported as ill were visible in all age groups. Among children, adolescents and prime working ages, the reported level of ailing was higher in the scheduled castes/schedules tribes (SC/STs) than in other caste groups. However, this SC/ST population was shown to be in an advantageous position in terms of risk of illness among late working ages and the elderly. Vulnerability to illness was highest among "other backward castes" (OBCs) and "others" with respect to those in late working ages and among the elderly population, respectively. Here, "others" refers to the "top" layer in the social group classification in India, characterized by historical advantage in terms of social and economic status in comparison with  the OBC and SC/ST populations  At the aggregate level, the proportion reported as ailing was highest among Christians, followed by Hindus and then by Muslims. The pattern was the same in the three initial age groups. In the late working ages, however, higher morbidity was reported among Muslims, while in the 60+ population, risk of ailing was reported to be much higher in Christians and Muslims than in Hindus.

There were large-scale differences in morbidity levels between self-reported and proxy-reported cases in the survey. Among those who reported on their own status, about 28% described themselves as ill, while only 20% of individuals reported on by other household members, due to absence during interviewer visits, were described as ill. As specified by the survey guidelines, information on the health of children was reported by the mother or any other adult household member. The proxy-reported morbidity level was lower than the self-reported morbidity level for all three adult age groups studied, emphasizing the need to account for this reporting-related variation in any morbidity analysis using this data set.

Differentials by economic status of the study population showed an increase in percent ill from 20% in the lowest MPCE quintile group to over 30% in the highest two MPCE quintile groups. Similar rich-poor differences in reported morbidity were visible in all age groups, with the exception of late working ages. At the aggregate level, seasonal differences were negligible, but for children, the risk of illness was higher between January and March than from April to June. The proportion ailing was only 18% in Northern Kerala, while it was 30% in Southern Kerala. This regional picture remained the same for all age groups examined.


[Table 3]


Multivariate Analysis

Results of binomial logistic regression analysis, with risk of reporting as ill in the last 15 days as the dependent variable and selected independent variables, are presented in Table 3. Analysis is performed for all age groups under study as well as for all ages combined. The odds ratio presented in Table 3 indicates the variation in likelihood of reporting as ill within each independent variable, when the effect of all other independent variables in the model are kept constant.

All Ages

The regression analysis confirms the J-shaped relationship between age and reported morbidity level in the population. Odds ratio indicates that the likelihood of illness was 3.6 times more among the elderly than among children. Gender differences were insignificant. The likelihood of reporting as ill was 17% higher in rural populations than in their urban counterparts. Caste-wise differences became pronounced while controlling for other factors. Compared with the socially advantaged "other" castes,  the SC/STs' and OBCs' risk of reporting as ill was 47% and 23% higher, respectively. Significant religious differences were noted, with reported risk of being ill highest among Christians, followed by Muslims and Hindus.

The other important observation in this analysis is that the likelihood of reporting as ill is 65% higher in self-reported than in proxy-reported cases. Little difference was observed in the risk of reporting as ill between the lowest three MPCE quintile groups, but the 60-80 MPCE quintile groups reported experiencing a higher risk of illness compared with the lowest MPCE quintile groups. Again, the differential risk of reporting as ill between the richest and poorest quintiles was insignificant. Seasonal variations were significant, with reported morbidity marginally higher during January to March than between April and June. Multivariate analysis also confirmed the regional divide in proportion reported to be ill, with this risk in Southern Kerala nearly double that in Northern Kerala.

Age-Wise Differentials in Morbidity

Among children in the 0-9 age group, religious, seasonal and regional differences were significant. Christian children were reported at significantly higher risk of illness than their counterparts from other religious groups. It is to be emphasized that seasonal variations were significant only in the case of children, where the likelihood of illness was 32% higher between January and March than from April to June. Odds of reporting as ill was nearly 2.4 times more in Southern than in Northern Kerala. As already known, the majority of reported illnesses in the 0-9 age group were short-duration acute ailments, which is the reason for finding the environment-sensitive variable, seasonality, significant in the model.

Multivariate analysis for the adolescent population shows that the variables place of residence, caste, religion and region have a significant effect on risk of illness. The likelihood of reported illness is 35% higher among adolescents in rural areas than among their counterparts in urban areas. As in the case of children, SC/STs and OBCs are reported to have higher morbidity than other castes. Similarly, Christian children were 80% more likely to be reported as ill than those from other religious groups. As expected, the differential in risk of illness between Northern and Southern Kerala is highly significant.

For the population in the prime-working-age group, the risk of illness was reported as 26% higher in rural than in urban areas. Socially and financially upward "other castes" were at lower risk of illness when compared with the OBC and SC/ST categories. In fact this caste-wise deprivation in health status is a serious issue capable of offsetting livelihood opportunities. The other significant predictor of reported illness in this age group is the self- and proxy-reported dimension, where odds of reporting as ill were 70% greater in self-reported cases than in proxy-reported ones. Regional patterns - whether the household was in Northern or Southern Kerala - were  highly significant.

In the late working ages (40-59 years), the period that generally marks the onset of chronic conditions, the risk of illness was 46% higher among OBCs than SC/STs or other castes. Muslims were reported to be at higher risk of illness when compared with both Hindus and Christians. MPCE category-wise, risk of illness also become prominent from this age group onwards. As compared with the 0-20 MPCE quintile, the likelihood of reporting as ill was 94% and 69% more in the 60-80 and 80-100 MPCE quintile groups, respectively. Risk of reporting as ill was 93% higher in the population residing in Southern Kerala than those in Northern Kerala.

Analysis of the elderly population, aged 60 years and above, showed no gender, caste or rural-urban differences. However, the observation that Muslims had higher morbidity was significant in this age group. MPCE-wise differences were notable, with the highest MPCE quintile reported to be at higher risk than lower quintiles. This is due to the higher risk of communicable diseases among the top quintile groups compared with their lower-quintile counterparts. As noted in other age groups, morbidity among the elderly was reported as higher in Southern than in Northern Kerala.

Discussion

Analysis of reported morbidity, its composition and duration indicates that the potential risk of short-duration acute ailments and long-duration chronic ailments is much higher in Kerala than in other states. Number of days lived with a chronic disease is higher at all ages examined, which makes us rethink whether the Kerala population really enjoys a "good health" status compared with their counterparts elsewhere in India. Other secondary data sets show the prevalence of diabetes, hypertension, cancer and coronary heart disease to be greater in Kerala than in other states (RGI 2007). Similarly, the National Family Health Survey-II shows the prevalence of minor ailments like fever and cough to be greatest in Kerala. (International Institute for Population Sciences and ORC Macro 2000). The population of this state has gained considerably through mortality transition in terms longer life but appears to be more fragile than their counterparts in other parts of India. The other critical observation is the longer duration and treatment of acute ailments reported in Kerala, the reasons for which are unknown. The role of medical and nonmedical factors in prolonging the duration of ailment/treatment needs to be investigated separately. Kerala is well ahead of other states in physical access to private and public healthcare facilities (NCMH 2005). This, along with universal literacy, should have facilitated early diagnosis and detection of disease, leading to a shorter duration for acute ailments. But the reverse scenario noted here remains unexplained.

Some of the anomalies noted during aggregate-level analysis of reported morbidity are resolved through age-wise disaggregated analysis. If one factors for household-size-related bias in monthly per-capita consumer-expenditure quintile classification, the rich-poor differential narrows. Reported morbidity among children, adolescents and prime working ages is not associated with their economic status. At the same time, economic background becomes a significant predictor for population in late working ages and the elderly, due mainly to higher prevalence of chronic conditions noted among these subgroups. Significant caste-wise differences in adolescents and working ages indicates that health problems induced by income deprivation is a severe issue in socially disadvantaged subgroups. In the late working ages, the risk of ailing is higher among OBCs. The present study shows that the reported health status of Hindus is better than that of Christians and Muslims. It is encouraging to note that inequalities along religious lines do not exist in the prime-working-age groups. But relative risks are highest among Christians in age groups that are more vulnerable to acute ailments, and among Muslims in age groups where vulnerability to chronic ailments is high.

Another notable observation is that only children are vulnerable to ailments associated with seasonal variations. The health system should be equipped to meet this varying demand for preventive and curative healthcare services across seasons. Regional differences have not declined from the levels noted in the 1990s and are found across all age groups. The historical lag in exposure to both public and private health facilities between Northern and Southern districts in Kerala (Kutty 2000; Sadanandan 2001) could partly explain this divide. However, factors such as population density and a larger share of habitats along rivers or coastal areas expose the population in Southern Kerala to ailments in greater number than their counterparts in Northern Kerala. Furthermore, from a methodological point of view, analysis confirms that use of proxy respondents in these kinds of health-interview surveys has contributed to an underestimation of morbidity. Since insisting on self-reporting criteria is not feasible in this type of large-scale survey, self/proxy reported status should be used as a control variable in any analysis based on this data set, considered the only large-scale survey data on reported morbidity in India.

About the Author

T.R. Dilip, Lecturer, Centre for Development Studies, Ulloor Trivandrum, India

T.R. Dilip, Lecturer, Centre for Development Studies, Prasanth Nagar Road, Ulloor Trivandrum-695011, India, E-mail: dilip@cds.ac.in/ diliptr@hotmail.com, Fax: 91-471-2447137

Acknowledgment

Thanks are due to the National Sample Survey Organisation, Ministry of Statistics and Programme Implementation, Government of India, for providing this data set for research at the Centre for Development Studies. I sincerely acknowledge that comments and suggestions from three anonymous referees have contributed to the improvement of this article.

References

Deaton, A. and C. Paxson. 1998. "Economies of Scale, Household Size and the Demand for Food." Journal of Political Economy 106(5): 897-930.

Dilip, T.R. 2002. "Understanding Levels of Morbidity and Hospitalisation In Kerala, India." Bulletin of World Health Organisation 80(9): 746-51.

Gumber, A. and P. Berman. 1997. "Measurement and Pattern of Morbidity and Utilisation of Health Services: Some Emerging Issues from Recent Health Surveys in India." Journal of Health and Population in Developing Countries 1(1): 16-43.

International Institute for Population Sciences (IIPS) and ORC Macro 2000. National Family Health Survey (NFHS- 2)1998-99, Mumbai: IIPS

Kannan, K.P., K.R. Thankappan, V.R. Kutty and K.P. Aravindan. 1991. Health and Development in Rural Kerala: a Study of Linkages between Socio-Economic Status and Health Status, Trivandrum: Kerala Shastra Sahithya Parishad.

Kerala Shasthra Sahithya Parishad. 2006. Kerala Padanam. Kozhikode: Kerala Shasthra Sahithya Parishad (KSSP).

Kumar, B.G. 1993. "Low Mortality and High Morbidity in Kerala Reconsidered." Population and Development Review 19(1): 103-21.

Kunhikannan, T.P. and K.P. Aravindan. 2000. Changes in Health Status of Kerala 1986-1997. KRPLLD Discussion Paper No. 20. Thrivananthapuram: Centre For Development Studies.

Kutty, V.R. 2000. "Historical Analysis of Development of Health Care Services in Kerala." Health Policy and Planning 15(1): 103-9.

Michael E.J. and B. Singh. 2003. "Mixed Signals from Kerala's Improving Health." The Journal of Royal Society for Promotion of Health 123(1): 33-8.

Mohindra, K.S., S. Haddad and D. Narayana. 2006. "Women's Health in a Rural Community in Kerala, India: Do Caste and Socio-Economic Position Matter?" Journal of Epidemiology and Community Health 60: 1020-6.

Murray, C.J.L. and L.C. Chen. 1992. "Understanding Morbidity Change." Population and Development Review 18(3): 481-503.

National Commission on Macro Economics and Health (NCMH). 2005. Financing and Delivery of Health Care Services in India. New Delhi: Ministry of Health and Family Welfare, Government of India.

National Sample Survey Organisation. 1998. Morbidity and Treatment of Ailments. Report No. 441. New Delhi: Ministry of Statistics and Programme Implementation, Government of India.

National Sample Survey Organisation. 2006. Morbidity Health Care and the Condition of Aged. Report No. 507. New Delhi: Ministry of Statistics and Programme Implementation, Government of India.

Nananeetham. K, and M. Kabir. 2006. Health Status of Kerala: A Life Course Perspective. Project Report Submitted to IDPAD, Thiruvanathapuram: Centre for Development Studies.

Panikar, P.G.K. and C.R. Soman. 1984. Health Status of Kerala, The Paradox of Economic Backwardness and Health Development. Trivandrum: Centre for Development Studies.

Panikar, P.G.K. 1999. Health Transition in Kerala. KRPLD Discussion Paper No. 10 Thiruvanathapuram: Centre for Development Studies.

Rajan I.S., U.S. Mishra and P.S. Sarma. 2003. "Demography of Indian Ageing, 2001-2051." Journal of Aging and Social Policy 15(2&3): 11-30.

Registrar General of India (RGI). 2007. Baseline Survey Report 2004. New Delhi: Office of Registrar General of India, Government of India.

Sadanandan R. 2001. "Government Health Services in Kerala: Who Benefits." Economic and Political Weekly 36(33): 3071-7.