Outpatient care utilization in urban Kerala, India
1Institut national de santé publique du Québec (INSPQ), Canada 2Groupe de recherche interdisciplinaire en santé (GRIS), Département de médecine sociale et préventive, Faculté de médecine, Université de Montréal, Canada 3Centre for Development Studies (CDS), Thiruvananthapuram, Kerala, India 4Groupe de recherche interdisciplinaire en santé (GRIS), Département de médecine sociale et préventive, Faculté de médecine, Université de Montréal, Canada
Correspondence: Jean-Frédéric Levesque, Unité de santé internationale Édifice Saint-Urbain, 3875 Saint-Urbain, 5e étage Montreal (Quebec), H3W 1V1, Canada. Tel: +1 (514) 528-2400 ext. 3216; Fax: +1 (514) 528-2512; E-mail: jflevesq{at}santepub-mtl.qc.ca
Context: Kerala is characterized by a high density of public and private health infrastructure. While less inequality in access has been reported in this Indian state, few studies have looked at problems found within cities. Escalation of costs of private services and reduced public investments could generate some inequalities in access for the poor.
Objective: To assess factors associated with utilization and source of outpatient care in urban Kerala, and to discuss policy implications with regards to access to care.
Methods: A multilevel analysis of individual and urban characteristics associated with utilization and source of outpatient care was conducted using data from a 1995–96 survey by the National Sample Survey Organisation on health care in urban Kerala.
Results: There is a high level of utilization (83.6%) of allopathic medical services. Controlling for illness severity and age, utilization thereof was lower for the very poor (OR 0.13 [0.03; 0.49]), inhabitants of medium towns (OR 0.20 [0.05; 0.70]), and inhabitants of cities with a lower proportion of permanent material (pucca) houses (0.21 [0.06; 0.72]). Among all users, 77% resorted to a private source of care. Utilization of a private provider was less likely for the very poor (OR 0.13 [0.03; 0.51]) and individuals from casual worker households (OR 0.54 [0.30; 0.97]), while it was more likely for inhabitants of cities from both low public bed density districts (OR 4.08 [1.05; 15.95]) and high private bed density districts (OR 5.83 [2.34; 14.53]). Problems of quality and accessibility of the public sector were invoked to justify utilization of private clinics. A marked heterogeneity in utilization of outpatient care was found between cities of various sizes and characteristics.
Conclusion: This study confirms high utilization of private outpatient care in Kerala and suggests problems of access for the poorest. Even in a context of high public availability and considering the health transition factor, relying on the development of the private sector to respond to increasing health care needs could create inequalities in access. Investing in the public urban primary care system and ensuring access to quality health care for the poorest is warranted.
Key Words: access to health care, poverty, developing countries, primary health care, urban health
1This poverty line represents an indexation for 1995–96 of the most recent per capita poverty line (1993–94) suggested by the India Planning Commission.
2This type of measure corrects for the overestimation of poverty introduced by per capita measures of poverty. In a context such as Kerala, where the distribution of income across households shows lesser variability and with a concentration of households spread around the poverty line, the use of adjusted monthly per capita poverty consumption expenditure can identify the poorest households. Sensitivity analyses have confirmed that the very poor concentrate well under the per capita poverty line. The equivalence scale used attributed a value of 1 for the first adult, 0.7 for each subsequent adult and 0.5 for every member below 18 years of age (OECD 1982).
3Multilevel modelling allows for the simultaneous estimation of individual and contextual effects and takes into account the extent to which individual responses are correlated through membership in clusters of higher levels, in our case urban units. It provides an appropriate partitioning of variance between individual and urban levels to generate unbiased estimates (Snijders and Bosker 1999).
4All analyses were performed using restricted iterative generalized least-square (RIGLS) with the second order and penalized quasi-likelihood (PQL) approximation method. Variables were kept in the models when considered the main outcomes of interest, when statistically significant (95% CI excluding the null value) or, if not significant, when they showed an impact on other significant variables.
5Illness episode refers to the complete duration of the illness, benign illness refers to an illness for which no confinement to bed was ever reported during the episode, while a severe illness involves confinement to bed at some point during the episode.
6The first models included a random intercept and level 2 (urban) variance components. This so-called ‘empty’ model did not include any explanatory variables and enabled us to determine the extent to which the nesting of individuals within urban contexts explained a significant part of the variance in the outcome of interest. The second models tested individual-level variables, while the final models tested the addition of urban-level variables. The second and third models were compared with the ‘empty’ model using the deviance chi-square test statistic.
7Pucca houses or permanent houses are those whose walls and roofs are both made of permanent materials. The walls are either built with burnt bricks, metal sheets, stone or cement concrete. Roofing materials include tiles, slate, shingle, corrugated iron, zinc or other metal sheets, asbestos cement sheets, bricks, lime and stone, stone and RBC/RCC or concrete. Kutcha or temporary houses are dwellings whose walls and roofs are made of materials which need frequent replacement. Walls may be made of grass, leaves, reeds, bamboo, mud, unburnt brick or wood. The roof may be made from grass, leaves, bamboo, thatch, unburnt bricks or wood.
8Those who preferred private care perceived that public facilities are in inconvenient locations (Nandraj et al. cited in Dilip and Duggal 2004), that private services are more adequate (23% of respondents), in closer proximity (15%) and that private doctors behave more appropriately (13%) (Kunhikanan and Aravindan 2000). The reasons for not using government institutions in this study were: no treatment available (10%); no medicines (14%); no doctor available (10%); solicitation of bribery (5%); or premises not clean (3%). A study of poor urban dwellers in Mumbai suggested long waiting hours, long distances and too brief contacts with the doctor as reasons for not using government services (Yesudian 1994).