Association Between Child Immunization and Availability of Health Infrastructure in Slums in IndiaImmunization and Health Centers in India Slums FREE

Despite immunization being one of the most cost-effective public health interventions, in India, the latest national survey shows that only 44% of children younger than 5 years are completely immunized.¹ Within India, there are disparities in urban-rural immunization coverage. Although coverage in urban areas is higher (58%) than rural (39%), coverage among the urban poor (largely living in urban slums) is 40%.^{2 - 3} Indeed, intraurban disparities in health are not unique to India and have been shown to exist in other cities in developing as well as developed countries.^{4 - 5} Existing meager urban public health services are rendered negligible by rapid urbanization and a faster growth of urban poverty.⁶ In the last decade, the slum population in India grew by more than 5%,⁷ with the currently estimated 80 to 100 million⁸ urban poor expected to double in the next 10 years.⁹ Improved immunization services are important for slum populations not only because the residents are underprotected against vaccine-preventable diseases but also because they are overexposed to the risk because of high density and poor environmental conditions.¹⁰

While prior research has extensively documented the factors that influence the demand for immunization in India,^{11 - 14} an examination of the relationship between the availability of health infrastructure and childhood immunization outcomes in low-income countries remains limited.¹⁵ In this study, we examined whether availability and proximity to health centers that provide immunization services is associated with the probability of being immunized among children living in slums in India. We also examined whether the presence of slum-based outreach support workers impacts immunization coverage, based on effective demand and supply linkage approaches that have demonstrated improved reach of health services.^{16 - 17} Specifically, we investigated whether presence of an anganwadi worker (AWW) improves immunization coverage. The AWW is the community-based functionary of the Integrated Child Development Scheme (ICDS), a flagship program of the Government of India since 1975 to reach out to children younger than 6 years, with child education, health, and nutrition services. The AWW assists the health functionary in improving demand and timely uptake of services through identifying and mobilizing the target group from her catchment area of 1000 to 1500 population.

Data for this study came from the baseline survey collected by the US Agency for International Development Environmental Health Project Urban Health Program in slums of Agra, India, in 2005. Agra is representative of a fast-growing city with a 1.3 million total population in 2000. More than 50% (0.85 million) of the city's population resides in 393 slums.¹⁸ In this study, we conducted a secondary data analysis. The study was reviewed by the Harvard School of Public Health institutional review board and was considered exempt from full review because the study was based on an anonymous public-use data set with no identifiable information on the survey participants.

The Urban Health Program collected data using structured interviews with a sample of ever-married women aged 15 to 44 years and having a child younger than 2 years to understand maternal and child health issues, demographics, and household characteristics. A separate slum questionnaire was used to collect information on availability of health amenities in each slum. The slum questionnaire was completed with a key informant, who was an educated person and familiar with health-related issues. For our analysis, we matched the mother file with the slum file to get information on the immunization status of each child, demographics, household characteristics, availability of health infrastructure, and other slum-level characteristics.

The Urban Health Program conducted a 2-stage systematic random-sampling procedure to select respondents. At the first stage, a sample of slum clusters was selected followed by the selection of households. From a total of 3359 households identified for the survey, 3075 mothers (91.5%) completed interviews. Refusal rate in the survey was low (0.3%). Other reasons for loss of sample were finding the house locked, mother not at home, postponed interview, and destroyed dwelling.

To study complete immunization status, we restricted our data to those of 1728 children older than 10 months. According to the World Health Organization guidelines, Indian children are considered fully immunized if they receive 1 BCG vaccine injection to protect against tuberculosis; 3 doses each of diphtheria, pertussis, and tetanus (DPT) and polio vaccines; and 1 measles vaccine by age 12 months.¹⁹ BCG vaccine should be given at birth or at first clinical contact; DPT and polio require 3 vaccinations near 4, 8, and 12 weeks of age; and measles vaccine should be given at or soon after reaching 9 months of age.¹⁹ Standard practice is to calculate immunization rates for children aged 12 to 23 months; however, since the immunization schedule can be complete after 9 months of age, we did not exclude children 10 to 12 months of age. We restricted analysis to the last child for greater accuracy. Our regression models included at least 98% of our sample.

To collect information on immunization status, the interviewer copied the date for each received vaccination, if the vaccination card was available from the mother. If the mother could not show a vaccination card, she was asked if her child had received vaccinations. If any vaccination had been received, the mother was asked whether the child had received a vaccination against tuberculosis (BCG); against DPT; against polio; and against measles. For DPT and polio, information was obtained on the number of doses of the vaccine given to the child. In such cases, mothers were not asked the dates of vaccinations. Immunization status was measured as “complete” if the child had received 1 dose of BCG vaccine, 3 doses each of DPT and polio vaccines, and 1 dose of measles vaccine; “partial” if any 1 or more vaccines were missing; and “not” if no vaccine was received. Similar categories have been used in other studies.^{14 ,20} We analyzed immunization results for children aged 10 to 23 months.

Presence of an urban health center (UHC) within 2 km of the slum was our primary exposure of interest and was measured by creating a categorical variable for presence or absence of a UHC within a 2-km range from the slum. We categorized measures of distance since the values obtained from the survey with regard to distance were not normally distributed. Further, given that slums are densely populated areas, the notion of distances can be problematic. A criterion of 2 km was based on our knowledge of distance that slum dwellers commonly walk for their daily chores and at which no cost of traveling is involved to obtain services, thereby increasing chances of availing services. This categorization has been used by another study examining similar exposures.¹⁵ Distance was not reported by each household; rather it was reported by key informants, such as the AWW or a schoolteacher from the slum, and in that sense, it is truly measuring whether the slum has proximal availability of a health service center or not. We used presence of an operational anganwadi center (AWC), a child care center located within the slum area itself, as the focal point for the delivery of services by the AWW as a proxy to measure presence of the AWW. We constructed an indicator variable for presence or absence of an AWC in the slum. Our available sample had missing values for 1 slum and 9 households interviewed from that slum.

Multinomial logit regression techniques that accounted for multistage survey design were estimated to model the probability of a child being not, partially, or completely immunized, after controlling for relevant child, maternal, household, and slum-level variables. Child-level covariates included age, birth order, and sex of the child; maternal covariates included age, education level, and occupation; and household-level covariates included religion, social category, and standard of living index. Recognizing that all slums are not alike in terms of their available public and private health services, health infrastructure, and community efforts, we also included slum-level covariates of presence of government hospitals and urban dispensaries within 2 km from the slum and presence of organized efforts, such as the slum development committees, self-help groups, and nongovernment organization within a slum (Table 1).

Religion and caste of the child were based on those of the head of the household. Caste was grouped as general, scheduled caste and scheduled tribe, or other backward classes. Economic condition of the child's family was determined using the standard of living index. The standard of living index is an asset-based summary measure calculated by considering different household characteristics of the house type, toilet facility, source of lighting, main fuel for cooking, source of drinking water, separate room for cooking, ownership of house, ownership of agricultural land, ownership of irrigated land, ownership of livestock, and ownership of durable goods by the household. This index was calculated using methods similar to those followed by the International Institute for Population Sciences/ORC Macro index used in the National Family Health Surveys in India.¹

Stata for Windows version 10.1 (StataCorp, College Station, Texas) was used to estimate our models. Results are presented as relative risk ratios (RRRs) with 95% confidence intervals (CIs). The RRRs are ratios of absolute risk but for the specific comparison of the outcome in question against the chosen reference. We chose “not immunized” to be our reference category and present the likelihood of a child being completely or partially immunized.

The study cohort included 1728 children 10 to 23 months of age. Forty-one percent of children had received no immunization at all and only 14% were completely immunized. Complete immunization prevalence was higher in slums closer to the UHC and where there was a presence of an AWC (Table 1). Distribution of covariates across different levels of UHC and AWC was similar (Table 2).

In unadjusted models, proximity to a UHC and presence of an AWC showed an increased probability of being immunized (Table 3 and Table 4); children living within 2 km of a UHC were more likely to receive all immunizations (RRR, 1.85; 95% CI, 1.20-2.86) or some immunizations (RRR, 1.80; 95% CI, 1.34-2.42) relative to children living in slums not within a 2-km range of a UHC. Similarly, children living in slums with an AWC were more than twice as likely to be completely immunized (RRR, 2.08; 95% CI, 1.35-3.22) and were 42% more likely to be partially immunized (RRR, 1.42; 95% CI, 1.01-2.00).

Adjusting for covariates, the likelihood of complete immunization increased for those slums proximate (within 2 km) to a UHC (RRR, 2.03; 95% CI, 1.12-3.66). Similarly, the likelihood of being partially immunized also increased (RRR, 2.33; 95% CI, 1.55-3.50) for those within 2 km of a UHC. Of the various health facilities, the effect of a UHC was substantially larger than other health facilities. The likelihood for complete (RRR, 2.47; 95% CI, 1.61-3.80) and partial (RRR, 1.56; 95% CI, 1.10-2.10) immunization of children with presence of an AWC was substantially higher. The results for other child, maternal, and household covariates in the models showed that there were a number of other demographic and socioeconomic predictors of childhood immunization. The most important predictors in our models included sex of the child, maternal literacy, social category, and standard of living index (Table 3 and Table 4). We also found that children of working mothers were less likely to be completely immunized (RRR, 0.48; 95% CI, 0.21-1.08) and partially immunized (RRR, 0.46; 95% CI, 0.28-0.78).

The most salient finding is that children living in slums proximate to a UHC or those having presence of an AWW are more than twice as likely to be completely immunized. Our study highlights a need for government attention to health services for the urban poor. The existing urban health delivery system remains inadequate to respond to the needs of the urban poor population. In India, urban areas have less than 4% of government primary health care facilities.²¹ With a total urban population of 285 million in 2000, the available urban primary health infrastructure translates to 1 health facility for about 0.15 million urban population.²² Other government initiatives, such as the ICDS, have also had limited expansion in urban areas. An analysis of rural and urban ICDS coverage shows that the AWC to total population ratio is less (1:1260) in rural areas and almost 5 times (1:6114) the rural ratio in urban areas.²³ Distribution of urban public health and ICDS services is worse in smaller cities.²⁴ Besides a lack of availability, UHCs are often not located close to slums.²⁵ Availability is also likely to be worse in the city periphery.⁶ Similarly, several AWCs are not located in or near slums.²² Access to a wide network of private health care facilities in urban areas may be an advantage solely for the relatively better off. Lack of services along with a degraded slum environment, overcrowding,²⁶ and high mobility within slums²⁷ increases the vulnerability of slum dwellers to infectious diseases. Therefore, improved immunization in slums may not only decrease mortality associated with vaccine-preventable disease but also limit the number of outbreaks and cases and decrease severity of disease.²⁸

In our study, we also found that sex, maternal literacy, social category, maternal occupation, and standard of living were important “demand-side” predictors in the immunization status of children, which was found in other studies as well.^{2 ,13} While some studies²⁹ have shown a significant role of health workers in reducing sex bias, we found that, despite adjusting for the role of health services and presence of health workers, girls are less likely to be immunized than boys.

While we controlled for confounding by including several covariates, we did not have information on the population of each slum, which may influence placement of a health facility and immunization coverage. Calculating from the total population of 1.3 million in Agra and the total number of UHCs, each of the 15 UHCs covers approximately 0.1 million, that is, twice that of its desired/planned population coverage. This may influence the quality of services and the low current immunization coverage rates. We would expect to see a greater impact if the population under each UHC was 50 000. The data also lacked information on the incidence of vaccine-preventable diseases among children; thus, we are unable to discuss the impact of immunization coverage. Our measurement of the distance of all urban health facilities is based on the judgment of the slum key informant. This may lead to potential misclassification. However, it is likely to be low since all key informants were educated and familiar with the health infrastructure. Our estimates are representative of proximity of a health facility to the slums. Measurement of immunization status includes “recall” data that may result in error in measurement of outcome. However, this error is less likely because of substantial rigor in training of interviewers and in data collection, as described in the “Methods” section. Our results using recall are more representative of the actual situation since excluding the sample that did not have immunization cards would, in fact, bias our results by drawing conclusions for only a few mothers who were able to preserve cards in small slum households that often lack safe places for storage. Through this study, we are not able to comment on the cost-effectiveness of expansion of UHCs and ICDS functionaries and the quality of services provided by the facility or the health worker, which may also impact costs and coverage. Finally, since the study was restricted to 1 Indian city, its generalizability to other cities in India and other countries will need to be tested.

Despite these limitations, in this study, the presence of a UHC and AWW was positively associated with immunization status of children in slums. These results suggest that greater coverage of slums by UHCs and ICDS workers may lead to increasing immunization among children.

While this study suggests important policy implications for improving the health of slum dwellers in fast-growing Indian cities, as urbanization and urban poverty gain proportions with more than half of the world's population now living in urban areas and 1 of every 3 urban dwellers in developing countries living in slums,⁹ these findings may have significance for other developing countries. With the shift in locus of poverty from rural to urban areas, improving health services for the urban poor cannot be neglected to meet the global health challenges.

Correspondence: S. V. Subramanian, PhD, Department of Society, Human Development, and Health, Harvard School of Public Health, 677 Huntington Ave, Kresge Bldg, 7th Floor, Boston, MA 02115 (svsubram@hsph.harvard.edu).

Accepted for Publication: September 2, 2009.

Author Contributions:Study concept and design: Ghei and Subramanian. Acquisition of data: Ghei and Agarwal. Analysis and interpretation of data: Ghei, Subramanyam, and Subramanian. Drafting of the manuscript: Ghei and Subramanian. Critical revision of the manuscript for important intellectual content: Ghei, Agarwal, Subramanyam, and Subramanian. Statistical analysis: Ghei, Subramanyam, and Subramanian. Administrative, technical, and material support: Ghei, Agarwal, and Subramanian. Study supervision: Subramanian.

Financial Disclosure: None reported.

Funding/Support: Dr Subramanian is supported by National Institutes of Health Career Development Award NHLBI K25 HL081275.