A Dynamic, Multi-level Analysis of Recent Immunization Trends in Colombia

McQuestion, Michael, and Robert E. Jones
Working paper no. 1995-13


Primary health care programs, including childhood immunizations, are intensively utilized by the majority of Latin American households. During 1985-1991, governments used national PHC infrastructures to implement the Regional Polio Eradication Initiative. Unprecedented vaccine coverage levels were attained through a combination of mass campaigns, house-to-house vaccinations and improvements in routine immunization services. Little is known, however, about the effects of these interventions on immunization demand; whether they reached so-called high-risk households and, if so, whether program effects were sufficient to offset the negative risk factors. Using the household production of health model, this paper examines the probability and timing of full immunization over this period in one case country, Colombia. Information on the immunization status of 3609 vaccine-eligible children born 1985-90 was extracted from Colombia’s 1990 Demographic and Health Survey. Annual immunization coverage estimates from the Colombian Ministry of Health for 1985-90 for 148 sample municipios were appended to each child record, along with household-level data. Initial non-parametric regressions showed that five of six observed risk factors (maternal age <20, no maternal education, short previous birth interval, lack of a safe toilet, female headship and loss of a previous child) negatively influenced full immunization probability.

Multivariate logit models showed that high-risk households were significantly less likely to obtain immunization cards (a proxy for exposure to the routine immunization program), despite rising cardholdership rates over the period. Based on recall or card data, high-risk children without cards were significantly less likely to be fully immunized than were high-risk children with cards. A treatment effects model of full immunization probability showed that cardholdership effectively offset all but one risk factor (previous birth interval <24m). Among 1376 cardholders, waiting times to full immunization fell monotonically over the period. Local program coverage of 80% or higher and prior prenatal use both increased the hazard of full immunization. However, three of five maternal occupational categories decreased the hazard, as did no maternal education, consensual versus legal union and lack of a safe toilet. The results show that demand for routine immunizations rose over the period; that high-risk households were less likely to use the routine program but that exposure to the routine program effectively increased their demand for immunizations. To maximize health benefits, future interventions should aim to selectively recruit high-risk households into the ranks of routine immunization users.