The Need for Biomarkers in Surveys, the Case of Immunization (303637)

Keywords: vaccination coverage, recall bias, biomarkers, data linkage

Estimating vaccination coverage is challenging in resource-poor settings where accurate records are sparse. Little is known about how coverage estimates based on maternal recall differ from those based on health cards for the same children. Even less is known about how these measures compare with actual rates of seroconversion, which could be used to estimate of effective coverage of immunizations.

This study compares the accuracy of maternal recall, health card documentation, and antibody presence in vaccination coverage estimates in the poorest 20% of the population in the state of Chiapas, Mexico. A sample of 4,700 households with children under-5 and women of reproductive age were surveyed. A pre-survey census was carried out within segments which had been randomly selected with probability proportional to size. As part of a large household survey, anthropometric measurements were conducted for all children under five and dry blood spot samples for the detection of measles antibodies were collected from children aged 12--23 months.

Results suggest that maternal recall, child health cards and antibody tests generate differing estimates of immunization coverage. Recall differs from card-based estimates of vaccination coverage by up to 40 percentage points. There are considerable differences in measles immunization coverage as assessed by the presence of measles antibodies versus other survey sources, highlighting weaknesses in card accuracy, card coverage, and vaccine administration.

Current national estimates of immunization coverage based on the combination of maternal recall and children’s health cards may be over-estimating actual protection against vaccine-preventable diseases. Correcting for biases in recall and card coverage may produce more accurate estimates of intervention coverage.