01–1.07), but these effects disappeared after adjusting for travel time. The only significant predictor see more of immunization rates in the final model was season, with lower rates observed in the rains than in the dry season (HR = 0.86, 95% CI: 0.81–0.92). This large-scale survey of young children in Kilifi District, Kenya showed very high immunization coverage for all recommended vaccines, with 98.9%, 95.7%, 95.6% and 89.7% of subjects with vaccine cards receiving BCG, three doses of pentavalent, three doses of OPV, and measles vaccines by the age of 1 year, respectively. Only 14% of enrolled

subjects did not have vaccine cards available for examination. In this group, reported coverage was three to seven percentage points lower for all doses of vaccine (except OPV0), but remained >90% for BCG, DTP-HepB-Hib3, OPV3 and >80% for measles. The wide discrepancy between maternal reporting and card data for OPV0 coverage is specific to this vaccine, and may reflect poor recall for the period immediately after delivery. The reliability of mothers’ histories was previously evaluated in this setting among 18 children enrolled in a small immunization coverage survey, showing that 100% of mothers correctly recalled the first dose of DTP, 94% the second dose and 88% the third dose [9]. Evidence from other regions is conflicting, with some studies suggesting that maternal recall has low accuracy [22], [23], [24], [25], [26] and [27].

Most of these studies were conducted in industrialized countries and data from Kilifi, Egypt [23] AZD5363 price or Sudan [28] may be more relevant for our analysis. Regardless of the reliability of maternal recall, we calculated that even with 0% coverage in children without cards, overall coverage for BCG, Pentavalent-3 (or OPV3) and measles would attain 85%, 82% and 77%, respectively; these values would increase

to 92%, 89% and 84% with 50% coverage in children without cards. In addition DNA ligase to recall bias, our results may be subject to survivor bias because we only sampled children who were alive and 6–11 months of age at the time of the last Epi-DSS census. The 2006 birth cohort had an infant mortality ratio of 37 per 1000 live births (unpublished data, Kilifi Epi-DSS): even if none of these children were vaccinated, BCG, pentavalent-3, and measles coverage would only be reduced to 95%, 92% and 86%, respectively. Together, these results strengthen the evidence from earlier, smaller studies conducted from 2002 to 2004 [9], and attest to the success of the Kenyan EPI in reaching a large majority of children in Kilifi. They also concur with data from the 2008 Kenya Demographic and Health Survey (unpublished data, Kenya 2008 DHS) and WHO/UNICEF joint estimates [29] that showed approximately 95% coverage with BCG, 85% with Penta3, and 85–90% with measles vaccine on a national level. We sought to investigate spatial variations in immunization coverage, and found that these were relatively limited in the study area.