Determinants of vaccine coverage and timeliness in a northern Pakistani village

Abstract

The incidence of vaccine preventable disease in Pakistan remains high despite a long-standing Expanded Program on Immunization (EPI). We describe vaccine completeness, timeliness and determinants of coverage from a remote rural cohort (2012-2014). Vaccination histories were taken from EPI records. Vaccination was complete if all doses were received according to the EPI schedule and timely if doses were not ≥3 days early or ≥ 28 days late. Three models are presented: a multivariable logistic regression of household demographic and socioeconomic factors associated with complete vaccination, a multivariable mixed effects logistic regression assessing whether or not the vaccine was administered late (versus on-time), and a mixed effects multivariable Poisson regression model analysing the interval (in days) between vaccine doses. Of 959 enrolled children with full vaccination histories, 88.2 and 65.1% were fully vaccinated following either the pentavalent or DPT/HBV schedules if measles was excluded; coverage dropped to 50.0 and 27.1% when both doses of measles were included. Sixty-four (6.7%) were unvaccinated. Coverage and timeliness declined with subsequent doses. Migrating into the village after 1995 (95%CI 1.88 to 5.17) was associated with late vaccination. Being male, having an older father, and having parents with at least some formal education reduced the likelihood of a late dose. The interval between doses was consistent at 5 weeks (compared with the 4 weeks recommended by EPI). None of the socio-demographic variables were related to the likelihood of receiving full coverage. Vaccine coverage in Oshikhandass was higher than national averages. Measles vaccine coverage and timeliness were low; special consideration should be paid to this vaccine. The local vaccination schedule differed from the EPI, but the consistency suggests good local administration.

Fig 1

(left) Model predicting the odds ratio of completing the EPI schedule, (center) odds ratio of receiving a vaccine dose late and (right) the risk rate of the interval (in days) between doses. Continuous variables were centred on their mean value and scaled to unit standard deviation; coefficients therefore reflect the difference per one standard deviation from the mean. Solid points indicate that the 95% confidence (or credibility) interval for a coefficient did not include one.

Fig 2. The density of the interval between doses for multi-dose vaccines.

The line type indicates which interval (e.g., solid lines indicate the interval between the first and second dose) and the color indicates the different vaccines. The vertical line indicates a five-week interval.