Abstract

Background: In 2014, the team responsible for updating the World Health Organization (WHO) Vaccination Coverage Cluster Surveys: Reference Manual conducted pilot survey work in Bangladesh to shed light on different methods being considered for the manual.

Methods: Vaccination status of 420 children ages 0-11 months (m) and 451 ages 12-23m living in five rural and five semi-urban clusters in Bangladesh was assessed with a combination of three rounds of caregiver interviews, review of the children’s home-based vaccination records (HBRs), and review of health facility-based vaccination records (FBRs) when available. Caregivers also completed a brief literacy assessment. Study staff used photos of HBRs and FBRs to establish confirmed values of children’s date of birth, dates of vaccination, and gender.  The data recorded during household interviews was compared with confirmed values to calculate data entry error rates. Concordance was assessed between vaccination evidence from caregiver recall and from documented sources.  Self-reported caregiver education was compared with literacy results.  And accuracy of vaccination recall was assessed as a function of literacy.

Results: Interviewers made a data entry error in 3.3% of vaccination dates and in 1.5% of opportunities to record the child’s gender. Comparisons of caregiver recall with HBR and FBR demonstrated a relatively high level of caregiver accuracy for most doses although caregivers tended to under-report coverage for later doses in a series. Caregivers with better literacy were more likely to have their child’s vaccination card and were better at remembering their child’s vaccination status.

Conclusions:  This pilot study offers several suggestions for study teams doing vaccination coverage survey work. For example, if a survey’s goals include estimation of timeliness and characterization of missed opportunities, the team will want to find ways to double-check dates without revisiting homes or health facilities. When time and/or resources are scarce, survey teams may consider prioritizing matching FBRs for children of less literate caregivers and for those from rural populations.

Keywords:  Vaccination coverage, survey design, pilot survey, Bangladesh, data entry errors, data concordance