Validity of parental recall to estimate vaccination coverage: Evidence from nine districts in Malawi

Abstract

Abstract Introduction During household surveys, vaccination coverage is commonly estimated through vaccination cards and parental recall. Although data from vaccination cards are more reliable than parental recall, both approaches are prone to selection and information bias. At times, vaccination cards may not be available because of loss or misplacement necessitating the use of parental recall as alternative. In this study, the validity of the vaccination coverage from these two sources were compared. Individual and household level factors associated with recall bias were also assessed. Methods The vaccination coverage of the parental recall and vaccination card were calculated separately for each of the vaccines. The level of agreement was computed between the estimates from the parental recall and vaccination cards. Sensitivity and specificity of parental recall were computed. The study also examined the factors that would be associated with recall bias. Multiple logistic regression model by vaccine type were fitted where odds ratios and 95% confidence intervals were reported. Results The vaccination coverage for BCG was 98.6 for card-based and 98.1 for parental recall. The vaccination coverage for OPV was 98.9 for card-based and 98.1 for parental recall. For PCV, it was 99.6 for card-based and 97.4 for parental recall. For measles was 84.1 for card-based and 88.0 for parental recall. The results show a high level of agreement between parental recall and card-based (>97%) across all vaccines. The parental recall bias was minimal ranging from 1.13 to 6.66. The sensitivity of parental recall was almost 100% with low specificity. Factors such as parental and child age was associated with parental recall bias for PCV and measles Conclusion The study has demonstrated and supported the need to use the parental recall to estimate the vaccination coverage for different vaccine types which can be used instead of or in the absence of card-based data or records.