Validity of maternal recall for estimating vaccination coverage in children aged 12 to 23 months in the health district of Podor in 2020 (Senegal)

Abstract Methods The study was cross-sectional. Mothers of children aged 12 to 23 months represented the study population. Cluster sampling was carried out. Bacillus Calmette- Guerin vaccine (BCG), hepatitis B vaccine birth dose (HepB-BD), third dose of oral polio vaccine (OPV3), third dose of diphtheria, pertussis, tetanus, hepatitis B and haemophilus influenzae type b vaccine (Penta3) and first dose of measles-rubella vaccine (MR1) were used to assess the validity of maternal recall compared to HBR or HFR. Sensitivity, specificity, concordance, and Cohen's kappa and prevalence-adjusted bias-adjusted Kappa (PABAK) were estimated. Logistic regression revealed the determinants of recall bias. Results A total of 832 mothers were interviewed. Concordance ranged from 79–83.8%; the recall bias from 16.2–21%, the sensitivity from 80.6–86.7%, the specificity from 54–74.5%, the Kappa coefficient from 24.8 to 37.6 and the PABAK from 58 to 67.6. Concerning the vaccination status for BCG (OR = 2.38; p-value = 0.034), HepB-BD (OR = 2.2; p-value = 0.031) and Penta3 (OR = 2.93; p-value = 0.017), recall bias occurred more frequently in rural areas. The vaccination status of male children for HepB-BD (OR = 0.64; p-value = 0.034) was less subject to recall bias. In addition, the lack of prenatal and postnatal education on childhood vaccination was found to be a determinant of recall bias regarding vaccination status for BCG (OR = 2.94; p-value = 0.007) and Penta3 (OR = 1.96; p-value = 0.046), respectively. Conclusion Recall bias is relatively high for all vaccines. Maternal recall would be less valid than HBR or HFR. Further studies would be needed to confirm or refute these results.