Page 14 - IJPS-8-2

International Journal of
Population Studies Accuracy of mother’s reporting on child immunization in Yemen

and 2012. Similar results were found for measles, where vaccination that was consistent within this round. It is
campaigns showed very positive results in controlling important to emphasize that for a particular child, their
the spread of the disease in the period between 2006 and information on vaccines comes from different sources.
2009. For instance, it is possible for a child to have a consistent
As can be noted, efforts to increase national vaccination date at which BCG vaccine was taken but having no record
coverage against the diseases covered by the EPI are part on the card about third dose of polio and their mother
of a long-standing strategy to accelerate the reduction of reported them as not vaccinated against polio. This means
child mortality in Yemen. According to UNICEF Yemen, that for the same child, it is possible to have information
several national and subnational vaccination campaigns from either the vaccination card (through a readable or
took place in 2012 and 2013. The most recent vaccination not readable date of vaccination) or the mother’s report
campaigns are especially important when analyzing depending on the vaccine to be considered. In other
the youngest children – aged <2 years old who should words, card availability information varies obviously from
have taken all vaccines by the time they were 1 year old child to child, but more than that, it varies from vaccine
(UNICEF and IPC-IG, 2014). to vaccine even when considering the same child. Because
of this type of vaccination history, the number of sampled
2. Data and Methods children varies depending on the vaccine that is being used
to analyze the accuracy of the mother’s report.
2.1. The dataset
The NSPMS is the only nationally representative 2.2. Methods
longitudinal household survey available for Yemen where The methodology for verifying the accuracy of mother’s
members of 6397 households were interviewed on a reports on child vaccination is adapted from Langsten
quarterly basis during a 12-month period between October and Hill (1998), and it is merely descriptive. First, we
2012 and September 2013. During the four rounds of the excluded from the analysis children with inconsistencies
NSPMS, caretakers were asked to show the vaccination in the dates of vaccination copied from their vaccination
card for every child under age five. The interviewers copied card in round 3. That is, we excluded children whose (1)
the vaccination information from the card onto the NSPMS dates of vaccination were before their own birth and (2)
questionnaire. If there was no vaccination card, the mother vaccination dates in one round were after the date of the
was asked to recall whether the child had received each of interview of that round. In addition, some children also
the vaccines, and in the case of multiple doses, interviewers lack information concerning their vaccination status,
also asked how many times the vaccine was given. Given which means either “mother doesn’t know,” “mother doesn’t
the longitudinal nature of the NSPMS, households were remember” or “children not vaccinated.” For the majority
visited 4 times over a 12-month period. These sequential of children with missing information, the interviewer had
visits allowed the survey to improve the information on seen their vaccination card, and we assumed that these
children’s vaccination histories, as they increased the children were not vaccinated according to the card.
likelihood of having access to vaccination cards and to find Then, we analyzed the possible sources of disagreement
better informed interviewees on the children’s vaccination between the mother’s report in round 1 (R1) and the
histories. card information in round 3 (R3). Since data collection

This paper focuses on Bacillus Calmette-Guérin (BCG) improved over time, it is important to mention that
and the third dose of polio vaccines (polio 3 ). Typically, our definition of consistency criteria on mother’s recall
rd
indicators of children’s vaccination coverage consider considers the round 3 information as our benchmark.
children aged 12 – 23 months as their target population. After checking the consistency of the mother’s report, the
This is important because one would expect the memory following metrics were calculated: accuracy, sensitivity,
error of mothers to be greatly reduced when considering specificity, and predictive values. These measures are
the immunization schedule of very young children. commonly used in research to validate and compare
instruments (Selimuzzaman, Ullah, and Haque, 2008;
There were 1369 children aged 12 – 23 months in round
3 of the NSPMS data. It is important to emphasize that our Miles, Ryman, Dietz, et al., 2013).
sample comprises children aged 12 – 23 months in round 3 The accuracy of mothers’ reporting corresponds to
whose card was seen in this round and whose information the percentage of children whose mothers reporting
on vaccines in round 1 was taken through the mother’s information matched the information registered in
report. In addition, concerning the dates of vaccination their vaccination card: (TP + TN)/(TP+FP+FN+TN).
obtained in round 3, we only kept children with dates of Sensitivity is the percentage of vaccinated children who

Volume 8 Issue 2 (2022) 8 https://doi.org/10.36922/ijps.v8i2.1274