Page 15 - IJPS-8-2

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

were reported by their mother as vaccinated: TP/(TP+FN). 3. Results
Specificity is defined as the percentage of nonvaccinated
children who were reported by their mothers as not 3.1. Vaccination coverage by source of information:
vaccinated: TN/(FP+TN). The positive predictive value mother’s report versus vaccination card
(PPV) is the percentage of children reported as vaccinated Figure 1 shows the percentage of vaccinated children
who were actually vaccinated: TP/(TP+FP). Finally, the – against BCG and polio (third dose), fully vaccinated
negative predictive value (NPV) is the percentage of (had received BCG, measles and third dose of polio and
children reported as not vaccinated who were actually not pentavalent), undervaccinated (had not received at least
vaccinated: TN/(FN+TN). In each of these equations, the one of them) and even not vaccinated at all – among those
acronyms mean the following: TP=true positives, TN=true whose information on all vaccines was available through
negatives, FP=false positives, and FN=false negatives. The their mothers’ reports and the percentage of vaccinated
definition of true or false (positives or negatives) refers to children among those whose information on all vaccines
the consistency between information on vaccination cards was available on their vaccination cards. It is important to
and mother’s recall, taking into account that the former is emphasize two features of this figure. First, it only includes
our gold standard. children with available information on all vaccines (either
Based on these measures of agreement, we evaluated from mother’s report or vaccination card). Although
the level and pattern of accuracy of mothers’ recall on their having all children with complete information on their
children’s vaccination histories by comparing mother’s vaccination status would be the ideal situation, only 62%
reporting information in round 1 with the vaccination of children aged 12 – 23 months who have vaccination
card shown in round 3. It is important to emphasize cards show information on all vaccines taken, and the
that validity refers to the degree to which the test or an percentage of children aged 12 – 23 months who have a
estimate based on a test is able to determine the true value vaccination card is as low as 42%. Second, it is important
of what is being measured (in this case, vaccine coverage). to highlight that the sample of children used in this table
In this sense, sensitivity and specificity (and therefore the does not exactly correspond to the sample of children used
accuracy itself) are properties inherent to the criterion of in our empirical exercise. The idea of Figure 1 is giving
the evaluated test and do not vary except by technical error. a more general idea of how the estimates in vaccination
Predictive values, however, depend on the prevalence of the coverage may vary depending on the source of vaccination
phenomenon in the study population. The PPV increases information (whether it is mother’s report or vaccination
with prevalence, while the NPV decreases. Thus, when card).
the phenomenon is rare, the PPV is low, since most of the The most striking aspect of these numbers is the
positive information results from mothers of unvaccinated significantly higher percentage of vaccinated children
children, representing false positive results. On the other among those with vaccination cards in comparison to
hand, NPV is high at low prevalence (Fletcher, Fletcher, those whose mothers reported their vaccination status.
and Fletcher, 2014). For instance, approximately 83% of 557 children aged 12

Figure 1. Percentage of vaccinated children among those aged 12 – 23 months in round 3 with information available on all vaccines taken according to the
source of information – Yemen, 2013. Source: IPC-IG and UNICEF (2014), round 3.

Volume 8 Issue 2 (2022) 9 https://doi.org/10.36922/ijps.v8i2.1274