Completeness and accuracy of data in Zimbabwe's national PMTCT program health facility registers: findings from a patient level data quality audit
Background: Prevention of
mother-to-child transmission of HIV (PMTCT) remains a priority in Zimbabwe. Complete
and accurate data are required to reliably monitor PMTCT targets. In 2014, the Elizabeth
Glaser Pediatric AIDS Foundation (EGPAF) conducted a patient level data quality
audit (DQA) of the national PMTCT program to measure completeness and accuracy
of data in facility registers, which are used to measure PMTCT program
performance.
Methods: A descriptive cross-sectional
study was conducted in 43 randomly selected health facilities. Patient level
data for pregnant and lactating women and infants aged six weeks to six months
were abstracted from patient-held medical cards and facility registers. In
addition, exit interview findings with antenatal (ANC) and postnatal (PNC) women
were compared with data on patient held cards. A data element for example age was
complete if patient held card and facility register were documented, and
accurate if documentation on the two sources were the same. Reasons for
discordance were explored through interviews with healthcare workers (HCWs). Data
were analyzed using STATA 12.
Results: Records for 292 ANC and 266
PNC women were reviewed. Table 1 summarizes average completeness and accuracy
of data in facility registers by facility type.
Overall completeness was 83% for ANC and 71% for PNC; overall accuracy was
75% for ANC and 66% for PNC. Completeness and accuracy of ANC data for clinics
and rural hospitals were higher than referral facilities (mission, district and
provincial hospitals). Incompleteness and inaccuracy were largely a result of
health workers documenting on patient held cards only and not updating facility
registers; largely due to multiple registers and high workload. Figure 1
summarizes contribution of documentation practices to incompleteness. Completeness
based on data from patients'' interviews and patients held cards was 100% for
all data elements; accuracy was above 90%.
| Health Facility Type | Completeness of ANC data elements | Completeness of PNC data elements | Accuracy of ANC data elements | Accuracy of PNC data elements |
| Clinic | 91% | 69% | 78% | 63% |
| Rural Hospital | 94% | 65% | 89% | 60% |
| Mission Hospital | 85% | 76% | 76% | 72% |
| District Hospital | 68% | 74% | 54% | 66% |
| Provincial/Central Hospital | 75% | 70% | 76% | 67% |
| Average | 83% | 71% | 75% | 66% |
[Reasons for incompleteness]
Conclusions: The variation in completeness and accuracy by facility type calls for targeted on-site mentoring and coaching of facility level health workers. Health worker documentation in patient held cards without updating facility registers was the major determinant of incompleteness and inaccuracy. There is need to ensure that health workers document in facility registers in order to accurately measure PMTCT service uptake.
E. Tachiwenyika1, R. Musarandega1, M. Murandu1, M. Chideme1, C. Muchuchuti1, B. Mutede1, I. Shabbir2, R. Cathcart2, A. Mahomva1, L. Katirayi3
1Elizabeth Glaser Pediatric AIDS Foundation, Technical Department, Harare, Zimbabwe, 2Elizabeth Glaser Pediatric AIDS Foundation, Strategic Information and Evaluation, Washington DC, United States, 3Elizabeth Glaser Pediatric AIDS Foundation, Technical Department, Washington DC, United States