Quality of malaria data in public health facilities in three provinces of Mozambique

Background Malaria data reported through Mozambique's routine health information system are used to guide the implementation of prevention and control activities. Although previous studies have identified issues with the quality of aggregated data reported from public health facilities in the country, no studies have evaluated the quality of routine indicators recorded in health facility registries. This study addresses this issue by comparing indicators calculated from data from exit interviews and re-examinations of patients with data based on registry records from health facilities in order to measure the quality of registry data and data reporting in three provinces in Mozambique. Methods Data were collected from 1,840 outpatients from 117 health facilities in Maputo, Zambezia, and Cabo Delgado Provinces interviewed and examined as part of a malaria-specific health facility survey. Key indicators based on exit interview / re-examination data were compared to the same indicators based on records from health facility registries. Multivariable regression was performed to identify factors associated with indicators matching in re-examination / exit interview data and health facility registries. Aggregated indicators abstracted from facility registries were compared to those reported through the routine health management information system (HMIS) for the same time period. Results Sensitivity of exit interview / re-examination data compared with those recorded in facility registries was low for all indicators in all facilities. The lowest sensitivities were in Maputo, where the sensitivity for recording negative RDT results was 9.7%. The highest sensitivity was for recording positive RDT results in Cabo Delgado, at 75%. Multivariable analysis of factors associated with agreement between gold standard and registry data showed patients were less likely to be asked about having a fever in the triage ward in Maputo and Cabo Delgado (adjusted Odds Ratio 0.75 and 0.39 respectively), and in the outpatient ward in Cabo Delgado (aOR = 0.37), compared with the emergency department. Patients with positive RDT were also more likely to have RDT results recorded in all three provinces when patients had been managed according to national treatment guidelines during initial examination. Comparison of retrospective data abstracted from facility registries to HMIS data showed discrepancies in all three provinces. The proportion of outpatient cases with suspected and confirmed malaria were similar in registry and HMIS data across all provinces (a relatively low difference between registry and HMIS data of 3% in Maputo and Zambe ' zia), though the total number of all-cause outpatient cases was consistently higher in the HMIS. The largest difference was in Maputo, where a total of 87,992 all-cause outpatient cases were reported in HMIS, compared with a total of 42,431 abstracted from facility registries. Conclusion This study shows that care should be taken in interpreting trends based solely on routine data due to data quality issues, though the discrepancy in all-cause outpatient cases may be indicative that register availability and storage are important factors. As such, simple steps such as providing consistent access and storage of registers that include reporting of patient fever symptoms might improve the quality of routine data recorded at health facilities.