At Arua Regional Referral Hospital in northern Uganda, the figures that health managers rely on to plan maternal services are only half trustworthy. A new study reviewing twelve months of delivery records found that submitted Health Management Information System (HMIS) reports matched the hospital’s on-site maternity registers in just six of the twelve months reviewed — a completeness rate of 50 percent. In the other six months, discrepancies between what was recorded in source documents and what was reported into the system pointed to incomplete or inconsistent reporting of deliveries at one of the country’s busiest referral facilities.
The study, published in Discover Health Systems, was a mixed-methods investigation carried out in 2023 at Arua Regional Referral Hospital, the only regional referral hospital in the West Nile sub-region. The facility serves twelve districts, including refugee communities, and receives referrals from parts of the Acholi sub-region as well as from South Sudan and the Democratic Republic of Congo, generating high volumes of maternal and child health data. Researchers from the hospital surveyed 92 staff members drawn from a total population of 120, selected through a combination of purposive sampling for administrators and records staff and simple random sampling for clinical personnel in the maternity ward, antenatal care clinic and family planning clinic.
Quantitative data were collected with a structured self-administered questionnaire uploaded to the KoboCollect mobile platform, a choice the authors say reduced data entry errors through automated skip patterns and validation rules. Qualitative insights came from key informant interviews with four individuals drawn from hospital administration, the records department, the district health office and the Ministry of Health. Secondary data were extracted from unit registers, monthly HMIS reports and the DHIS2 digital system using a document review checklist. Instruments were pre-tested at Kuluva Hospital in Arua District before the main data collection began.
The analysis combined descriptive statistics with Spearman’s rank correlation, computed in SPSS v26 and R v4.5.1, with qualitative transcripts coded thematically in NVivo. Because the capacity-building variables were ordinal and the correlation analysis rested on only twelve monthly observations, the researchers set significance at p < 0.05 and treated the correlational results as exploratory. Data completeness was defined narrowly: a month counted as complete only when delivery figures in the source registers exactly matched those reported in HMIS submissions, with any discrepancy classified as incomplete reporting. The authors note that this measure overlaps with data accuracy and served as a proxy indicator of completeness in routine reporting, consistent with similar HMIS verification studies.
The correlational findings pointed clearly toward supervision and verification. Supportive supervision showed a moderate, statistically significant positive correlation with data completeness (ρ = 0.598, p = 0.040), while routine data quality assessments showed an even stronger significant association (ρ = 0.714, p = 0.009). HMIS training, by contrast, showed a positive but non-significant correlation (ρ = 0.488, p = 0.108), and an aggregate capacity-building measure correlated positively with completeness but without statistical significance (ρ = 0.600, p = 0.453). The authors caution that with only twelve monthly data points, statistical power is limited and the estimates may not be stable.
The survey data revealed an uneven pattern in how often capacity-building activities actually took place. Supportive supervision visits and data quality assessments were primarily conducted quarterly, reported by 78.3 percent and 79.3 percent of participants respectively. HMIS training happened far less often: the largest share of respondents, 40.2 percent, said trainings occurred annually, and 31.5 percent said bi-annually. About 80.4 percent of participants said their facility received regular supportive supervision visits, but only 31.5 percent reported receiving regular feedback on the quality of the reports they submitted — a gap that surfaced repeatedly in the qualitative interviews.
Three themes dominated the qualitative material. First, participants described inconsistent HMIS training and persistent knowledge gaps, particularly around revised HMIS tools and indicators. One key informant explained that the HMIS 105 report forms are filled in triplicate but staff had not been told how this should be done or where the three copies should go, and that renamed data elements and indicators posed challenges for timely gathering, processing and reporting. Second, participants flagged limited feedback and supervision challenges: data quality assessments were conducted periodically, usually quarterly, and often by non-governmental partners who focused on a few data elements aligned with their project goals, leaving other data elements and units unverified. Third, workload and data management constraints emerged as a structural burden, with some health workers delaying documentation and leaving compilation to records staff at the end of the reporting period.
That last theme offers a concrete mechanism for the discrepancies the document review uncovered. When frontline clinicians postpone filling in registers, the monthly compilation falls to a small number of records officers working against the clock, increasing the likelihood of both over- and under-reporting. One key informant described how data collection for all facility data elements is coordinated by records officers who enter HMIS indicators into DHIS2, with a biostatistician stepping in when quality problems are discovered. The study’s participants were predominantly female (68.0 percent), half came from the maternity department, and 53.3 percent were enrolled midwives — the cadre whose daily documentation ultimately determines whether the reported numbers match reality.
The discussion situates these results within a broader evidence base. Previous research in Kaduna State, Nigeria, reported substantial improvements in data completeness after structured training of primary healthcare workers, and evidence from Malawi links training, supportive supervision and regular feedback to better data quality. The Arua findings refine that picture: training alone appeared insufficient, apparently because it was irregular and left staff struggling with revised tools, whereas supervision and data quality assessments — which embed verification and accountability into routine work — tracked more closely with complete reporting. The authors emphasise that the quality and scope of supervision, particularly feedback and follow-up, may matter as much as its frequency.
Important limitations frame how far these conclusions travel. The cross-sectional design permits only associations, not causal claims, and some measures relied on self-reported information vulnerable to recall or social desirability bias. Facility registers served as the reference standard for validating HMIS reports, yet those source documents may themselves contain inaccuracies. The study did not examine the quality, content or duration of training received, did not explore variation in staffing levels or workload across departments, and assessed only completeness — not timeliness or the accuracy of recording maternal complications. The correlation analysis, based on twelve monthly data points, should be read as exploratory rather than confirmatory. Within those bounds, the practical message is specific: strengthening consistent supervision, routine data verification and structured feedback mechanisms may do more for the reliability of maternal delivery data than occasional training events, and the authors suggest that district health teams, facility in-charges, biostatisticians and the Ministry of Health each have defined roles in making routine maternal health data trustworthy enough to act on.
Subject of Research: Capacity-building practices and HMIS data completeness for maternal delivery services at a Ugandan regional referral hospital
Article Title: Effects of capacity building interventions on Health Management Information System data completeness for maternal delivery services at Arua regional referral hospital Uganda
Article References: Ahimbisibwe, J., Nakayinga, R., & Ekakoro, N. (2026). Effects of capacity building interventions on Health Management Information System data completeness for maternal delivery services at Arua regional referral hospital Uganda. Discover Health Systems, 5(1), Article 71. https://doi.org/10.1007/s44250-026-00411-1
Image Credits: AI Generated
DOI: 10.1007/s44250-026-00411-1
Keywords: HMIS, data completeness, maternal health, supportive supervision, data quality assessment, Uganda, DHIS2, midwives, health information systems, capacity building, referral hospital, mixed methods
News Source: Ophelia Keating. (October 29, 2026). Research report: Effects of capacity building interventions on Health Management Information System data completeness for maternal delivery services at Arua regional referral hospital Uganda. Scienmag.
