Malaria control and elimination heavily rely on the vigilance of health systems’ human resources to adequately and ardently perform routine surveillance activities, both passive and active [1]. In sub-Saharan Africa, routine surveillance of all reported diseases has adopted a de facto management structure, hinged around district-level headship, through which the mainstay reporting system, introduced around 2015, derives the name District Health Management Information System – Version 2 (DHIS-2) [2, 3]. As such, duty performance of health management information systems (HMIS) surveillance activities by district or city health management officials (DHM), is critical. In high endemicity settings, malaria surveillance activities at the district involve the collation of data from district-supervised health facilities, monitoring quality of these data, and submitting the abstracted data to the national central authorities - mainly National Malaria control divisions/Programs (NMCP) under the leadership of national ministries of health.

Enormous investments have been made to strengthen HMIS, through the development of increasingly complex data collection tools and derivative indicators and the establishment of electronic reporting systems, for example [4, 5]. Despite recent improvements, some inconsistencies still exist in reporting, data errors, and timeliness, among others [6, 7]. Notably however, perspectives and practices of frontline HMIS data handlers in district-level health offices, who play a data gatekeeper role between health facilities and national health authorities, have hardly been assessed. Improved understanding of these aspects at this management level may foster adapted surveillance support systems and highlight key target points in the pathway to surveillance transformation into a distinct intervention. This study therefore, aimed to assess the attitude and practices of frontline officials in HMIS data recording, reporting, analysis and use, and associated support supervision, at district level.

Routine HMIS data plays a central role in surveillance-based evidence generation and health facility monitoring [1, 8]. Efforts towards its improved interpretation and thereby use include integration of HMIS with supportive data from other sources [9, 10]. Notably for Uganda, the prevailing malaria reduction and elimination strategic plan 2021–2025 set as its fourth of six strategic objectives, to have “malaria programming at all levels guided by and based on robust data and evidence” [11]. Despite all efforts of good intention to improve HMIS data, generate evidence through advanced analyses, or caution about the data itself, ignoring the role of HMIS data handlers may perpetuate limitations to its full utility. Moreover, majority of factors associated with improved HMIS data use, including prioritization of data or promotion of a data-use culture, skills training, support supervision, regular performance review, planning, governance, and resource availability, are all people-centered [1, 12–14]. Extensive progress has been made in and attention placed on advancing modeling approaches to better interpret HMIS data with commendable benefits [15–18]. However, the potential role of top-level managers in the state of routine HMIS data, considering their perceptions and day-to-day actions, remains little understood or appreciated. An improved understanding of this perspective could facilitate the identification of capacity needs and the establishment of accountability mechanisms, as well as, improved data ownership, interpretation, and use [19].

A formative cross-sectional assessment of attitudes and practices in management of routine data, among the DHM HMIS managers, presented a great opportunity to examine processes undertaken and unearth potential pitfalls and/or areas for improvement and intervention.

The study involved a cross-sectional survey, conducted between January and June, 2024, using a semi-structured interviewer-administered questionnaire.

DHM directly involved in routine HMIS reporting were the study population from among whom, one participant per jurisdiction was approached, recruited and interviewed. All district/city health team members involved in HMIS reporting were eligible for recruitment into the study, if agreeable to consent. DHM offices were visited by a research assistant and from each office, one member of the DHM team that is actively engaged in HMIS reporting was recruited into the study, except Kampala city. Kampala health management is further decentralized into five divisions. Here, one member of each division’s health team, actively engaged in HMIS reporting, was recruited into the study. Particular interest was placed on biostatisticians, HMIS focal persons, and district-level health officers, but in the absence of all these, a designate handling HMIS data for routine reporting purposes was considered.

A semi-structured, study validated questionnaire was developed and implemented on the Kobo Toolbox platform [21] to be administered by trained study team members to consenting participants. The questionnaire was designed in English with both closed and a few open-ended questions categorized into the four topical sections of: a) Data recording and review, b) Data reporting c) Support supervision, and d) Data analysis and use. In each of these sections were questions addressed to attitude and practices. Participant demographics, including age, cadre, gender, and level of education were also captured at the start of the interviews.

Field team members were trained on the study protocol, the informed consent process, and administering of the questionnaire. Questionnaires were conducted in electronic format by the trained field workers, using android tablets with data regularly uploaded to a remote Kobo server. All interviews were conducted in English, with no need for translation, since all participants were expected to be sufficiently fluent in English.

District copies of the most recent monthly and support supervision reports, as well as documented data queries generated from review of received data were also sought and reviewed by the study team. Particular focus was paid to identify a sample of health facilities (one per level, including health centres II, III, IV, and hospital, as well as a private-for-profit (PFP facility) per district/city in these documents. Support supervision was assessed in two directions, first between DHI and districts and second between districts and their associated health facilities

The primary outcomes included malaria routine reporting attitude and practices among district-level HMIS managers, handling routine/monthly reports that are regularly compiled and submitted to the district by health facilities.

To assess attitude, questions addressing considerations made prior to key actions, desired frequency of key actions, desired formats of action outputs, and perceptions of HMIS in decision-making, were addressed.

To assess practices, questions addressing actions and time to actions, such as: quality assessment, follow-up on quality concerns, and process tracking were used

Results showed that more than one person at each district-level health office was actively engaged in routine HMIS reporting activities, including regularly reviewing, for quality, the data reports submitted to the district. Moreover, district-level managers were able to enlist the support of facility-level data staff whenever necessary, a clear indication of wide-spread collaboration on data and routine surveillance within the districts. At this performance level, however, up to a quarter of managers interviewed did not document the issues, on quality or accuracy, identified in the data reports received from their health facilities when reviewed at their district-level health offices. Whilst majority of non-documenters of data issues opted for the more pragmatic approach of calling the corresponding facility staff for clarifications, this practice, if inconclusive at the time of identifying an issue, increases the chances of issues remaining perpetually unresolved with little to no likelihood of follow-up. This may subsequently propagate data errors or missingness to next reporting levels, affecting overall data quality. Furthermore, with less than one third of the queries documenters able to easily access their queries record, the impetus to address these queries may need further strengthening through say, a standardized procedure of documenting and resolving identified data issues. Overall, a lack of defined standards and/or procedures to document, track, and monitor data quality and thereby mitigate data issues was evident.

The surveillance system in Uganda boasts of high health facility reporting rates, corroborated by this study [22]. However, results show that up to a quarter of the realized reporting depends on extensive follow-up with reporting health facilities by district-level managers. This highlights a demand for resources, such as time and communication costs, to perform this follow-up, but also that interventions to improve voluntary timely submission at the data sources may be vital. Importantly also, there need to be mutually beneficial feedback loops for DHM and other levels–not simply to correct mistakes, but also to the impact of the submitted data.

Relatedly, a majority of district-level managers completing their data entry into DHIS-2 within the first 15 days of the new month directly reflects the time during which all follow-up for data submission was ensured. This suggests that the first half of the new month is a rather engaging time for district-level HMIS managers to ensure full surveillance compliance and therefore, may not be suitable for other activities, such as those requiring these managers’ involvement outside of their duty stations.

Following data submission, an overwhelming majority of district-level managers received queries from the highest level, DHI, ostensibly due to a combination of vigilance at the higher level and laudable functionality of the data management system at auto-generation of queries. However, there was low use of queries’ documentation, with much less than half of participants able to find or show their queries record. This amplifies the urgency for heightened vigilance with queries-follow-through by these district-level managers, given a fairly high average query rate of four per facility. Such vigilance may lead to further improvement of data quality, if intentionally maintained.

For internal use, an encouraging majority of district-level managers (nearly 70%) conducted analysis of their reported data monthly, a higher analysis regularity than the quarterly desired by a slight majority. The internally generated analytical output formats were closely mirrored by desired analytical output formats, be they externally sourced. A notable difference was in the quest for geographical outputs at 27% versus nearly absent practice to generate them. These findings suggest a good mix of district-level managers’ awareness, skills, and confidence around the use of summary tables, normal channel plots, trend plots, and pie charts, their need for advanced skills in generating the same. Moreover, they may indicate widespread demand for analytical feedback that incorporates already well understood result formats such as the normal channel plots. These plots, comprising time series of the monthly median and the upper 3rd quartile of case counts, are used to observe for time points where monthly counts exceed the 3rd quartile—indicating an epidemic—and are recommended by WHO for epidemic monitoring [23]. Overwhelmingly, however, the difference between nearly absent use of maps (3%) as an internally generated analytical output and desire for maps (26.5%) to be included in externally sourced analytical outputs was profound. This may be suggestive of either low awareness of maps as an important means of surveillance data interpretation or a skills gap for this moderately recognized analytical tool and output format, among district-level managers.

Whilst, the different analytical outputs, internally generated or desired to be outsourced, weren’t each linked to intended specific application(s) or purpose(s), an open-ended discussion of general uses of these outputs raised vital insights. The most cited, and therefore, most cross-cutting uses of surveillance analytical outputs, included: general and health facility performance review; planning, decision-making, and resource allocation; as well as, identification of gaps or areas for improvement and supporting health facilities. This broad array of widely recognized uses of routine data analytical outputs across districts suggests a high level of regard and utility of these routine data by HMIS managers across Uganda. This could be supportive evidence towards achieving the data-use strategic objective of the prevailing malaria strategic plan 2021–2025, at the districts level [11]. Among the notable but little mentioned data-uses, however, were: supplies re-distribution, that refers to situations where some facilities find themselves overstocked with a particular commodity, at a time when other(s) are stocked out, and the overstocked share(s) with those stocked out; targeted mentorship that may point to human resource capacity building, particularly given that problem of high staff turnover was raised by several participants; and, support supervision. Unfortunately, the very low consideration of data to inform support supervision correlates with the apparent linking of support supervision primarily to resource availability rather than identified gaps, which could be an area for systemic improvement.

To ensure full adherence to the required data reporting standards at any given level, support supervision, from the next higher level internal or otherwise, is extremely critical [24, 25]. Support supervision enables problem-root-cause identification concurrently with collective solution development at the highest proximity to the source [26]. Whilst DHI support supervision of the district-level managers is happening, there are gaps between desired (district) and actual (DHI) regularity of this supervision as well as in documentation of findings or recommendations when visits happen. Districts largely felt that a visit once every quarter from the DHI would be preferred to ensure good performance. However, a lot still needs to be done since approximately half of district-level managers reported having not received a single supervision visit through the past year and an even greater majority had no tools to record findings or recommendations of such a visit.

Regarding district-level managers’ provision of support supervision to their facilities, a slightly smaller difference was observed between their current practice and their desired frequency. This closeness of perception and practice performance could arguably have been subject to desirability bias. However, with the main reason for the quarterly timescale for majority in both the desired and practiced time-scale, being ensured alignment with resource availability and reporting timelines, it’s unlikely that desirability significantly influenced participants’ responses. Nevertheless, the apparent widely entrenched idea of linking supervision activities to resource-available cycles may risk establishing seasonal vigilance across the reporting system rather than intentionally sustained efforts to improve evidence generation and health service provision.

Merits of this study notwithstanding, there were limitations worthy of mention. First, whilst several open-ended questions were included in this survey, a fully comprehensive qualitative explanation or interpretation of the observed perceptions and practices may not be possible, limiting potential insights drawn. However, numerous questions provided complementary findings to each other and together with the open-ended discussion results, these ensured consolidated learnings from this study. Secondly, most studies assessing perceptions and practices often consider quantitative performance on these. However, this was not feasible in this study due to the small number of district-level offices and thereby, participants. Nonetheless, such quantitative estimates predominantly inform knowledge performance, which was not a practical focus of this study. Thirdly, the small numbers further limited capacity to assess and infer any potential associations between observed perceptions or practices and potential factors, albeit without detriment to the credence of the main study findings. Lastly, whilst mistakes in the data were a key part of this assessment, the actual mistakes identified by DHM officials were not particularly examined to assess their extent or classify them so as to identify patterns and therewith infer possible solutions.

More than one individual manages district-level routine reporting for malaria, with engagement of health facility officials common where needed and the first 15 days of the new month a critical window to ensure complete data reporting. Analytical outputs from routine data, generated by participants remain heavily monitoring-focused and very limited towards impact assessment suggesting a skills gap or underutilization of routine data. The current strong linking of support supervision to resource availability cycles limits its capacity to shore up data quality and general health system improvement. These findings point to the need to: establish standardized data-query handling procedures; improve health system-wide appreciation and adoption of support supervision; as well as, institutionalize advanced analytical skills’ training aligned with key routine data use-cases, to ensure cutting-edge evidence, improved data use, and thereby, impactful decision-making.