It emerged from our study that the malaria ES system of Tambacounda health district had some strengths but also shortcomings in its attributes which were not favorable to the reduction of morbidity and mortality related to malaria.
We surveyed 27 health facilities with catchment areas covering the entire district. Most of these structures were headed by state-enrolled nursing assistants (44.4%) attesting to the low level of qualification at the head of the SDP. The managers of these structures had considered that the malaria ES system was acceptable despite the fact that the acceptability attribute was subjective. It can, however, be attested by the completeness rate of the reports of the SDP surveyed that were good. This could be related to the filling time of the management tools which was correct according to the heads of health structures. Regular monitoring of malaria data at coordination meetings also contributed to performance against the completeness of the reports. In a study conducted in Nigeria, Visa Tyakaray and colleagues also demonstrated that the country's malaria ES system was acceptable [8]. The same was true of Shaojiin Ma's study of Thailand's malaria epidemiological surveillance system where users of the system felt it was also acceptable [9]. In the study on the documentation of epidemiological surveillance of malaria in the river valley in Senegal and in the suburbs of Dakar the same results were found with an epidemiological surveillance system for malaria assessed as acceptable [10].
In our work, even if the system was representative with the possibility of illustrating the data in all the health structures surveyed, only 33.3% of them displayed malaria cases on graphs. The possibility of representativeness was mainly related to primary collection tools that were simple to inform and took into account the characteristics of time and place. These results were contrary to those of the Nigeria study where the malaria surveillance system in Kano State collected data that was not representative of all health facilities, as data from two tertiary health facilities and all private facilities were not included in the state's general data [8]. In Kaduna State, Nigeria, work has also shown that the malaria surveillance system excludes data from private health facilities and distorts the comprehensive representation of cases across locations [11]. Thus, in our work, the representativeness of the surveillance system could also be affected by the under-reporting of Community data which are not considered in the weekly reports but rather in the monthly morbidity reports of health structures.
Of the 27 SDP surveyed in our work, nearly three-quarter of those responsible for these structures felt that the malaria ES system was simple. This finding is similar to that of studies conducted in Oyo and Kaduna States, Nigeria and Chipinge District, Zimbabwe, where surveillance systems were found to be simple and flexible [8, 12, 13]. However, this contrasts with studies in Brazil, Zimbabwe and Angola where systems were complex to operate [14, 15]. At the Tambacounda health district level, this simplicity of the malaria ES system was supported by data collection with easily understandable tools and simple reporting to the next level. In addition, the malaria ES system was also flexible with tools adapted to the epidemiological profile of the Tambacounda health district, which is in a malaria control zone. In all health facilities, the available management tools considered the diagnosis of malaria (with the RDT results columns in the registries) and a possibility of adapting these tools in the event of the introduction of new variables related to ES. These results were like those of Visa Tyakaray et al in Nigeria and those of Chipinge in Zimbabwe and Ebonyi state in Nigeria which also showed that malaria ES systems were flexible and could adapt to changes in malaria data entry [8, 13, 16].
On the other hand, the stability of the malaria ES system in the Tambacounda district was lacking because in more than half of the SDP surveyed, as soon as the person in charge of the structure was absent, there was a discontinuity of the ES service. This failure is much more accentuated by the non-involvement of other health workers in surveillance, as the tools were available in all the SDP visited.
In this work, although all health facility managers took care of suspected and/or confirmed cases of malaria and transmitted the information to the next level, the data produced on the spot were not analyzed by the providers themselves. The majority (85.2%) were unaware of the usefulness of epidemiological surveillance for malaria and, in turn, did not use the data to make decisions. When training health workers on ES in general, and during supervision activities, the focus is not on the aspects of analysis and interpretation of data that very often lead to failures in the analysis of data on site by the providers who produce it. This was not in line with the results of Visa Tyakaray's work which had shown that in Kano State in Nigeria epidemiological surveillance data were used for decision-making and the formulation of policies that guide the functioning of the surveillance system [8].
Even if the data were not used by the providers themselves, the system in place was very responsive with a speed in the transmission of information to the next level that was done in a harmonized way and a diagnostic and care device available in all health facilities. This performance could be explained by the reminders made every Monday morning by the district malaria focal point concerning the transmission and entry of data in the platform but also by the periodic sharing of the level of data transmission in the district electronic working group and during the coordination bodies. This speed of transmission of information in a harmonized manner and the regular monitoring by the district were much more supported using electronic means which improves the speed and facilitates access to epidemiological data thus allowing a faster analysis and response [17].
In addition, the regular monitoring of RDTs and ACTs through RDT reporting sheets and drug stock sheets had largely contributed to the availability of these inputs in all health structures surveyed for a good responsiveness of the malaria ES system.
Despite the good performance in terms of information transmission (96.3% of the SDP in our study), the malaria data transmitted weekly to the district level via the DHIS2 platform, were not consistent with those of morbidity reported monthly in the SDP report. In many health facilities, community data (health boxes and HCP) and private data were left behind in the transmission of malaria ES data, thus explaining the under-reporting in the ES part with a gap of 7,611 malaria cases throughout 2021. Irregular supervision with on-site data verification had certainly exacerbated this data discrepancy between ES and malaria morbidity. In a study conducted in Kaduna State, Nigeria, the results also showed that the malaria surveillance system excluded data from private health facilities [11].
However, there are some limitations to our work. First, it was not representative of all the health districts of Senegal. However, it can be very useful in other contexts, since in Senegal the health districts have the same functioning of the routine epidemiological surveillance system for malaria. In addition, the Community component was not also considered.
Finally, this study could also present a desirability bias because, the participants' responses to self-reported questionnaire can be biased to project a better picture of their performance.