During the implementation of a health facility based surveillance system, there were significant improvements in several key steps of malaria case management at the six sentinel sites. The biggest impact was in the proportion of patients with suspected malaria who were referred for diagnostic testing. At the onset of the surveillance system a majority of suspected malaria cases were treated empirically without referral for diagnostic testing. However, utilization of laboratory service greatly improved as 97% of patients with suspected malaria were referred for diagnostic testing during the final three months of evaluation. Treatment practices were also significantly improved in terms of not prescribing antimalarials in patients with negative diagnostic tests, prescribing antimalarials in patients with positive diagnostic tests, and prescribing AL for patients with laboratory confirmed malaria.
The quantitative impact of the surveillance system can be estimated using a simple hypothetical approach comparing observed malaria treatment practices with expected treatment practices assuming no changes in our key indicators after the first three months of observation. Between September 2006 and March 2010, a total of 229,375 cases of suspected malaria were captured by the UMSP surveillance system. Assuming the proportion of patients suspected of malaria with a lab test done and the proportion of patients prescribed antimalarials with positive and negative lab tests would have continued at the same level as the first three months of data collection, the implementation of the surveillance system resulted in 58,678 fewer antimalarial treatments prescribed. In addition, the UMSP almost doubled the number of prescriptions of antimalarials for lab confirmed cases of malaria (32,505 expected vs. 64,366 observed).
The first critical step for improving malaria case management is the referral of patients with suspected malaria for laboratory diagnostic testing. Several studies from Africa have reported less than 50% of patients suspected of having malaria undergo diagnostic testing even when these services are available 
. The surveillance program described here benefitted from a six-day integrated team-based training course of health care workers conducted shortly after the program was implemented. However, even after this training course, the proportion of suspected malaria cases referred for laboratory testing remained just over 50% 
. It was not until after over three years of ongoing surveillance and continued supervisory visits that consistent levels of over 90% of suspected malaria cases referred for laboratory testing at all the sites were achieved. Several lessons were learned over the course of these three years. Patience was required as empiric treatment of malaria without diagnostic testing has historically been part of the national policy in most African countries and deeply ingrained in the teaching of health care workers. Indeed, it is only in the last year that the WHO has made a clear recommendation for the laboratory confirmation of diagnosis in all patients suspected of having malaria before treating in situations where diagnostic testing is available 
. Another important factor was support from the Ugandan Ministry of Health and district focal persons in advocating for the utilization of laboratory services. Feedback and setting targets were also important for encouraging health care workers and building confidence in the value of having a test result for making treatment decisions. Finally, ensuring the laboratories at the sentinel sites were well equipped to handle the large numbers of patients referred for laboratory testing was essential. At most of the sites this primarily involved support for microscopy, which included advocating for adequate laboratory personnel and ensuring adequate supplies needed for making blood smears. At some of the sites, the utilization of RDTs also played a role, especially in areas of unstable transmission intensity where the need for diagnostic testing can fluctuate, and at times overwhelm the capacity for microscopy. However, the role of RDTs was limited by their availability.
The primary objective of the surveillance program was to generate unbiased and precise estimates of the SPR by increasing the utilization of diagnostic testing among cases of suspected malaria. These data are provided to the Ministry of Health and other stakeholders in the form of monthly reports which are also posted on a public website (http://umsp.muucsf.org/
). Although evaluations of trends in SPR were beyond the scope of this report, increasing the use of diagnostic testing provided an opportunity to evaluate the impact of the surveillance program on improving antimalarial treatment practices. The use of diagnostic testing may improve patient care in parasite-positive patients, allow for the identification of parasite-negative patients in whom another diagnosis should be sought, reduce the use of unnecessary antimalarials, and provide confirmation of treatment failures. Approximately 1 in 10 patients with a positive diagnostic test were not prescribed antimalarials during the initial period of the surveillance program. This appeared to be due to clinicians making decisions about treatment prior to receiving the diagnostic test results based on informal discussions with clinic staff. Through continuous training and supervisory visits, clinicians were encouraged to wait for the laboratory result before making treatment decisions resulting in a significant reduction in the proportion of parasite-positive patients not prescribed an antimalarial. A much more common problem early in the surveillance program was the practice of prescribing antimalarials in patients with a negative diagnostic test. Indeed, studies across a wide range of epidemiologic setting in Africa have documented that 35–79% of patients with negative diagnostic test result were still prescribed antimalarial drugs 
. This seemingly irrational treatment practice can be difficult to change as demonstrated by a study in Tanzania where the introduction of RDTs and basic training did not lead to a reduction in overuse of antimalarial drugs 
In this surveillance program, the proportion of patients with a diagnostic test done who were appropriately prescribed antimalarial therapy increased after the JUMP training program, however, some sites failed to sustain these improvements or showed declines. Only after three years of the surveillance program were we able to reach levels greater than 90%, although two sites continue to prescribe antimalarials in up to 20% of patients with a negative diagnostic test. Again, promoting rational antimalarial treatment practices took patience, continual feedback to the health care providers, and support from government officials at the Ministry of Health and district level. In the era of ACTs, limiting the unnecessary use of antimalarials becomes a high priority as this will help maintain drug supplies, reduces health system costs 
, and might reduce opportunities for the selection of drug resistant parasites. The surveillance program was less successful in promoting the use of AL in parasite-positive patients. Although there were some modest gains, AL treatment practices varied widely over time and between sites. Although data on the reasons for not prescribing AL were not collected systematically, informal discussions with health care workers suggested that the primary factor responsible was drug stock-outs. As ACTs are being rolled out in large numbers around Africa, the ability to maintain a consistent drug supply has become a major issue and has been cited as a major factor in health care workers choice of antimalarials in Uganda and Kenya 
There are several important limitations of this study that should be pointed out. The surveillance program was not implemented as a controlled experiment, therefore causal inferences between the intervention and improvement in indicators of improved malaria case management should be made with caution. Secondly the various components of the surveillance program were not implemented in a systematic fashion. Rather improvements to the program were made over time based on experience, need, and a “trial and error” basis. Indeed, the surveillance program did not establish a pre-specified list of qualitative or quantitative goals in terms of interventions, but rather interventions were developed and implemented as a means of continuously improving indicators of malaria case management. Finally, data collected were limited to the practices of health care workers and did not include exit interviews or follow-up surveys. Therefore it is unknown whether improvements in health care worker performance lead to improved patient outcomes. Several “downstream” factors such as proper dosing, successfully filling prescriptions, adherence to medications, and treatment seeking practices after leaving the clinic are all important for successful malaria case management.
In summary, although the absence of a “control group” limits the ability to make causal inferences, the experience of UMSP provides evidence for the utility of a health facility-based sentinel site malaria surveillance system that produces high quality data in Africa given that adequate resources are available. In addition to improving the capacity to monitor trends in malaria morbidity and measure the impact of control interventions in these selected sites, there is added value by improving malaria case management for large numbers of patients. Indeed, surveillance itself should be considered an intervention and an integral part of any malaria control program. Success of the program did not occur overnight, but rather required patience, flexibility, feedback from heath care workers, and continuous support from the government and funding agency. Although the malaria surveillance program described here has not been expanded beyond the sentinel site health facilities, lessons learned from this program should benefit other initiatives aimed at improving malaria case management in other health care facilities and provides a demonstration project for changing the practices of health care workers in Africa.