Laboratory malaria diagnosis is increasingly receiving much attention due to observed high rate of misdiagnosis  and adoption of more expensive anti-malarial drugs. However, previous studies have shown that the existing health laboratory system in Tanzania is incapable of maintaining good laboratory facilities that can support appropriate diagnosis of malaria and other infections that would lead to proper management of patients [6, 21]. Although few facilities do provide laboratory services, insufficient trained laboratory personnel and clinicians' diagnostic practices are likely to be important obstacles in appropriate malaria diagnosis. These obstacles are common in other malaria endemic areas as reported from a study in Ghana .
This study has shown that the majority of laboratory personnel were not adequately trained and some were lacking professional qualifications. It was also shown that training opportunities and refresher courses for laboratory staff were rarely available while poor working environment was cited by most of the laboratory staff as another constraint for provision of better malaria diagnostic services. Most of the constraints mentioned by laboratory personnel involved in this study have also been reported elsewhere in sub-Saharan Africa [15, 22].
In the current study, one third of the patients referred to the laboratory for malaria investigations were reported to have malaria. However, a survey to establish the accuracy of malaria microscopy in the same health facilities  revealed that the ability of the laboratory personnel in detecting malaria infection by microscopy was approximately 50% (Kappa value, κ = 0.489). This means that most of the patients reported to have malaria parasites did not actually have the parasites. The reported relatively high malaria slide positivity rate coupled with the observed presumptive diagnosis practices indicates high level of malaria misdiagnosis in the study health facilities.
Diagnosis of malaria using clinical presentation was evident in the current study. Nankabirwa and others  reported that such practices are highly sensitive in detecting malaria cases but their specificity was low. In the current study, less than half of the patients diagnosed to have malaria using clinical features were detected to have malaria parasites by microscopy. Likewise, the ability of fever to predict true cases of malaria was estimated at about one third of slide positive cases. Thus, managing fevers as malaria has a potential risk of malaria over-diagnosis and under-diagnosis of other febrile illnesses.
Despite the fact that laboratory malaria diagnostic services were available in all study health facilities, standard criteria for who to test was lacking and test results were underutilized in management of patients. Request for laboratory malaria test was unguided as some clinicians requested the test always while others ordered the test infrequently. Furthermore, less than half of the interviewed clinicians indicated to trust and use laboratory results whenever they were available while more than half indicated that test results have less influence on how they treat patients. Similar findings have been reported in Ghana . This observation agrees with other studies which showed that with optimal laboratory testing clinicians' perceptions and practice remains one of the major barriers to effective laboratory use[14, 16].
The majority of clinicians reported that results from their laboratories were reliable, but relatively few were always using them in malaria case management. It has been argued that, improvement in diagnostic sensitivity alone may not translate into improved patient care . Thus, it is proposed that training of clinicians and appropriate supportive supervision may possibly change diagnostic behaviour of the clinicians . In the current study, it was observed that some clinicians were not always requesting for malaria microscopy because laboratories service hours were relatively shorter than health care provision time. This might have provided a honest room for further justifying presumptive diagnosis. For the laboratories to support accurate disease diagnosis the working hours should as far as possible match with health care provision time.
Some studies [18, 19] have shown that patients or their care-takers prefer laboratory investigations before anti-malarial drug prescription and patients with positive malaria test tend to be more satisfied. This was evident in the current study in which some patients demanded for malaria laboratory testing without requests from clinicians. Moreover, patients with positive malaria test were more satisfied compared with those with negative results. Thus, it can be argued that, the findings that some patient requested malaria test before anti-malarial drug prescription is an indication that the role of patients in malaria over-diagnosis is minimal. However, the decision by laboratory personnel to examine some patients and deny others who visited laboratory without prior consultation with clinicians was not clear. This is an indication of lack of standard operating procedures in the respective laboratory facilities. It was also evident that, the time spent by patients waiting for their laboratory results was not critical, as most of them were satisfied with laboratory services. This may be partly due to the fact that majority of the surveyed laboratories were using Field's stain , a technique that takes relatively shorter time than Giemsa technique and also since laboratory services were few, patients had to appreciate services they were getting from the few available laboratories. However, holding a patient for more than one hour waiting for laboratory results which will not guide appropriate drug prescription is unjustified.
This study was designed to collect information from clinicians and laboratory personnel working at the selected health facilities at the time of investigators' visit. The laboratory personnel and clinicians were aware that their diagnostic practices are being observed and this might have modified their actual practices. These factors affect generalization of the findings of this study. However, the fact that these findings corroborate those of other similar studies elsewhere may still have local or regional relevance. The findings highlight the complexity of the factors that might be responsible for poor quality and low utilization of laboratory results for malaria diagnosis.