Treatment guided by rapid diagnostic tests for malaria in Tanzanian children: safety and alternative bacterial diagnoses

The features of malaria overlap with those of other common illnesses and, where diagnosis is based on clinical history and examination alone, this often results in a large degree of overtreatment for malaria [1]. For many years this has been considered as a price worth paying to ensure prompt treatment of malaria in young children, but over the last decade a number of important trends have driven a movement towards parasitological testing to guide the use of anti-malarial drugs [2]. These include the replacement of older anti-malarial drugs with the more expensive artemisinin combination therapy (ACT), the need to minimize selection pressure for drug-resistant strains of Plasmodium falciparum, the need for more reliable surveillance data as malaria has declined in many areas of Africa, and the relative neglect of alternative diagnoses to malaria [3‐7].

In parallel to this, accurate and affordable rapid diagnostic tests for malaria (RDTs) have become available over the last decade [8] and the 2010 WHO malaria treatment guidelines now propose that wherever possible all patients suspected of malaria should be tested by blood slide or RDT and only those with a positive test result should receive anti-malarial treatment [9]. While this should lead to a number of benefits, there is still relatively little direct evidence that non-malarial treatment in young febrile children in highly or moderately endemic areas is safe, with a concern that a significant number of malaria cases may be missed. The risks of 'test negative malaria' are consistently raised by clinicians cautious about this approach [10]. In addition clinicians need more evidence on the probabilities of different non-malarial diagnoses, especially those that require treatment with antimicrobial agents.

This study was conducted in a malaria endemic area of Tanzania to document the clinical outcomes and blood culture results of children with febrile illness whose treatment had been guided by the results of a commonly used RDT.

The study was conducted over one year from April 2009; 2592 children were screened, and 965 were eligible and enrolled (Figure 1). Of these, 158 were RDT-positive and treated with ALu and 807 were RDT-negative and received non-anti-malarial treatment. Baseline characteristics by RDT result are shown in Table 1; children who were RDT-negative were on average younger with a lower axillary temperature and more likely to have a history of cough or difficulty in breathing (Table 1).

RDTs had a sensitivity of 98.6% (95%CI 97.8-99.3) and specificity of 97.3% (95%CI 96.3-98.4) when compared to double-read research slide results; 14.3% of blood slides were positive with a negative predictive value of RDT results of 99.8% (95%CI 99.4-100).

Of the 807 patients who were RDT-negative, 765 (94.8%) remained in the study for the 14 days of follow-up at which time 461 (60.3%) of whom were reported by the caretaker to be 'completely well'. Of 42 patients who exited before the end of the study, the parents of 6 children administered an anti-malarial drug during the 14-day follow-up period and all of these were RDT-negative at their next research visit and a further six children became RDT-positive after enrolment and were treated with ALu, two of whom were slide-positive with 16,676 and 1,234 asexual parasites/μl respectively and all of whom were PCR-negative for P. falciparum DNA at enrolment. Twenty children were lost to follow-up of whom 6 (30%) were untraceable and 14 (70%) had left the study area but were reported (by telephone or friends) to be well at the time they exited the study. In addition, 12 children were admitted to hospital for whom details are provided in Table 2.

Diagnoses made by the study clinician at the first consultation are shown in Table 3. Almost half (383, 47.5%) of RDT-negative and 29 (18.4%) of RDT-positive children met WHO criteria for non-severe pneumonia and were prescribed amoxicillin and an additional 34 (4.2%) of RDT-negative and 19(12.0%) of RDT-positive children also met these criteria but were not treated with amoxicillin. There was no evidence of a trend of the proportion of children with raised respiratory rate or a recent history of 'cough or difficulty in breathing' with month of the study that might indicate increased prescriber confidence to withhold antibiotic treatment (p = 0.37 and p = 0.43 respectively). While 73% of respiratory rate values were even, there was no apparent preference for values at or above cut-off levels; the commonest values were 36/min (9.7% of values) and 48/min (15.4% of values).

A bacterial pathogen was identified from blood culture in 9/965 (0.9%) children in the study, 8 of whom were RDT-negative and one was RDT-positive but slide-negative (Table 4). In addition, an organism judged to be a contaminant was isolated in 56 (5.8%) of cultures. All of the bacteraemic children were successfully treated as outpatients and the majority (7/9) were treated with amoxicillin at the first consultation before the blood culture result was available. Clinical and laboratory features associated with bacteraemia are shown in Table 5, none of which reached 5% significance although 'current fever' (measured axillary temperature > = 37.5C) was positively associated with bacteraemia (p = 0.08) and with sensitivity and specificity of 77.8% (95%CI 75.2-80.4) and 46.5% (95%CI 43.4-49.6) respectively. In addition, with the exception of children with S. typhi, all of the children with bacteraemia were 12 months of age or younger

While age, haemoglobin level and presence of respiratory symptoms provide some guide to diagnostic probability, clinical predictors of malaria are still insufficient to guide treatment. The advent of RDTs provides the potential to target anti-malarial treatment on a scale that has hitherto been impossible. However, doubts remain in the minds of many clinicians about the safety of not treating parasite-negative patients for malaria, and these have probably contributed to the observation from a number of studies in Africa that up to half of those with a negative rapid test or blood slide result for malaria are treated with anti-malarials anyway. Additionally the question 'if it is not malaria what is it?' is important for framing a rational diagnostic and treatment strategy. Appropriate use of anti-microbial drugs is a key issue given the high mortality of children admitted to hospital with invasive bacterial disease but whose treatment is restricted by high levels of resistance to commonly used agents [14].

This study adds to other evidence that anti-malarial treatment based on parasitological testing misses few true cases of malaria. Njama-Meyer et al used expert microscopy to guide the treatment of 2,359 illness episodes in children under the age of 10 years in a malaria-endemic area of Uganda; fewer than 1% of initial blood slide readings were false-negatives, and only 13 (0.8%) of 1,602 slide-negative illness episodes became positive over the following even days, and none of these was severely ill [15]. D'Acremont et al followed up 603 RDT-negative children who had not been treated for malaria and only three children developed a positive result in the following week, all of whom were successfully treated [16].

Although the study results are reassuring, the policy of RDT-directed treatment for suspected malaria represents a substantial change in the delivery of routine care in Africa and some questions still remain. Firstly, it is impossible to distinguish between 'new' or 'missed' diagnoses of malaria during routine follow-up of RDT-negative patients. As observed in the current study, a significant minority of children will be admitted for malaria within days of being seen with a non-severe febrile illness and a negative RDT result. Although the PCR primers used in this study were not the most sensitive, the results suggest that the malaria diagnoses following enrolment into the study were 'new' infections and that it is rational and safe to treat RDT-negative children for a non-malarial illness. However, the perception of clinicians and parents may be that a child with malaria has been denied the best treatment as a result of application of new guidelines with RDTs. Given evidence of a strong preference for a diagnosis of malaria among parents and clinicians [17], such an interpretation is quite likely and may undermine efforts to implement the policy of RDT-directed anti-malarial treatment. To guard against this, clinicians should be reminded that 'no test is perfect' and to advise parents to bring their child back if they remain unwell. Secondly, there is still a possibility that presumptive treatment for malaria in all children with fever might provide a benefit due to intermittent prophylaxis that might protect against malaria and/or anaemia. Recent results of trials of intermittent treatment for malaria in children suggest that this is possible, and a cluster randomized trial of presumptive compared to RDT-directed anti-malarial treatment is in progress [18].

The commonest indication for anti-microbial treatment in our study was the IMCI diagnosis of non-severe pneumonia. While the overlap between malaria and pneumonia has been previously described both in severe and non-severe illness, the proportion of children meeting WHO criteria for non-severe pneumonia was higher than expected from previous studies [15, 19‐21]. Preferential recording by study clinicians in order to justify prescribing an anti-microbial drug seems unlikely as the number of children meeting the WHO definition was slightly higher than the number actually diagnosed. In addition, there was no evidence of a change in the proportion diagnosed with pneumonia as the study progressed, nor was there evidence of clustering of respiratory counts near the cut-off values that define raised levels. It's possible that 'cough or difficulty breathing' was over-interpreted or there may be local variations in the perception or reporting of respiratory problems in children [22]. In addition, the finger-prick undertaken approximately 15 minutes before examination may have resulted in transiently increased respiratory rates.

As new guidelines and RDTs are introduced across Africa it is possible that there will be a compensatory over-use of the IMCI diagnosis of non-severe pneumonia but this will be difficult to assess since there is no gold standard definition of the diagnosis of pneumonia. Even of the true pneumonias some will be viral, including the vaccine-preventable RSV pneumonias, where antibiotics will make little positive impact on outcome [23, 24]. In addition, studies of the quality of care suggest that IMCI procedures to diagnose pneumonia, in particular examining the chest and recording respiratory rate, are not well undertaken outside research settings [25, 26].

Consistent with at least one other study, we found that bacteraemia was uncommon in children with non-severe illness, especially in children who were RDT-positive [27]. However, blood culture lacks sensitivity and early recognition and treatment of blood stream infections has the potential to avert progression to severe disease with high associated mortality [14, 28]. While RDTs can improve targeting of anti-malarial drugs there is no comparable test for bacterial disease. The POC measure of blood lactate was increased in children who were RDT-positive compared to RDT-negative, but was unhelpful in distinguishing between children with and without invasive bacterial infection. While the small numbers of children with bacteraemia in our study reduced the likelihood of finding a statistically significant association with clinical and laboratory features, our findings are consistent with other studies that bacteraemia occurs predominantly in young children, up to three-quarters of whom present with a 'current fever'. However, these indicators lack both sensitivity and specificity and there is still a need for more accurate measures of bacterial infection; devices measuring acute phase proteins such as C-reactive protein or procalcitonin merit further research [29].

This study has a number of inevitable limitations. The study was conducted in hospital outpatients where the patient population may differ in a number of ways from those attending primary health facilities generally. The accuracy of RDT interpretation may have been higher than under routine practice, and certainly higher than routine microscopy in some clinics in the region. Clinicians were aware that they were being observed (in the sense that their findings and prescriptions were being recorded) and may have been more likely to follow IMCI guidelines and to prescribe antibiotics, thus over-estimating antibiotic usage that would occur under similar conditions in routine practice. The clinical care provided in the study followed IMCI guidelines and did not have access to an otoscope or urine examination resulting in possible neglect of otitis media or urinary tract infections. The lack of validating investigations for the IMCI diagnosis of pneumonia has already been discussed.

In conclusion, this study demonstrates that use of RDT's to direct the use of anti-malarial drugs in young children is a safe strategy and did not result in any missed diagnoses of malaria. However, new episodes of malaria do occur within a short period of an initial consultation in malaria endemic areas and these may undermine confidence in parasitological testing for malaria. Invasive bacterial disease is relatively uncommon in children with non-severe illness; while most cases occurred in infants with a current fever there is a need for more specific diagnostic tests to detect these infections at the point of care.

The authors declare that they have no competing interests.