Effects of revised diagnostic recommendations on malaria treatment practices across age groups in Kenya

Objective The recent change of treatment policy for uncomplicated malaria from sulfadoxine-pyrime-thamine to artemether-lumefantrine (AL) in Kenya was accompanied by revised malaria diagnosis recommendations promoting presumptive antimalarial treatment in young children and parasitological diagnosis in patients 5 years and older. We evaluated the impact of these age-specific recommendations on routine malaria treatment practices 4–6 months after AL treatment was implemented. Methods Cross-sectional, cluster sample survey using quality-of-care assessment methods in all government facilities in four Kenyan districts. Analysis was restricted to the 64 facilities with malaria diagnostics and AL available on the survey day. Main outcome measures were antimalarial treatment practices for febrile patients stratified by age, use of malaria diagnostic tests, and test result. Results Treatment practices for 706 febrile patients (401 young children and 305 patients ≥5 years) were evaluated. 43.0% of patients ≥5 years and 25.9% of children underwent parasitological malaria testing (87% by microscopy). AL was prescribed for 79.7% of patients ≥5 years with positive test results, for 9.7% with negative results and for 10.9% without a test. 84.6% of children with positive tests, 19.2% with negative tests, and 21.6% without tests were treated with AL. At least one antimalarial drug was prescribed for 75.0% of children and for 61.3% of patients ≥5 years with a negative test result. Conclusions Despite different recommendations for patients below and above 5 years of age, malaria diagnosis and treatment practices were similar in the two age groups. Parasitological diagnosis was under-used in older children and adults, and young children were still tested. Use of AL was low overall and alternative antimalarials were commonly prescribed; but AL prescribing largely followed the results of malaria tests. Malaria diagnosis recommendations differing between age groups appear complex to implement; further strengthening of diagnosis and treatment practices under AL policy is required.


Introduction
Antimalarial treatment with highly efficacious but expensive artemisinin-based combination therapy (ACT) is a recent key strategy to reduce the public health impact of failing monotherapies in sub-Saharan Africa (White 2006;WHO 2006). In 2004, Kenya changed its first-line treatment policy for uncomplicated malaria from sulfa-doxine-pyrimethamine (SP) to a specific ACT, artemetherlumefantrine (AL). Quinine became the treatment of choice for children below 5 kg, pregnant women, and as the second-line; SP was reserved only for intermittent preventive treatment in pregnancy; and amodiaquine (previous second-line treatment) was no longer recommended (MoH 2006a).
To rationalize the use of expensive ACTs, the WHO promotes parasitological diagnosis, except for young children in high malaria risk areas where, pending further evidence, the clinical consequences associated with not treating potentially false negative test results may outweigh OnlineOpen: This article is available free online at www.blackwell-synergy.com *Re-use of this article is permitted in accordance with the Creative Commons Deed, Attribution 2.5, which does not permit commercial exploitation. In 2006 the new AL policy was implemented countrywide and evaluated approximately 4-6 months after AL was delivered to health facilities. Descriptions of the implementation process and overall paediatric malaria case-management results were published earlier (Amin et al. 2007;Zurovac et al. 2008). We report here how agespecific malaria diagnosis recommendations were translated into routine malaria treatment practices at facilities with diagnostic capacity.

Methods
Between October and December 2006, we evaluated outpatient malaria case-management at government health facilities in four Kenyan districts (Kwale, Bondo, Kisii ⁄ Gucha and Makueni). The detailed explanation of survey methods is presented elsewhere . Briefly, the survey was a cross-sectional, cluster sample survey undertaken at all 193 government facilities in the four districts. Data at each facility were collected over 1 day using quality-of-care assessment methods including health facility assessments, health worker interviews, and exit interviews with caretakers and patients. During the exit interviews the study nurses collected information about patient's age, weight, temperature, history of fever, pregnancy status, main complaints, prior use of antimalarial drugs and if the visit was an initial or follow-up consultation. Information was also collected from patientheld records about routine diagnostic procedures requested and results reported, medications prescribed, and if the patient was treated as an outpatient or referred for hospitalization. The present analysis focused on routine antimalarial treatment practices for patients weighing ‡5 kg presenting for an initial consultation and on outpa-tient visit with a history of fever or axillary temperature ‡37.5°C. Pregnant women and patients ‡5 years presenting with another obvious cause of fever were excluded from analysis. Since malaria diagnosis recommendations differ between age groups, we stratified this analysis for patients below and above 5 years of age. Treatment practices were further stratified by use and results of malaria tests. Given the small number of patients at facilities with malaria RDTs, which precludes a meaningful analysis stratified by type of diagnostics, the combined results from all facilities are presented. To ensure comparable evaluation of treatment practices, the analysis was restricted to facilities where AL was available on the survey day. The precision of proportions [95% confidence interval (CI)] was determined adjusting for the cluster sampling.
The pattern of antimalarial treatments was similar among febrile patients below and ‡5 years ( aquine and SP were the most commonly prescribed antimalarial treatments and over 80% of patients were treated with antibiotics (Table 1).

Discussion
Kenya's policy to promote presumptive malaria diagnosis in young children and parasitological testing in patients ‡5 years should have resulted in substantially different patterns of malaria diagnostic and treatment practices between these age groups. However, the only difference found was somewhat higher use of malaria testing in patients ‡5 years (43%) than in young children (27%). In both age groups, any antimalarial drug was prescribed to the majority of patients with a positive test (98-99%), negative test (61-75%) and those without test performed (50-64%). AL, however, was rarely prescribed for all patients' categories (10-22%), except for those with a positive test result (80-85%). The observed pattern is likely to reflect modalities of AL implementation process in Kenya. Prior to the introduction of ACTs, several studies in Kenya (Zurovac et al. 2006a) and elsewhere (Barat et al. 1999;Holtz & Kachur 2000;Reyburn et al. 2007) reported high rates of negative test results treated with an antimalarial drug. As a response to the concern that the continuation of this practice under AL policy would waste new expensive drugs, particularly in older age groups (Zurovac et al. 2006b), the new guidelines unambiguously discouraged this practice for patients ‡5 years (Ministry of Health (MoH) 2006a). Simultaneously, prompt and effective presumptive treatment of young febrile children was promoted in line with international recommendations (Gove 1997;World Health Organisation (WHO) 2006). However, the translation of new guidelines into clinical practice faced several challenges. Firstly, although the guidelines recommended the use of malaria microscopy or RDTs for patients ‡5 years, there was no decision taken to procure and supply RDTs on larger scale. Secondly, the training of health workers greatly emphasized the high cost of AL demanding its rational use for test-positive adults using microscopy or RDTs, with less focus on providing prompt AL treatment for febrile children. Indeed, all health workers had been trained how to perform RDTs, their potential in peripheral facilities had been emphasized and even limited quantities of RDTs were delivered for training purposes (MoH 2006b). Thirdly, training messages commonly included incorrect information about expecting RDT supplies, compulsory testing before prescribing AL for all age groups, and adequate efficacy of non-recommended antimalarials (Wasunna et al. 2008). Finally, the national guidelines were somewhat ambiguous as they stated 'parasitological diagnosis is not prerequisite for the treatment' in febrile children. Therefore, it was not surprising that we found similar age-specific practices characterized by suboptimal use of malaria diagnostics and prescription of AL predominantly for the small subset of test positive patients.
Antimalarial treatment practices for test negative older children and adults deserve special attention. Although the use of malaria diagnostics was low, our previous concerns that health workers would massively overprescribe AL for these patients appear to be unfounded (Zurovac et al. 2006a,b;Reyburn et al. 2006). The change of policy to AL has not eliminated unnecessary use of other inexpensive antimalarial drugs, but the new policy, however, had an impact on rational use of AL in this patient group. Although this is an encouraging finding, it should be cautiously interpreted since the pattern might change with longer establishment of AL policy, removal of amodiaquine from the facilities, and more widespread availability of RDTs as demonstrated in Zambia (Hamer et al. 2007).

Conclusion
The translation of revised diagnostic recommendations into effective malaria treatment practices under the AL policy in Kenya faces several challenges. Parasitological diagnosis is underused in older children and adults, young children are still tested and the overall use of AL is low in both age groups with marked tendency for alternative antimalarial prescriptions. Prescribing of AL, however, largely followed malaria test results. Different age-specific diagnosis recommendations appear complex to implement. The most suitable solution for policy implementers would be to have the same recommendations for all age groups. RDTs are a potential solution for young children, as suggested recently in urban setting in Uganda (Njama-Meya et al. 2007); however more evidence across different settings is required to document if not treating test-negative children in high risk areas can produce significant cost-savings without resulting in harmful clinical consequences. Meanwhile, the priority for the Kenyan MoH should be strengthening of AL implementation activities based on presumptive treatment for young children and parasitological diagnosis, including RDTs, for patients 5 years and older.