Use of malaria rapid diagnostic tests by community health workers in Afghanistan: cluster randomised trial

Acute febrile illness is one of the commonest presentations to clinics in Asia, with malaria being one of multiple possible causes. Malaria rapid diagnostic tests (mRDTs) are increasingly used across health service settings to improve diagnosis and treatment of febrile illness by detecting malaria [1]. The World Health Organisation (WHO) recommends universal coverage with diagnostic testing for malaria to ensure that patients are appropriately prescribed artemisinin combination therapy (ACT) and other antimalarial drugs [2] and improving treatment of other causes of febrile illness. Wider use of mRDTs should also improve disease surveillance, particularly in areas of low to moderate transmission where malaria elimination is being considered.

There is now reasonable evidence to support WHO policy on the effect of using falciparum-specific RDTs in highly endemic areas in Africa, where almost all malaria is caused by Plasmodium falciparum [3‐7]. As a result, RDTs are now widely used in Africa, and their use is growing [8]. There is much less evidence from Asia, where more than one billion people live in malaria endemic areas [9] which are co-endemic for P. falciparum and P. vivax and where the proportion of febrile patients who have malaria is generally much lower than in Africa. Afghanistan is typical of much of low-resource south Asia, where P. vivax makes up around 85–95% of malaria cases [10].

Community health workers (CHWs) are often the first point of contact with healthcare services for those with fever or a history of fever. In common with much of south Asia, CHWs in Afghanistan have limited training and do not have access to laboratory testing. Currently malaria reported in this setting is diagnosed using clinical signs and symptoms alone which are indistinguishable from many other causes of fever. This is termed ’suspected malaria’ in training curricula and treatment guidelines and is often treated presumptively with chloroquine (CQ) and/or sulphadoxine-pyrimethamine (SP) since there is no way of distinguishing the species of malaria necessary to guide treatment. This approach leads to misdiagnosis and overtreatment of malaria, sometimes resulting in 100% mistreatment and antimalarial wastage in areas of very low malaria endemicity [11, 12]. In initial testing in health facilities where healthcare workers have formal training, an earlier study in this Afghan setting showed that diagnosis using mRDTs led to improved prescribing of antimalarial drugs at an individual level and an increase in prescribing of antibiotics [11].

We therefore conducted a two-arm, stratified cluster randomised trial amongst communities within the formal health service in two provinces of Afghanistan to investigate the effect of training and the use of mRDTs on accurate diagnosis and appropriate prescribing of antimalarials and antibiotics by CHWs compared to normal clinical diagnosis. A cluster randomised design was chosen because a patient randomised trial in the setting of CHWs is unlikely to give accurate estimates of effect. This happens because prescribers influence one another (a healthcare worker community effect), and perceptions of interventions may influence normal practice by the community (a patient community effect) [13].

Patients were enrolled between October 2011 and May 2012, 22 clinics were randomised and of the 256 CHWs, 222 (86.7%) received training in the study data collection methods, consented and enrolled patients. Those who did not enrol patients were either not present for training or did not give consent to participate. The trial ended when the sample size had been reached.

This trial in moderate and low endemic areas of Afghanistan, typical in both epidemiology and healthcare provision to much of low-resource south Asia, showed that provision of mRDTs to CHWs together with practical training on correct use led to a complex picture. Targeting of antimalarials (the primary outcome) improved significantly, mediated by much lower rates of prescribing antimalarials to patients without malaria parasites. This useful outcome was offset by a significant reduction in prescribing antimalarials to those who did have malaria, both vivax and falciparum. ACTs were given to only a quarter of cases of falciparum malaria in the intervention arm. Antibiotic prescribing increased in the intervention arm, possibly indicating a switch from overprescribing antimalarials to overprescribing antibiotics.

In comparison to PCR, the operational sensitivity of WHO-approved mRDTs when used by CHWs was poor, possibly because of infections with low parasite density, which are common in low-transmission settings. Limited laboratory testing has suggested that, in common with many other tests, whilst the mRDTs used in this study have good sensitivity for vivax malaria at high parasitaemias, it may decrease rapidly (to around 50% at low parasitaemias) [18]. However, the prescribing practices of CHWs were highly concordant with RDT result, suggesting that if tests appropriate to the local setting were available, there would be a substantial positive effect on prescribing of antimalarials. In African settings CHWs have proved willing to adapt practice to RDTs [19, 20]. CHWs start with a history of fever in considering malaria; in the absence of either clear alternative diagnoses, or a test, malaria is unlikely but almost impossible to exclude.

WHO currently recommends diagnostic testing with WHO-approved RDTs prior to prescribing antimalarials. Trials in Africa including those in the formal secondary setting [21], district health posts (similar to this setting) [22] and private sector providers [23] support this approach, although health worker adherence to test results is variable [6, 24]. Translating this to the much lower incidence settings where most malaria is vivax and parasite density will be lower should be undertaken with caution in the absence of relevant data. This trial supports the principle that CHWs will change prescribing behaviour and improve targeting in acute febrile illness with mRDTs and training, but raises significant practical concerns. In particular, without mRDTs with greater operational sensitivity in this epidemiological setting, the risks of missing true cases of malaria, including falciparum, are non-trivial. In Africa the risks of missing falciparum malaria are emphasised early in training, and studies show it is very rare for test positive cases not to be prescribed antimalarials [13]. In south Asia the true risk of malaria is lower, and it is possible that this drives behaviour [25].

The introduction of mRDTs, whilst reducing the overuse of antimalarial drugs, had the effect of increasing the use of antibiotics (co-trimoxazole) in malaria negative cases; this finding is similar to operational studies in Africa [5, 22], and recent studies suggest that this is a widespread concern in the context of rising antibiotic resistance as a consequence of overuse [26]. A previous individually randomised trial at a higher level of the health system suggested that the better trained personnel in those centres are more likely to diagnose malaria with mRDTs [11], but also tend to switch from overuse of antimalarials to higher use of antibiotics. Interventions are required which improve prescribing practices in patients with non-severe fever.

The study was designed to assess the effectiveness of the mRDT intervention under field operational conditions within a widely accessed part of the health service in malaria endemic areas of Afghanistan, and has the advantages and limitations of operational studies. The cluster design allowed a robust measurement of the effect of mRDTs on prescription of antimalarial drugs by CHWs and is accepted as a more reliable approach than individual patient randomised trials conducted in clinics, because clinicians influence one another. Training followed the approved national guidelines for the use of mRDTs, which closely match the WHO standard guidelines for training and use of mRDTs, and so is generalisable. The clinics chosen were representative of most of Afghanistan’s health service, and were similar to many CHW systems in low-resource south Asia. The main findings and conclusions should therefore be applicable to other countries in the region. The main limitations were imposed by having to work through relatively unskilled CHWs in a relatively unstable area. In particular it was not realistic to get blood slides made to supplement PCR data and/or to follow patients up into the community. PCR is significantly more sensitive than microscopy, but in a low-transmission setting it is unlikely that a PCR positive malaria case would not be clinically relevant [27]. If microscopy could have been reliably used as a gold standard in this clinical setting, then sensitivity of the test may have been reportedly higher. Lastly, a cost-effectiveness analysis was not part of this study, but it is an important consideration for policy decisions driving the placement of mRDTs at the facility and/or community level [28].

Use of mRDTs by CHWs has a potential advantage of improving surveillance of disease based on parasitological diagnosis. This is particularly relevant to malaria epidemiology in Afghanistan, which is heterogeneous. In the north of the country there is very little transmission, and elimination is being actively pursued. This requires an ability to detect and respond to outbreaks with preventative interventions such as insecticide-treated nets. However, the performance of the mRDTs at the community level may reduce the attractiveness of this tool in favour of more sensitive tests. In the east of the country , the malaria programme is still in the control phase, where mRDTs could provide improved surveillance and targeting of preventative measures.

The increase in antibiotic use in the intervention arm is potentially a concern. The World Health Assembly and UN General Assembly have recently highlighted the serious risk that is posed by antibiotic resistance. Overuse of antibiotics in south Asia is driving substantial antibiotic resistance [29]. If one of the effects of mRDTs is to lead to a switch from overprescribing of antimalarial drugs to overprescription of antibiotics, this would be a concerning unintended consequence.

In low prevalence areas such as Afghanistan, mRDTs for malaria have at least three potential roles: to identify the relatively uncommon cases of true malaria, especially falciparum malaria; to provide epidemiological surveillance data in areas considering elimination; and to reassure prescribers that not prescribing an antimalarial is reasonable. The main advantage of mRDTs at the community level is to reduce the overuse of antimalarial drugs in malaria negative patients. However, currently available mRDTs may miss a substantial proportion of cases of malaria. Although using mRDTs led to reductions in the number of inappropriate doses of antimalarials, mRDTs may also lead to overprescription of antibiotics. Any potential role for improving surveillance should consider the operational sensitivity of mRDTs, which may be too low for detecting some malaria infections; higher sensitivity tests may be useful in these settings. The risks identified in this study need to be balanced against the advantages of rolling out mRDTs amongst CHWs.