Daniel KYABAYINZE

Thesis summary

Background and rationale

Malaria is one of the leading causes of fever and mortality in sub-Saharan Africa with an estimated 655,000 deaths out of the 300 million cases reported in 2011. Malaria caused 16% of all the 4.2 million child deaths occurring in Africa in 2008. Uganda is ranked 6th worldwide for the absolute number of malaria cases registered. Malaria is endemic in 90-95% of the country and all residents are at risk of getting malaria. Malaria is responsible for 30-50% of out-patients visits, 15-20% of hospital admissions, and up to 14% inpatient deaths.

In a bid to save lives and improve case management, WHO in 2006 recommended for the first time parasitological confirmation before administering antimalarial treatment, with the only exception of children in areas of intense transmission. As the cost-effectiveness of this approach decreases with declining malaria transmission WHO has revised the malaria treatment guidelines in 2010, recommending universal parasitological diagnosis in all suspected cases [2]. Currently many African countries are working to implement the new treatment guidelines using microscopy and RDTs at primary health care (PHC) facilities, the latter when microscopy is not available.

RDTs are visual lateral flow tests that detect parasite-specific antigens in blood. Malaria antigens currently used as diagnostic targets are either specific to a Plasmodium species or conserved across the human malarias. P. falciparum-specific monoclonal antibodies for histidine-rich protein 2 (HRP-2) are more sensitive, affordable, and heat stable compared to the pLDH and Aldolase antibodies. Despite their simplicity, there are challenges for the implementation of RDTs in routine care settings. For example, some clinicians are reluctant to abandon presumptive malaria treatment among the high risk groups such as children under five years and pregnant women, there is limited human and infrastructural capacity, and operational accuracy of tests may vary with individual users. In addition, persistent antigenemia may limit the utility of RDTs in recently treated patients.

In this thesis we evaluated the process and impact of implementing RDTs at primary health care facilities in Uganda. Our specific objectives were i) to assess the health system capacity to incorporate RDTs into the routine management of febrile illnesses, ii) to determine the impact of RDTs on health workers' prescription behaviour, and (iii) to investigate the effectiveness of RDTs in detecting malaria among vulnerable populations (children and pregnant women) in Uganda.

Methods

Suitable study designs were used to answer each specific objective. We investigated health systems capacity to absorb parasite based diagnosis in peripheral health facilities using a cross-sectional survey. A total of 125 health facilities, selected using multi-stage cluster sampling were assessed for 1) tools, 2) skills, 3) staff and infrastructure, and 4) structures, systems and roles necessary for the implementation of parasite-based diagnosis of malaria. A quasi experimental ‘pre - post’ and ‘intervention - control’ analysis design was used to determine the impact of RDTs on health workers’ prescription behaviour. We measured the effectiveness of RDTs in the diagnosis of malaria during pregnancy among 433 women using multiple cross-sectional analyses. We documented the accuracy of HRP-2 and the duration of persistent antigenicity through a prospective longitudinal follow-up of 357 children treated for malaria.

Key results:

Health system capacity to incorporate diagnostics into routine health care delivery at Lower Level Health Care Facilities (LLHCFs)

We observed that LLHCFs had inadequate human and infrastructural capacity to effectively implement universal parasite-based malaria diagnosis. This was available at only 24 % of PHC facilities. Three-months’-long stock-outs of oral and parental quinine were reported at 48% of the health facilities. Up to 80 % of the approved staff positions were vacant and the majority (66 %) of the available staff were unskilled health care providers. Capacity building needs identified include: training of health workers in fever management, recruitment of qualified staff, strengthening the referral mechanisms, and development of supply chain and quality control systems. This study provides a bench mark for health systems strengthening activities against which successful implementation of parasite-based diagnosis of malaria can be tracked.

Feasibility and impact of RDTs on health workers' prescription behaviour

Using a cluster randomized design, we observed that RDT use reduced antimalarial drug prescriptions by 40% (RR: 0.62, 95 % CI 0.55-0.70). The highest reduction was in a hypo-endemic transmission setting (RR 0.46, 95% CI 0.51-0.53). The reduction was in part a result of training health workers and providing RDTs. On average, health workers offered a test to 90% of all malaria suspected out-patients across all transmission setting. However, 30% of the RDT-negative fever patients still received anti-malarial drugs and 50% for children under-five years old. One-day training on use of RDTs successfully delivered adequate skills and competences among health workers to perform RDTs in fever case management at LLHCFs. The average cost per health care worker for one-day training was USD 101 (range 92-112) and the main driver of cost was trainees’ travel and per-diems. For the large scale implementation of RDTs, there is a need to explore the modalities, effectiveness and cost-effectiveness of a cascade training model given the good peer to peer training observed at the health facilities. Our data demonstrate that RDT use is feasible at LLHCFs, and may lead to better targeting of malaria treatment. There is need to invest in advocacy, communication and social mobilisation (ACSM) to promote acceptance and use of RDTs.

Accuracy of RDTs in detecting malaria among vulnerable populations (children and pregnant women)

We investigated the accuracy and effectiveness of RDTs in detecting malaria during pregnancy and placental malaria infection at the time of delivery. We demonstrate that RDTs are accurate in detecting P.falciparum malaria during pregnancy and negative-RDT results are useful in excluding malaria infection among febrile pregnant mothers. We observed that HRP-2 RDTs are as good as microscopy in detecting placental malaria. Further, each of the methods used independently, under-estimated malaria in pregnancy compared to placental histology as the gold standard. A combination of RDTs and microscopy improved the sensitivity of diagnosing placental malaria, from 80.9% (RDTs alone) to 90.5%. Further research is required to identify peripheral markers of placental malaria infection. Therefore, preventive methods like intermittent presumptive treatment of malaria in pregnancy (IPTp) remain key since it is still difficult to correctly diagnose placental malaria in malaria hyper-endemic settings.

Operational accuracy of RDTs when used in the public health care delivery system

We observed that ICT Malaria Pf (ICT diagnostics, Cape town, South Africa) immuno-chromatographic test (ICT) had an acceptable accuracy (sensitivity of 98%, specificity of 72% compared to expert microscopy), suggesting that it is an appropriate test for operational use as a diagnostic tool where microscopy is not feasible. In a follow-up component we observed that, after successful treatment of malaria in children, persistent antigenicity had a mean duration of 32 days and it varied significantly depending on pre-treatment parasitaemia density. Therefore, good clinical evaluation is essential to interpret positive RDT results given the inability of RDTs to differentiate recent and new infections and response to antimalarial therapy in malaria-endemic settings.