Diagnostic performance of rapid diagnostic tests for the diagnosis of malaria at public health facilities in north-west Ethiopia
Abstract
enObjective
To assess the performance of RDTs against nested polymerase chain reaction (nPCR) for the diagnosis of malaria in public health facilities in north-western Ethiopia.
Methods
Cross-sectional study at public health facilities in North Gondar, Ethiopia, of 359 febrile patients with signs and symptoms consistent with malaria. Finger prick blood samples were collected for testing in a P. falciparum/pan-malaria RDTs and for molecular analysis. Sensitivity, specificity and predictive values were determined for the RDTs using nPCR as reference diagnostic method. Kappa value was determined to demonstrate the consistency of the results between the diagnostic tools.
Results
By RDTs, 22.28% (80/359) of patients tested positive for malaria, and by nPCR, 27.02% (97/359) did. In nPCR, 1.67% (6/359) and 0.28% (1/359) samples were positive for P. ovale and P. malariae, which had almost all tested negative in the RDTs. The sensitivity, specificity, positive and negative predictive values of RDTs for the diagnosis of malaria were 62.9%, 92.7%, 76.3% and 87.1%, respectively, with 0.589 measurement agreement between RDTs and nPCR. The sensitivity and specificity of RDTs for P. falciparum identification only were 70.8% and 95.2%, and 65.2% and 93.1% for P. vivax.
Conclusion
Although RDTs are commonly used at health posts in resource-limited environments, their sensitivity and specificity for the detection and species identification of Plasmodium parasites were poor compared to nPCR, suggesting caution in interpreting RDTs results. Particularly, in the light of expanded efforts to eliminate malaria in the country, more sensitive diagnostic procedures will be needed.
Abstract
frObjectif
Evaluer la performance des TDR par rapport à la réaction en chaîne de la polymérase imbriquée (nPCR) pour le diagnostic du paludisme dans les établissements de santé publique dans le nord-ouest de l'Ethiopie.
Méthodes
Etude transversale dans les établissements de santé publique dans le Nord Gondar, en Ethiopie, sur 359 patients fébriles avec des signes et des symptômes compatibles avec le paludisme. Des échantillons sanguins ont été prélevés par piqûre au doigt pour être analysés par des TDR pour P. falciparum/paludisme en général et pour l'analyse moléculaire. La sensibilité, la spécificité et les valeurs prédictives ont été déterminées pour les TDR en utilisant la nPCR comme méthode de diagnostic de référence. La valeur Kappa a été déterminée pour démontrer la concordance des résultats entre les outils de diagnostic.
Résultats
Par les TDR, 22,28% (80/359) des patients étaient positifs et par la nPCR, 27,02% (97/359) pour le paludisme. Par la nPCR, 1,67% (6/359) et 0,28% (1/359) des échantillons étaient positifs pour P. ovale et P. malariae; presque tous étaient négatifs par les TDR. La sensibilité, la spécificité, les valeurs prédictives positives et négatives des TDR pour le diagnostic du paludisme étaient de 62,9%, 92,7%, 76,3% et 87,1%, respectivement, avec 0,589 de concordance de résultats entre les TDR et la nPCR. La sensibilité et la spécificité des TDR pour l'identification de P. falciparum seulement étaient de 70,8% et 95,2% et pour P. vivax, 65,2% et 93,1%.
Conclusion
Bien que les TDR soient couramment utilisés dans les postes de santé dans les zones à ressources limitées, leur sensibilité et spécificité pour la détection et l'identification des espèces de parasites du genre Plasmodium sont faibles par rapport à la nPCR, suggérant la prudence dans l'interprétation des résultats des TDR. En particulier, au vu des efforts accrus pour éliminer le paludisme dans le pays, des procédures diagnostiques plus sensibles sont nécessaires.
Abstract
esObjetivo
Evaluar el desempeño de PDRs frente a la reacción de la polimerasa anidada (PCRa) para el diagnóstico de la malaria en centros sanitarios públicos del noroeste de Etiopía.
Métodos
Estudio croseccional en centros sanitarios públicos en Semien Gondar, Etiopía, de 359 pacientes febriles con signos y síntomas consistentes con la malaria. Se recogieron muestras de sangre mediante punción digital para realizar análisis mediante PDRs para P. falciparum/pan-malaria RDTs y análisis molecular. Se determinaron la sensibilidad, especificidad y valores predictivos de las PDRs utilizando la PCRa como método diagnóstico de referencia. Se determinó el valor Kappa para demostrar la consistencia de resultados entre las herramientas diagnósticas.
Resultados
Mediante las PDRs un 22.28% (80/359) de los pacientes dieron positivo para malaria, mediante PCRa un 27.02% (97/359). Mediante PCRs, 1.67% (6/359) y 0.28% (1/359) de las muestras eran positivas para P. ovale y P. malariae, la mayoría de las cuales habían dado negativo en las PDRs. La sensibilidad, especificidad, valores predictivos positivos y negativos de las PDRs para el diagnóstico de malaria eran 62.9%, 92.7%, 76.3% y 87.1%, respectivamente con una concordancia de 0.589 entre PDRs y PCRa. La sensibilidad y especificidad de las PDRs para la identificación de P. falciparum solo era del 70.8% y 95.2%, y del 65.2% y 93.1% para P. vivax.
Conclusión
Aunque las PDRs son comúnmente utilizadas en centros de salud de lugares con pocos recursos, su sensibilidad y especificidad para la detección y la identificación de especies de parásitos de Plasmodium era pobre comparada con la PCRa, lo cual sugiere la necesidad de ser cautelosos a la hora de interpretar los resultados de las PDRs. En particular, a la luz de expandir esfuerzos para eliminar la malaria del país, serían necesarias procedimientos diagnósticos más sensibles.
Background
Although malaria caused by Plasmodium falciparum remains the predominant species in Ethiopia, non-falciparum malaria is gaining importance. Close to one-third of malaria cases in Ethiopia are caused by P. vivax. P. ovale and P. malariae have only recently been identified as contributing a significant proportion of malaria cases 1, 2.
Malaria threatens the lives of 40% of the world's population 3. Each year, there are estimated 350–500 million clinical cases, with about 90% of these occurring in sub-Saharan Africa 3-5. Malaria is a leading communicable disease in Ethiopia: an estimated 57.3 million (68%) of the 84.3 million population live in areas at risk of malaria. The Federal Ministry of Health (FMOH) estimates that there are 5–10 million clinical malaria cases each year. In 2009/2010 health and health indicators report, 1581 malaria deaths were reported in Ethiopia 6.
In Ethiopia, the main species P. falciparum and P. vivax account for 60% and 40% of malaria cases, respectively 7, but only P. falciparum commonly causes severe disease in which the case fatality rate is about 10% in hospitalised adults and up to 33% in children under 12 years 8. In addition to the health impact, malaria has a significant impediment to the socio-economic development of the country. Fertile lowlands and major river valleys have not been populated and developed largely due to the high malaria burden in these areas.
In Ethiopia, early diagnosis and prompt treatment are the main strategies in malaria prevention and control and to reduce morbidity and prevent mortality 9. However, clinical diagnosis and empirical treatment have been the mainstay of malaria management in areas where laboratory facilities are not available. However, clinical diagnosis is notoriously unreliable and leads to overdiagnosis and overtreatment 10.
Although frequently not accessible in most peripheral health facilities in the country, Giemsa microscopy remains the gold standard for the laboratory diagnosis of malaria. However, it is time-consuming, requires trained personnel and needs careful preparation and application of reagents 11. As a result, rapid diagnostic tests (RDTs) have been introduced for the diagnosis of malaria in remote areas where there are no microscopy facilities. However, poor and varying performances of RDTs have been reported in Ethiopia 12. Therefore, this study aimed to assess the performance of RDTs as compared to nested polymerase chain reaction (nPCR), the most sensitive reference test, for the diagnosis of malaria at public health facilities in North Gondar, Amhara Regional State, Ethiopia.
Materials and methods
Setting and design
The study was conducted at Health Posts under Lamba Health Center, Maksegnet Health Center and Kola Diba Health Center in North Gondar, Amhara Regional State, Ethiopia. The altitude at these sites ranges from 1750 to 2100 m above sea level. According to the Municipal Health Bureau report, malaria is the most prevalent seasonal disease in these areas.
A health centre-based cross-sectional study was conducted from February to September 2014. Self-reporting febrile patients, in the age ranging from 1 year to 81 years, with signs and symptoms consistent with malaria attending either of the health posts were invited to participate. Patients who had received antimalarial drugs during the past 4 weeks prior to enrolment and patients unwilling to give consent and/or a blood sample were excluded from the study.
Study participants were recruited consecutively until a total of 359 patients with signs and symptoms consistent with malaria had been reached. Finger prick blood samples were collected from every participant and used to carry out RDT (CareStart™ Pf/Pan) analysis. CareStart™ Pf/Pan is targeted to detect P. falciparum histidine-rich protein-2 and plasmodial lactate dehydrogenase for the diagnosis of P. falciparum and other Plasmodium species, respectively.
Laboratory procedures
Two drops of blood from each participant were collected and transferred on filter paper (Whatman #903, GE Healthcare) labelled with the participant's study code and date. Each filter paper was dried individually to avoid any chance of contamination. The samples were then stored in small sealable plastic bags with desiccant and underwent molecular analysis at the laboratories of the Medical University of Vienna (MUV), Vienna, Austria.
A modified Chelex-based DNA extraction method using the InstaGene Whole Blood Kit (Bio-Rad Laboratories, Hercules, CA, USA) was used for the extraction and purification of Plasmodium DNA from the blood spots on filter paper. Parasite detection and species classification by nested PCR assay were performed for all samples as described previously 13, 14. The individual interpreting the nPCR results was blinded to the results of RDTs.
Data analysis
Sensitivity, specificity and predictive values were determined using SISA online statistical software 15. The kappa coefficient (Cohen's kappa coefficient as a measure of agreement for qualitative items) was determined to confirm the consistency of the results between the diagnostic tools.
Ethics
The study protocol was reviewed and approved by the Institutional Review Board of University of Gondar prior to the study. Informed consent and assent were obtained from all participants and/or their legal representatives after being translated and read in the vernacular language that the patients or the caretakers understood. Patients who tested positive for malaria by RDTs received immediate treatment according to the Ethiopian national treatment guidelines.
Results
Comparison of RDTs and nPCRs
Among the study participants, 22.28% (80/359) patients tested positive for malaria by RDTs of whom 12.53% (45/359), 5.85% (21/359) and 3.9% (12/359) were diagnosed as P. falciparum, P. vivax and mixed infection of P. falciparum and P. vivax, respectively. By nPCR, 27.02% (97/359) patients were malaria positive. Among RDT-negative samples, 12.9% (36/279) samples turned malaria positive by nPCR. In nPCR, the rate of mixed infection was 4.46% (16/359). The rates of P. ovalae and P. malariae infections were 1.67% (6/359) and 0.28% (1/359), respectively. Only one case of P. ovalae was reported as mixed infection of P. falciparum and P. vivax, while none of the P. ovalae and P. malariae infections were identified by RDTs (Table 1).
| RDT (n) | nPCR (n) |
|---|---|
| P. falciparum (45) | P. falciparum (34); negative (11) |
| P. vivax (21) | P. falciparum (2); P. vivax (10); positive of genus Plasmodium (1); negative (8) |
| Mixed P. falciparum and P. vivax (14) | P. falciparum (8); P. vivax (1); mixed infection of P. falciparum and vivax (4); P. ovalae (1) |
| Negative (279) | P. falciparum (17); P. vivax (6); P. ovalae (5); mixed infection of P. falciparum, vivax and malariae (1); mixed infection of P. falciparum and vivax (1); positive for genus Plasmodium (6); negative (243) |
- n, number of blood samples tested; nPCR, nested polymerase chain reaction; RDTs, rapid diagnostic malaria tests.
Diagnostic accuracy of RDTs compared to nested PCR for the diagnosis of malaria
Malaria rapid diagnostic tests had a sensitivity and specificity of 62.9% (95% CI: 50.6–75.1) and 92.7% (95% CI: 88.7–96.8), respectively, compared to the reference method, nPCR. Its corresponding positive and negative predictive values were 76.3% (95% CI: 64.4–88.1) and 87.1 (95% CI: 82.1–92.1). Overall, there was a moderate measurement agreement of test results between RDTs and nPCR (Kappa = 0.589).
RDTs had a sensitivity and specificity of 70.8% (95% CI: 56.7–84.9) and 95.2% (95% CI: 92.0–98.5) for the diagnosis and identification of P. falciparum, respectively. The corresponding sensitivity and specificity for the diagnosis and identification of P. vivax were 65.2% (95% CI: 40.4–90.0) and 93.1% (95% CI: 89.4–96.8), respectively.
Discussion
Early, prompt and particularly accurate malaria diagnosis down to species level even in remote areas is needed to provide prompt treatment and optimal case management of malaria 16, 17. Standard diagnostic methods, such as Giemsa microscopy and RDTs, have become commonly available at most health facilities, even in resource-limited environments. However, both techniques have inherent limitations and show relatively poor performance with low parasite densities 11, 18-20. The sensitivity of most RDTs used in routine diagnosis is strongly dependent on the parasite density and on the parasite species. The threshold of parasite density reliability giving positive results is also largely dependent on environmental factors, the brand of the RDTs, the targeted antigens, and a number of host factors. In previous publications the detection limits of RDTs has therefore been estimated at up to 200 parasites/μl of whole blood 21. Non-falciparum malaria remains a major challenge for most RDTs with sensitivities reaching only non-satisfactory levels. In addition, many RDTs are not designed to adequately identify parasite species other than P. falciparum.
The combination of antibodies detecting P. falciparum antigens with antibodies detecting so called pan-malaria antigens is a common, economical, and to a certain extent highly valuable approach to designing RDTs. However, this does not allow for the identification of mixed infections and leaves non-falciparum malaria notoriously underdetected.
While non-falciparum malaria (mostly P. ovale and P. malariae) is not very common in most African countries, Ethiopia reports one of the highest P. vivax burdens in the world. P. malariae infections, however, seem to be uncommon in the region. In fact this is the first report of a confirmed P. malariae infection in this part of Ethiopia. There also seems to be a clear trend towards a relatively higher prevalence of non-falciparum due to a strong focus on falciparum malaria in the current malaria control programme, as seen previously, for example in South-East Asia 22-24. Nevertheless, the elimination of P. falciparum is a goal in itself, and would be a major achievement even if that of the other species would take much longer. In this setting, the accurate diagnosis of non-falciparum malaria, preferably down to species level is therefore of utmost importance. This particularly applies to elimination settings in which non-falciparum malaria is likely to play an important role due to the inherent challenges associated with eliminating relapsing malaria.
The present study adds relevant information regarding the performance of currently used malaria RDTs in the study area. RDTs showed a sensitivity of only 62.9% and specificity of 92.7% regardless of Plasmodium species when using nPCR as a reference method. In practice, this means that more than 10% of all patient samples would have been missed as being malaria positive. The utilisation of Plasmodium genus and species-specific markers makes nested PCR an ideal confirmatory test for malaria diagnosis as it also allows for the detection of low density infections and even more importantly of mixed infections, which could not be identified by the RDTs used in this study 13.
A similarly poor performance of malaria RDTs has previously been reported from Gabon, Nigeria and China–Myanmar 25-27. A study from Zambia reported that community health workers read faint positive and invalid results of RDTs as negative suggesting that subjective interpretation may contribute to poor performance 28. An earlier study from Angola suggested that performance may also be related to the age of the patients with malaria 29. In the current study, 57.4% of the study participants were 18 years and older. Accordingly, the sensitivity of RDTs in study participants under 18 years (72.9%) was higher than older ages (50.0%). The fact that previous studies in Ethiopia and Burkina Faso reported a relatively good performance of malaria RDTs compared to Giemsa microscopy 30, 31 may have been influenced by the inherent limitations of Giemsa microscopy rather than good performance of RDTs 22.
In the current study, the RDTs consistently showed better performance in P. falciparum than in P. vivax (sensitivity 70.8% vs. 65.2% and specificity 95.2% vs. 93.1%). Similar results have been reported previously, for example from Belgium and the China–Myanmar border area. Throughout most studies the key factors contributing to the performance of RDTs are parasite density and Plasmodium species.
A factor strongly influencing the performance of HRP2-based RDTs in certain regions of the world seems to be the genetic variability of the PfHRP2 gene which results in P. falciparum infections not being detected in spite of relatively high parasite densities 32. This has previously been reported from a number of countries in Latin America 33, 34. In how far this could influence RDTs results in Ethiopia remains to be seen.
The relatively poor performance of RDTs in our study underlines the urgent need for new diagnostic tests that are highly sensitive and specific but also economical and field deployable. Nested PCR, which was used as the reference test in this study, may provide adequate performance but is clearly not suited for use in resource-limited environments. Loop-mediated isothermal amplification (LAMP), a relatively recent development that considerably simplifies molecular analysis, may be a major step towards making molecular techniques available in such settings. In the meantime, the performance of RDTs could profit from improved training for healthcare workers and from optimising integrated diagnostic approaches with the ultimate goal of supporting malaria elimination programmes in Ethiopia 35-37.
One limitation of this study is that malaria microscopy was not performed and parasite density therefore not reported.
Acknowledgements
The authors' are very grateful to Institute of Specific Prophylaxis and Tropical Medicine, Medical University of Vienna for funding this study. We would also thank Health Posts' staff and the study participants for their cooperation in providing the necessary laboratory sample.
