Evaluation of standard diagnostic rapid test kits for malaria diagnosis among HIV patients in Kano, Nigeria

Background Malaria diagnosis among HIV-positive patients is uncommon in Nigeria despite the high burden of both diseases. Objectives We evaluated the performance of a malaria rapid diagnostic test (MRDT) against blood smear microscopy (BSM) among HIV-positive patients in relation to anti-retroviral treatment (ART) status, CD4+ count, fever, cotrimoxazole prophylaxis and malaria density count. Method A cross-sectional study involving 1521 consenting randomly selected HIV-positive adults attending two ART clinics in Kano, Nigeria, between June 2015 and May 2016. Venous blood samples were collected for testing with MRDT, BSM, and CD4+ T cells count by cytometry. Biodata and other clinical details were extracted from patient folders into an Excel file, cleaned, validated, and exported for analysis into SPSS version 23.0. Sensitivity, specificity, predictive values of MRDT were compared with BSM with a 95% confidence interval. Results Malaria parasites were detected in 25.4% of enrollees by BSM and 16.4% by MRDT. Overall sensitivity of MRDT was 58% and specificity was 97%. Cotrimoxazole prophylaxis and fever status did not affect MRDT sensitivity and specificity. Unexpectedly, the sensitivity was highest at parasite density count of less than 500 cells/µL. At CD4+ T cells count over 500 cells/µL the sensitivity was higher (62.4%) compared to 56% at less than 500 cells/µL. In the non-ART group sensitivity was higher (65%) compared to those on ART (56%) but the specificity was similar. All differences were significant for all variables (p < 0.05). Conclusion Although the MRDT specificity was good, the sensitivity was poor, requiring further evaluation for use in malaria diagnosis among HIV-malaria co-infected persons in these settings.


Introduction
Malaria remains a major public health problem in Nigeria with 97% of the population at risk. 1 In 2016 Nigeria had about 57.3 million cases of malaria with over 112 000 deaths reported. 2 Although malaria prevalence in Nigeria varied between states, a 60.6% prevalence has been reported in a population-based study across five local government areas of Kano State. 3 HIV prevalence among adults in Nigeria was about 2.9% with about 3.2 million persons living with HIV in 2016. 4 The greatest burden of disease due to both HIV and malaria (predominantly Plasmodium falciparum) occurs commonly in sub-Saharan Africa 5 and the rates of co-infection make interaction between both diseases a global health focus. 5,6,7 A key factor for effective management of malaria is early and accurate diagnosis. The global impact of malaria has prompted an increase in the development of diagnostic strategies. The updated World Health Organization (WHO) recommendations indicate that all suspected cases of malaria must be confirmed with a laboratory diagnostic test before treatment. 8 This enables differentiation of malarial and non-malarial fevers, to prevent unnecessary use of antimalarial drugs. 9,10 This is very crucial among HIV-positive patients, where febrile illnesses other than malaria are common. Opportunistic infections and other febrile illnesses mimic malaria in HIV patients. 5 the disease is more prevalent. Hence, the use of rapid diagnostic tests (RDT) in such endemic areas is recommended. 16,17 Furthermore, part of the national malaria control strategy plan for 2014 to 2020, with a vision of having a malaria-free Nigeria, is to test all care-seeking persons suspected of suffering from malaria using RDT or microscopy. 1 Data from reviewed literature showed that very few studies have evaluated the validity of RDT for malaria diagnosis in the general population in Nigeria. 14,15,18 Thus far, only one study has evaluated the performance of a malaria RDT (MRDT) among 387 HIV-positive persons in Nigeria. 19 Overall MRDT sensitivity, specificity, positive and negative predictive value have not yet been properly evaluated especially among HIV-positive individuals. 20,21 The objective of the study was to evaluate the performance of a commercially available MRDT (based on P. falciparum histidine-rich protein type-2 and Plasmodium lactate dehydrogenase), the Standard Diagnostics (SD) Bioline MRDT, compared to the gold standard, microscopy, in the diagnosis of malaria among HIV-positive participants with respect to malaria parasite density, CD4+ T cells count, cotrimoxazole prophylaxis use, anti-retroviral treatment (ART) status and occurrence of fever. The choice of SD Bioline was on the premise that this rapid test kit is the most commonly used MRDT for routine malaria diagnosis in Kano State.

Ethical considerations
The protocol was reviewed by Kano State Hospitals Management Board local ethical committee and written approval was given (study approval number: HMB/ GEN/488/1 of 17/04/2015). Participant consent was obtained from all respondents.

Study setting and population
This was a cross-sectional study carried out from June 2015 to May 2016 at two health facilities in Kano City, Nigeria, a malaria endemic area. 3 The facilities are Infectious Diseases Hospital Kano (IDH) and Murtala Mohammed Specialist Hospital Kano.
The population of interest in this study were HIV-positive adult participants who routinely visit the two health facilities to access ART services. They consisted of male and female (non-pregnant) patients (on ART and non-ART) aged 18 years and older enrolled in an ART programme who had not been on an antimalarial drug in the past 14 days and who willfully consented to participate in the study.

Data collection and sampling technique
A total of 1521 consenting participants were randomly selected and blood specimens were collected and analysed for the presence of malaria parasite using both RDT and microscopy techniques and density of malaria parasites was determined. Additionally, relevant socio-demographic and clinical data including clients' code, age, gender, use of cotrimoxazole prophylaxis, ART status (receiving or not receiving treatment), and episodes of fever were retrieved from both patients' folders and Lafiya Management Information System (LAMIS ® ) software with the support of trained data entry clerks. We ensured data confidentiality through participant coding and restricting access to the data to only research team members.

Laboratory analysis (rapid diagnostic test, blood microscopy and parasite density determination)
From 1521 consenting HIV-positive participants, 4 mL venous blood sample was aseptically collected using sterile vacutainer needle and holder into an ethylenediaminetetraacetic acid (EDTA) anticoagulant tube. The blood samples were mixed properly to avoid clotting before laboratory investigations.
We used the MRDT kit (Code 05FK30) as described by the manufacturers (Standard Diagnostics Bioline Korea, 2013).
For microscopy slide method, both thin and thick smear slides were prepared and examined. The microscopy procedure adopted was described by WHO. 23 To ensure the quality of testing, MRTD and microscopy were performed by independent laboratory personnel and MRTD results were blinded to microscopists. All MRTD kits were stored according to manufacturer recommendation and used before the expiry date. The two microscopists responsible for slide examinations in both facilities have attended WHO training on malaria microscopy. Additionally, 10% of the slides were randomly selected for retesting. About 2% of the results were discordant between the first and second readers and were subjected to further evaluation by a third reader before a final decision was made.
Malaria parasite density for each positive smear was calculated using individual white blood cell count according to this formula: malaria parasite count = Number of parasite count × patient actual white blood cell count/μL/Number of white blood cell count 200 or 500.

Data processing and analysis
Data were reviewed, cleaned and validated in a Microsoft 2013 Excel file (Microsoft Corp., Redmond, Washington, United States) and analysed using Statistical Package for Social Scientists (SPSS version 23.0; IBM, Armonk, New York, United States). The sensitivity, specificity, positive predictive value and negative predictive value of MRDT were estimated against blood smear microscopy (BSM), the gold standard, at 95% confidence interval in relation to (1) malaria parasite density count, (2) fever (presence or absence), (3) CD4+ T cells count, (4) co-trimoxazole prophylaxis use, and (5) ART status (currently on ART or not taking ART). In addition, differences in performance of MRDT against BSM were analysed (overall and for sub-groups of the five variables mentioned above) based on level of 'equality of marginal positives' in disease classification between the two using McNemar tests and significance level was set at p < 0.05.

Study population
This study included a total of 1521 HIV-positive participants consisting of 1074 (70.6%) women and 447 (29.4%) men. The mean age was 37.20 with a standard deviation (SD) of 10.41 years.
Data on clinical information showed that the majority (84.7%) were on ART. An almost equal proportion of the participants (n = 772, 50.8%) were on daily dose co-trimoxazole prophylaxis for prevention against opportunistic infection including malaria parasite infection while 749 (49.2%) were not on co-trimoxazole prophylaxis. About (7%) complained of fever. The mean current CD4+T cells count of the participants was 401.35 cells/µL ± 239.24 cells/µL, and ranged from 6 cells/µL to 1736 cells/µL. The majority (n = 1038, 68.2%) of the respondents had a CD4+T cells count lower than 500 cells/µL, while 31.8% (n = 483) had CD4+ T cell counts of 500 cells/µL or higher. The mean malaria parasite density was 265 cells/µL ± 31.8 (SD) cells/µL with a range of 20 cells/µL to 2500 cells/µL. A majority (87.8%) had malaria parasite density less than 500 cells/µL, while the rest (12.2%) had parasite density 500 cells/µL or higher.

Performance of malaria rapid diagnostic test in comparison with standard microscopy Different malaria parasite densities
The MRDT showed a sensitivity of about 72% among HIVpositive patients with malaria density count between 0 and 499 cells/µL and greater than 1000 cells/µL. Unexpectedly, a low sensitivity of about 34% was observed when density was between 500 and 999 cells/µL (Table 3).
MRDT performance was significantly different from BSM findings, regardless of malaria parasite density count of participants (p < 0.005).

Presence or absence of fever
The MRDT showed slightly lower but similar sensitivity in cases with and without fever, but similar specificity for both groups of participants. However, positive predictive value was higher (95.6%) among clients that complained of fever compared to those without (88.1%) fever (Table 3). In those with or without fever, MRDT performance was significantly lower compared to BSM findings (p < 0.0)

Co-trimoxazole prophylaxis
MRDT performed similarly in sensitivity and specificity in both groups, although higher positive predictive value was observed among participants who were not on co-trimoxazole ( Table 3). Regardless of participant's co-trimoxazole status, the performance of MRDT was significantly lower in  sensitivity and specificity compared to BSM findings (p < 0.001) in both groups.
The observed differences in performance were statistically significant (p < 0.001).

Anti-retroviral treatment status
The MRDT revealed a higher (65%) sensitivity among HIV participants who were not on ART compared (56%) to those on ART. A similar pattern of higher positive predictive value (98%) was observed among the non-ART group compared to 87% among the ART group (Table 3). For both groups of participants (ART vs non-ART), the observed differences in performance of MRDT were significantly

Discussion
We detected malaria parasites in 25.4% of participants by BSM and 16.4% by MRDT. Overall sensitivity of MRDT was 58% and specificity was 97%. The sensitivity and specificity were similar irrespective of co-trimoxazole and fever status. At malaria parasite density count of under 500 cells/µL, sensitivity was 73% and between 500 and 999 cells/µL, sensitivity was 34%. At CD4+ T cell count over 500 cells/µL the sensitivity was higher (62.4%) compared to 56% at under 500 cells/µL. In the non-ART group sensitivity was higher (65%) compared to those on ART (56%) but the specificity was similar.
Although blood smear microcopy still remains the gold standard for malaria diagnosis, several limitations ranging from infrastructure and equipment to competence still exist in endemic areas like Nigeria. These justify the need for an alternate method like MRDT. 15 Since WHO recommended the use of MRDT, several test kits have flooded the market, hence the need for constant field validation to verify manufacturers' claims. This is critical, particularly among HIV-positive patients in whom other febrile illnesses and infections mimic malaria. 5 The prevalence of malaria in HIV-positive patients in this study was 25.4% based on microscopy compared to 16 21 It is worth noting that in this study, participants were recruited at routine follow-up visits, regardless of suspicion of malaria unlike the other studies comparing performance of BSM and MRDT in HIVpositive patients suspected of suffering from or having symptoms of malaria. For example, the Tanzania 24 and Burkina Faso 21 studies looked at HIV-positive children suspected of malaria, while those from Lagos, 19 Uganda, 10,20 and Malawi 25 looked at adults with HIV attending outpatient clinics suspected of having malaria. Thus, the direct comparison of proportions positive for malaria in our studies with the others must be considered with caution.
MRDT overall sensitivity and specificity reported in this study was similar to a 55.4% sensitivity reported in Lagos, Nigeria, but varied in specificity (90.3%). 19 Contrary to our results, a 100% sensitivity and similar specificity of about 95.4% was documented in a Burkina Faso study among HIVpositive participants. 21 Other studies on MRDT performance among HIV-positive population in Malawi, 25 and two different studies in Uganda 10,20 have shown higher sensitivity of over 94%. In these studies, the specificities averaged over 97% like in ours except for the Malawi study that reported about 51% specificity. 25 The 2.5% false positive results observed with the use of MRDT may be an indicator of residual antigenemia. The average time that histidine rich protein-2 (pFHRP-2) remains positive after resolution of parasitaemia is about 2 weeks, although it has been shown that the protein can take over a month to clear. 26 Using MRDT, a false positivity rate of 5.9% had been reported in Burkina Faso among HIVpositive patients.
The 27.55% false negative results observed might be partly due to deletion or mutation of HRP2 gene in the malaria parasite, which is the most common target for MRDT. 27,28 It has been reported to exhibit a high level of polymorphism. 29,30 This is an important factor that may affect the performance of MRDT based on antigen detection. 31,32 In addition, false negative results have been associated with blood samples taken beyond the period of malaria fever paroxysms. 33 P. falciparum isolates without the HRP2 gene have been shown to be important contributors to false negative HRP2-based MRDT testing. 34,35 Moreover, only Plasmodium falciparum releases HRP2; therefore, the presence of non-falciparum infection as another malaria species co-infection in Plasmodium falciparum may also give a false negative result. Other factors that might possibly affect MRDT field performance are shelf life and heat stability. 36,37 In this study, the test kits were stored based on manufacturers' recommendations and used before the expiry date on the MRDT kits.
Unexpectedly, with the MRDT, we observed a lower sensitivity of 34% among HIV-positive participants with a malaria density count between 500 cells/µL and 999 cells/µL compared to a 73% sensitivity when the malaria density count was less than 500 cells/µL. However, the specificity at between 500 cells/µL and 999 cells/µL was 97.5%. This was contrary to a previous study in Lagos, Nigeria, that observed an increase in sensitivity from 90.9% at parasite densities greater than 200 cells/µL to 97.6% at densities greater than 500 cells/µL. 19 Generally, at lower parasitaemia, variability in sensitivity is more common. 20,38,39 Similar observation of false negative MRDT results at parasite counts even higher than 1000 cells/µL have been reported in other studies 19,25,40,41 and may be attributed to prozone effect observed with immunochromatographic tests such as MRDT. 42 All the same, many MRDTs today can achieve excellent sensitivity and specificity for P. falciparum at parasite density lower than 500 cells/µL. 11 Besides this study, and to the best of our knowledge, only one other study has assessed performance of MRDT among HIVpositive patients based on febrile status in Africa. 10 MRDT performance in our study showed similar sensitivity and specificity as reported by Mills et al., 10 in febrile or non-febrile http://www.ajlmonline.org Open Access cases in rural Uganda. Apparently, MRDT reliability is independent of febrile status.
MRDT performance was similar in terms of sensitivity, specificity and positive predictive value regardless of whether a participant was on co-trimoxazole prophylaxis or not. Information on similar studies is not available; however, with microscopy a relatively higher prevalence has been reported on those without co-trimoxazole prophylaxis compared with those on it. 19,22,43 MRDT performance revealed a higher sensitivity among participants with a CD4+ T cell count greater than 500 cells/ µL compared to a lower sensitivity among participants with a CD4+ T cell count under 500 cells/µL. Although information on similar studies is not available, it has been reported in a similar setting that HIV-positive participants with CD4+ T cell counts 250 cells/µL or higher are significantly less likely to have patent parasitaemia. 19 The MRDT revealed a higher sensitivity among HIV-positive participants who were non-ART compared to those on ART. Lower incidence of malaria among HIV-positive adults on ART have been reported in different settings 10,43,44 as well as reports of antimalarial properties of some antiretroviral drugs. 45,46 Based on other reports in the general population, the use of MRDT method as an alternative to BSM in malaria endemic areas is recommended for epidemiological studies. However, the performance varied depending on species of the malaria parasite, level of parasitaemia, and immunity. 47 For some of these reasons, such recommendation must be taken with caution in HIV-malaria co-infected persons, even though no differential effect of HIV infection on MRDT performance (specifically the SD Bioline antigen-based method) has been reported in literature. 25

Strengths and limitations
This study was sufficiently powered having a large sample size of 1521 HIV-positive participants across the two health facilities. Apparently, this may be the first study to assess the performance of MRDT in an African setting of significantly high HIV and malaria co-morbidity in association with variables like immune status, parasitaemia and some clinical (co-trimoxazole prophylaxis, fever and ART status) factors.
The cross-sectional design of this study was a limitation as it was difficult to determine how the potential antimalarial effect of co-trimoxazole prophylaxis and ART could have affected MRDT performance over time. Note that we observed better diagnostic accuracy of MRDT among participants who were neither on co-trimoxazole prophylaxis nor on ART. Secondly, we did not measure individual body temperature to correlate it with participants' reports of fever.

Conclusion
As observed in this study, the sensitivity of SD Bioline, which is based on P. falciparum histidine-rich protein type-2 and Plasmodium lactate dehydrogenase, is lower than WHO recommendation for MRDT. 48 Thus, further evaluation is required to determine its suitability in malaria diagnosis among HIV-malaria co-infected patients in these settings.
It is worth speculating that a similar observation with the MRDT could have been observed in non-HIV-positive persons.

Trustworthiness, Reliability and Validity
The data reported in this current article reflect findings of work done on evaluation of standard diagnostic Bioline rapid test kits for malaria diagnosis among HIV patients in Kano, Nigeria, by our research team members who all participated in the study design, execution, collection, analysis of data and report writing.
The research outcomes presented in this report were from the diagnostic evaluation of standard diagnostic Bioline rapid test kits for malaria diagnosis among HIV-positive participants in Kano, Nigeria, at Infectious Diseases Hospital and Murtala Mohammed Specialist Hospital Kano. Although previous studies have reported on the evaluation of rapid malaria diagnosis against gold standard microscopy methods with a focus on the general population, it is worth noting that data among HIVpositive persons under different conditions are not readily available. Hence, this study sought to address paucity of data on performance evaluation of standard diagnostic Bioline among HIV-positive patients in relation to various factors such as fever status, CD4+ cell count, ART status, co-trimoxazole prophylaxis and malaria density.
The outcomes of this study and proposed recommendations could be useful in public health programme planning in such settings where HIV and malaria burden is high. Apart from disease prevalence data reported, malaria density count and performance of rapid test kits was also evaluated in addition to other clinical and useful variables. This may serve as a baseline upon which comparison may be made in future.