Target condition being diagnosed
Malaria is a life-threatening illness, caused by the asexual form of the parasitic protozoan Plasmodium. Most cases of malaria are uncomplicated, commonly presenting with fever and sometimes with other non-specific symptoms including headache, and aches and pains elsewhere in the body (Gilles 1991; WHO 2003). A few people develop severe malaria, with confusion, weakness, coma and other life-threatening complications. Malaria is curable, and early, prompt and accurate diagnosis followed by appropriate treatment helps to reduce illness and death, (WHO 2003) and is central to current malaria control policy (Bell 2006; WHO 2005).
The two most common species of malaria parasite are Plasmodium falciparum and Plasmodium vivax. P. falciparum malaria is by far the most common type of malaria in Africa, and is also endemic in parts of Asia and South America. It is the most common cause of severe malaria, is responsible for almost all malaria deaths, and can cause other complications such as anaemia and, in pregnancy, low birth-weight babies. Vivax malaria is a relapsing form, which is rarely fatal but can cause serious anaemia in children. Less common human malaria parasite species include P. malariae and P. ovale.
In 2008, there were between 190 million and 311 million cases of malaria worldwide (WHO 2009a). Around 85% of these cases were in Africa; 10% were in South East Asia; 4% were in the Eastern Mediterranean region; and 1% were in South America (WHO 2009a). In the same year, there were between 708,000 and 1,003,000 deaths from malaria; 89% were in Africa and 85% were children under the age of five years (WHO 2009a).
People who are repeatedly exposed to malaria infection develop a partial and incomplete immunity. In highly endemic areas, those most at risk are children under the age of five, who have not yet had the chance to develop immunity. In less endemic areas, or areas of seasonal or epidemic transmission, older children and adults are also at risk due to less developed immunity. Travellers from non-endemic to endemic countries are at highest risk because they have no immunity at all.
Rapid diagnostic tests (RDTs) (WHO 2003) detect parasite-specific antigens in a drop of fresh blood through lateral flow immunochromatography (WHO 2006). The World Health Organization (WHO) currently lists 96 commercially-available test kits meeting ISO13485:2003 manufacturing standards (WHO 2009). RDTs do not require a laboratory or any special equipment (WHO 2006); they are simple to use and can give results as a simple positive/ negative result, within 15 minutes (Talman 2007). RDTs are therefore, in general, suitable for remote areas with limited facilities and relatively untrained staff. However, they have a limited shelf life, and need to be kept dry and away from extremes of temperature. They may also fail to detect malaria in cases where there are low levels of parasites in the blood, and false positives are possible due to cross reactions or gametocytaemia (infection with the sexual stage of the parasite only) (Kakkilaya 2003).
RDTs use antibodies to detect one or several antigens. The most commonly used antibodies react to histidine-rich protein-2 (HRP-2), aldolase and plasmodium lactate dehydrogenase (pLDH) (Talman 2007). HRP-2 is a marker for P. falciparum, while pLDH antibodies can be specific for P. falciparum, or P. vivax, or may detect all species (including P. ovale and P. gambiae) or other combinations of these species. Aldolase antibodies are pan-specific, detecting all types of malaria parasite but not differentiating between them. Until recently, there were seven main types of commercially-available test, using different antigen combinations as described in Table 1 below (Bell 2006).
Since this classification was developed, the following test types have also become available.
- Pan pLDH antibodies only, with the following possible results: no malaria; P. falciparum, P. vivax, P. ovale and/or P. malariae; and invalid (as for Type 7 tests).
- P. vivax-specific pLDH antibodies only.
- pLDH antibody lines detecting P. vivax, P. ovale and P. malariae in combination.
The different test types detect different malaria species and combinations of species; the choice of RDT used will therefore depend on which species are endemic in the area. Table 2 shows the type of tests that are appropriate for use in the different malaria 'zones' of the world.
HRP-2 can stay in the blood for up to 28 days after starting antimalarial therapy (Kakkilaya 2003). Because of this 'persistent antigenaemia', it is not possible to use these tests for assessing parasite clearance following treatment, and false positive results may be found in patients who have recently been treated for malaria. In contrast, pLDH is rapidly cleared from the blood following parasite death; in fact, it may clear more rapidly than the dead parasites (WHO 2009), but may persist in the presence of gametocytes.
Microscopic examination of Giemsa-stained thick and thin blood films remains the conventional laboratory method for malaria diagnosis, but needs to be conducted by microscopists with adequate training and equipment. Microscopic examination displays a good sensitivity and specificity, and allows species and stage differentiations and quantification of parasites, all of which are important in assessing the disease severity and prescribing appropriate therapy. Intensive examination is more likely to reveal parasitaemia, so the test is carried out by examining a fixed number of fields; infections may be missed if slides are not examined carefully (Wongsrichalanai 2007). Very low parasitaemia may be missed even by good quality microscopy; the limit of detection of thick smear microscopy has been estimated at between around four and 20 asexual parasites per μl, although under field conditions a threshold of between 50 and 100 asexual parasites per μl is more realistic (Wongsrichalanai 2007). On the whole, false positive results are the result of poor slide preparation or reading (Wongsrichalanai 2007).
Molecular DNA amplification via polymerase chain reaction (PCR) is the most accurate method of detecting parasites in the blood. It eliminates observer error and is more sensitive at low levels of parasitaemia, with limits of detection as low as 0.004 asexual parasites per μl (Hanscheid 2002; Snounou 1993). However, whether this increased ability to detect low level parasitaemias makes it a better diagnostic test is uncertain, as sub-microscopic parasitaemias are of unknown clinical significance and the prevalence of asymptomatic sub-microscopic infection is high in some areas (May 1999). In addition, PCR may be prone to false positive results due to contamination of samples if laboratory standards are not sufficiently high. PCR is currently not widely available outside of research settings, as it needs specially-trained technicians and a well-equipped laboratory.
Diagnostic tests for malaria in endemic areas are now recommended as routine by the WHO in all patients suspected of malaria before any treatment begins (WHO 2010). This is due to a shift in drug treatment policy away from cheap, often relatively ineffective, drugs, towards artemisinin-based combination treatment (ACTs), which are highly effective, expensive, and need to be used properly to prevent resistance developing.
There is a long-standing recognition that good quality, standard malaria microscopy is relatively expensive and difficult to deliver in many basic, primary health care settings in developing countries, while RDTs for malaria have now become widely available and affordable.
RDTs in malaria could dramatically increase access to prompt diagnosis in primary health care. The question of how a package of care (diagnosis using RDTs with positive cases treated with drugs versus presumptive treatment of all cases with symptoms suggestive of malaria) impacts on health outcomes is to be addressed in a separate forthcoming review (Odaga 2011). However, important policy questions remain to be answered
a) How well do RDTs perform compared to the previous standard of microscopy in diagnosing symptomatic patients?
b) What are the differences in accuracy between different types of commercial test, and individual brands of commercial tests?
This information will help to inform choice, although factors, such as price, product consistency, stability and shelf life will also influence those decisions.
This review is the first of a series of three reviews: the second will examine the accuracy of RDTs for diagnosing uncomplicated P. vivax and other non-falciparum malaria; and the third will assess trials that incorporate RDTs into treatment protocols (Odaga 2011). Previous published reviews have examined travellers only (Marx 2005) or just one particular test (Cruciani 2004).
To assess the diagnostic accuracy of RDTs for detecting clinical P. falciparum malaria (symptoms suggestive of malaria plus P. falciparum parasitaemia detectable by microscopy) in persons living in malaria endemic areas who present to ambulatory healthcare facilities with symptoms of malaria, and to identify which types and brands of commercial test best detect clinical P. falciparum malaria.
Investigation of sources of heterogeneity
We planned to investigate heterogeneity in relation to the index test (by commercial test, test type and grouped by HRP-2/pLDH) and reference tests (microscopy vs PCR), as well as the study participants' age, endemicity of malaria, and geographic area (by continent).
Criteria for considering studies for this review
Types of studies
Studies evaluating one or more RDTs in a consecutive series of patients, or a randomly-selected series of patients, were eligible. Where the report did not explicitly state that sampling was consecutive, but consecutive sampling was judged most probable, the study was included. Studies were excluded if they did not present sufficient data to allow us to extract or calculate absolute numbers of true positives, false positives, false negatives, and true negatives. Studies were also excluded if they were not available in English, or if they presented insufficient information to fully assess their eligibility.
Studies recruiting people living in P. falciparum malaria endemic areas who attended ambulatory healthcare settings with symptoms of uncomplicated malaria. This included patients attending malaria clinics with self-assessed symptoms.
We excluded studies if participants:
1. were non-immune persons returning from endemic countries or were mainly recent migrant or displaced populations from non-endemic or very low endemicity areas;
2. had been treated for malaria and the test was performed to assess treatment outcome;
3. had symptoms suggestive of severe malaria;
4. did not have symptoms suggestive of malaria;
5. were recruited through active case finding (for example, door-to-door surveys).
In studies with broader inclusion criteria but which presented results stratified by subgroups, we included the data relevant to our inclusion criteria. If studies included some participants with severe malaria, and data specific to a subgroup of participants with uncomplicated malaria could not be extracted, the study was included if 90% or more of the participants had uncomplicated malaria.
Studies evaluating any immunochromatography-based RDTs specifically designed to detect P. falciparum malaria.
Commerical tests no longer available were included because they may use the same antibodies, and very similar technology, to tests that are currently available or may become available in the future. Older and more recently available versions of the same test, and tests available in both dipstick and cassette format, were included separately. Late prototype tests corresponding to one of the commercially-available types were also included.
Studies were included regardless of whether they made comparisons with other RDT tests.
Studies aimed to detect P. falciparum malaria parasitaemia. Studies that presented RDT results relating only to all types of malaria without distinction by species, but where over 98% of malaria infections by reference standard were associated with P. falciparum, were included in this review and analysed as for P. falciparum.
Studies were required to diagnose P. falciparum malaria using at least one of the following two reference standards.
- Conventional microscopy of thick blood smears, thin blood smears or both. Presence of asexual P. falciparum parasites of any density was regarded as a positive smear.
- PCR test.
We required that the reference standard was carried out on blood samples taken at the same time and from the same person as the index tests. Where studies used more than one reference standard, we presented data relating to comparisons with each.
Search methods for identification of studies
We used a single search strategy for all reviews in the series.
To identify all relevant studies, we searched the following databases using the search terms and strategy identified in Appendix 1. The date of the last search was 14 January 2010.
Cochrane Infectious Diseases Group Specialized Register; MEDLINE; EMBASE; MEDION; Science Citation Index; Web of Knowledge; African Index Medicus; LILACS; IndMED. We used the following MeSH, full text and keyword terms: malaria, Plasmodium, reagent kits, diagnosis, diagnostics, RDT, dipstick, MRDD, OptiMal, Binax Now, Parasight, Immumochromatography, antigen detection, antigen test, Combo card. We restricted the searches to human studies. We did not limit the search by language or publication status.
Data collection and analysis
Selection of studies
A single selection procedure was initially used to identify studies for inclusion in either of the two diagnostic test accuracy reviews in the series. The inclusion criteria between the reviews differed only in the target condition and parasite species. Parasite species was therefore the last aspect of the study characteristics to be assessed. One author (KA) initially assessed the titles identified by the search, excluding those obviously irrelevant to the diagnosis of malaria using RDTs.
Letters, review articles, and articles clearly irrelevant based on examination of the abstract and other notes were next excluded and the eligibility of the remaining potentially relevant articles was judged on full text publications independently by two authors (KA SJ) using a proforma. These excluded studies are listed in Characteristics of excluded studies. Any discrepancy was resolved by discussion. Where agreement could not be reached, we consulted a third author (PG or PO). Where it remained unclear whether a study was eligible for inclusion, it was excluded, and we excluded study reports in non-English language reports for logistical reasons.
Studies were named according to the surname of their first author and the year of publication. The study naming used in this review uniquely identifies multiple study cohorts from within each study report (for example as 'Bell 2001a' and 'Bell 2001b'), each of which use different reference standards or present data separately for more than one population with different characteristics. A slightly different notation (for example, 'Singh 1997(a)' and 'Singh 1997(b)') was used to refer to completely separate studies published by an author of the same name in the same year. Note that more that one RDT may be evaluated in each study cohort, thus the number of test evaluations exceeds the number of study cohorts, which exceeds the number of study reports.
Data extraction and management
A standard set of data was extracted from each study cohort, using a tailored data extraction form. Two authors from a pool of three (KA SJ CMN) independently extracted data, and any discrepancies were resolved by discussion. In cases of studies where only a subgroup of participants met the review inclusion criteria, data was extracted and presented only for that particular subgroup. Where two versions of one reference standard or index test were used, for example local clinic and expert standard microscopy or field versus laboratory testing, only the one most likely to yield the highest quality results was included in the review.
For each study, we systematically extracted data on the characteristics of the study, as shown in Appendix 2.
For each comparison of index test with reference test, data were extracted on the number of true positives, true negatives, false positives and false negatives in the form of a two by two table. RDT results are dichotomous; microscopy results were deemed positive at any level of asexual P. falciparum parasitaemia; and PCR results used the cut-off points presented by the study authors. Gametocyte-only parasitaemia was considered negative; where a study was unclear on how they had classed gametocyte-only parasitaemia, they were assumed to have used the same classification as ourselves and the data were included in the study. In cases of minor disagreement (within 2%) between two by two table data presented in a study report and reported study sample sizes or calculated accuracies, the data in the table were taken as correct. In cases where there was a large discrepancy, the data were not included in the review.
Data were extracted (Smidt 2008) using current manufacturers' instructions in interpreting the RDT results. P. falciparum only and P. falciparum as part of a mixed infection were not distinguished and were classed as positive. Non-falciparum malaria only was classed as negative for this review. Where study authors interpreted test results or presented data differently, we used all the information presented in the paper to extract data consistent with our own methods; if we were unable to do this, we did not include the data in the analyses.
Reference standard positive was defined as 'P. falciparum or mixed infection' and reference standard negative as 'no malaria parasitaemia or non-falciparum malaria parasitaemia only'.
Assessment of methodological quality
Two authors from a pool of three (KA SJ CMN) independently assessed the quality of each individual study using the checklist adapted from the QUADAS tool (Whiting 2003). Each question on the checklist was answered with a yes/no response, or noted as unclear if insufficient information was reported to allow a judgement to be made, and the reasons for the judgement made were documented. The criteria used are summarized in Appendix 3.
Statistical analysis and data synthesis
The comparisons made in this review can be considered in a hierarchy. The highest level comparison groups tests by antibody type (HRP-2 versus pLDH) and is formed by combining the test types into two groups: HRP-2 antibody-based (Types 1, 2, 3 and 6) and pLDH antibody-based (Types 4 and 5). However, the data on each test type is classified in the primary studies according to commercial brands. In order to provide a coherent description of the studies contributing to each analysis, the results are structured first by grouping studies according to their commercial brand, then grouping brands to form test types, and finally grouping test types by antibody.
The analytical strategy thus compared the test accuracy of commercial brands within each test type before making comparisons between test types, and then between antibodies. Comparative analyses first included all studies with relevant data, and were then restricted to studies that made direct comparisons between tests with the same participants, where such studies existed.
For each test type, we plotted estimates of the observed sensitivities and specificities in forest plots and in receiver-operating characteristic (ROC) space. These plots demonstrate the variation in accuracy between studies.
Meta-analyses were undertaken where adequate data were available. Hierarchical summary ROC models (HSROC) that included a random-effects term for variation in accuracy and threshold between studies, and non-symmetrical underlying ROC curves, were fitted. The average operating point for each test was identified on each curve, and average sensitivities and specificities computed. Comparisons between tests were made by adding a covariate for brand, test type or antigen to the accuracy and threshold parameters, assuming a common underlying shape. The impact of test type and antibody on the variability of random-effects of accuracy and thresholds was also investigated and separate variance terms included where required. The significance of the difference in test performance was assessed by a likelihood ratio test comparing models with and without covariate terms for accuracy and threshold. Where inadequate studies were available to estimate all parameters, the HSROC model was simplified by assuming a symmetrical shape to the summary ROC curve or fixed-effect estimates.
Where more than one commercial test of the same type was tested on the same patients against the same reference standard, we selected one type at random from the analysis by test type, in order to avoid bias due to inclusion of the same participants more than once in the analysis. We included both types in any analyses comparing commercial brands.
Investigations of heterogeneity
We investigated heterogeneity for Type 1 tests because this was the only test category for which there were sufficient studies available. We investigated variation in sensitivity and specificity by adding to the meta-analysis models covariates indicating the following characteristics: age group; P. falciparum endemicity; continent where the study took place; and adequacy of the reference standard.
Age group was classified as: children only; adults only; mixed adults and children; and 'not stated'. Studies including, for example, all ages over the age of five years were classified as 'mixed adults and children'. The age cut-off between adults and children was as used by the study authors.
Endemicity was divided into two categories: high and low. We classified endemicity as 'high' if described by the authors as 'holoendemic', 'hyperendemic' or 'high'; and 'low' if described as 'hypoendemic', 'mesoendemic', 'low' or 'epidemic-prone'. In the case of studies where a reported endemicity was not available, we imputed endemicity using geographical location information provided in the report. This involved mapping the location using 'Google Earth' onto country maps of mean parasite rate in children aged two to ten years in 2007 (Hay 2009) provided by the Malaria Atlas Project (www.map.ox.ac.uk). An example map produced during this process is shown in Figure 1. Study sites with a mean parasite rate of less than 50% were classified as 'low' endemicity to correspond with endemicities of hypoendemic and mesoendemic; study sites with a mean parasite rate of 50% and above were classified as 'high' (Hay 2008). Where the endemicity was unclear and borderline between 'high' and 'low' we assigned it 'high'. Where multiple sites of differing endemicity class were included, and separate results by site were not available, the endemicity assigned to that study was 'mixed'.
|Figure 1. Example map showing P. falciparum malaria endemicities and study locations|
For continent classification, where multi-site studies were conducted across continents and results were not available for different sites separately, the location of the study was classified according to the continent with the largest number of participants.
Sensitivity analyses were undertaken to investigate the impact of the reference standard method (PCR and PCR-adjusted microscopy) on the results obtained by microscopy alone.
Results of the search
The search identified 3971 titles, of which 3418 were excluded on the basis of title alone. A further 168 were excluded without obtaining full-text articles; 29 were excluded because they were letters; and 139 were excluded on the basis of their abstract. We were unable to obtain one article in full-text form. Full-text articles were retrieved for 384 titles, of which 307 were excluded: 254 because they were initially assessed as ineligible; 17 because the reports did not present sufficient detail for us to be sure of their eligibility or ineligibility; 18 because they were available only in non-English languages; 12 because we were unable to extract absolute numbers of true positives, false positives, false negatives and true negatives; and six because they did not present data on P. falciparum malaria, although they were eligible for other reviews in this series.
Two further studies were included as they were identified as eligible during an earlier, scoping stage of the review process but were not identified by the final search.
A total of 74 unique studies described in 79 study reports are therefore included in the review. However, as some of these studies were divided for the purposes of the review (for example, because they used two different reference standards or were conducted in two communities with differing characteristics), 89 separate study cohorts are identified. Fourteen of the 89 study cohorts evaluated more than one test: one compared seven tests, three compared three tests and ten compared two tests. Thus, there are a total of 111 test evaluations reporting a total of 60,396 test results. Microscopy was the reference standard for 104 test evaluations, PCR-adjusted microscopy for two and PCR alone for five,
Sixty-five study cohorts (40,062 participants) assessed the accuracy of Type 1 tests using microscopy as the reference standard; 16 study cohorts (13,010 participants) did the same for a Type 4 test, eight for a Type 2 test (3397 participants), five for a Type 3 test (958 participants) and three for a Type 5 test (1777 participants). Seventy-five cohorts (43,307 participants) assessed the accuracy of HRP-2 antibody-based tests and 19 cohorts (14,787) assessed the accuracy of pLHD antibody-based tests. Only four studies used PCR and one used PCR-adjusted microscopy. A summary of the numbers of studies assessing each RDT type using microscopy, PCR or PCR-adjusted microscopy is shown in Table 3.
Methodological quality of included studies
The overall methodological quality of all included study cohorts is summarized in Figure 2. Just over 50% clearly included a representative spectrum of participants attending ambulatory care settings with symptoms suggestive of malaria; the majority of the remaining studies were unclear, in most cases because they had not described the sampling methods. Around 40% reported an acceptable reference standard, 40% were unclear about the microscopy method, and 20% reported an unacceptable quality reference standard (heterogeneity relating to this criteria is investigated below). As expected, almost all the included studies reported avoidance of partial verification and differential verification, and all reported avoidance of incorporation bias. Around 65% of study cohorts reported blinding of the reference standard to the results of the index test, and around 70% reported blinding of the index test to the results of the reference standard. Only around 25% of studies reported on uninterpretable results while around 60% either explained any withdrawals or were clear that there were no withdrawals.
|Figure 2. Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.|
Twenty-four of the included studies gave details of the number of uninterpretable or invalid RDT results. Eight reported no uninterpretable RDT results; one reported that 14% of tests needed to be repeated; and 15 reported small numbers of uninterpretable test results (<1% to 5%), which were excluded from the analysis. Four studies reported small numbers of uninterpretable microscopy slides, which were excluded from the analysis.
Four key quality items (representative spectrum, adequate reference standard, blinding of reference test, and index test) are used to evaluate each RDT type in Table 4. A lower proportion of those studies assessing Type 1 and Type 4 RDTs reported an adequate reference standard than those assessing other RDT types (P=0.05) (only 25% of Type 1 evaluations and 29% of Type 2 evaluations were judged to be adequate).
PRIMARY COMPARISONS - MICROSCOPY AS THE REFERENCE STANDARD
HRP-2 antibody-based tests
Type 1 tests
There were 71 evaluations of Type 1 RDTs verified with microscopy (based on data from 40,062 individuals in 65 cohorts described in 55 publications); forty-one were conducted in Africa, 28 in Asia and two in South America. The median sample size was 269 (range 30 to 7000), and the median prevalence of falciparum malaria parasitaemia was 30% (range 1% to 92%). Only nine of the 71 evaluations were undertaken exclusively in children under the age of five. Ten different RDT brands were evaluated: Paracheck-Pf (27), ParaSight (17), ICT Malaria Pf (16), ParaHIT-F (4), PATH (2), Determine Malaria Pf (1), Rapid Test Malaria (1), Diaspot Malaria (1), New mini-Pf (1), and Hexagon Malaria (1). The earliest study was published in 1996, with the majority published between 1999 and 2007.
Sensitivities of the tests ranged from 42% to 100%, specificities from 65% to 100% (Figure 3). The meta-analytical average sensitivity and specificity (95% confidence interval (CI)) were 94.8% (93.1% to 96.1%) and 95.2% (93.2% to 96.7%), respectively, but heterogeneity was noted between studies. Comparing the ten RDT brands in an analysis of the 71 evaluations revealed no statistically significant differences (P = 0.18), although differences may be masked by the high between study heterogeneity ( Table 5, see Appendix 4 for extra figures). In an analysis restricted only to the four brands evaluated in more than 1000 patients (Paracheck-Pf, ParaSight, ICT Malaria Pf, ParaHIT-F), pairwise comparisons indicated that ICT Malaria Pf was significantly more sensitive than Paracheck-Pf and ParaSight-F (97.7% compared to 93.3% and 94.2%, respectively), whilst ParaHIT-F was significantly more specific than Paracheck-Pf, ParaSight-F, and ICT Malaria Pf (98.9% compared to 95.7%, 94.5% and 94.5%, respectively) (see Appendix 5). However, these differences were small and are based on between-study comparisons, so may have been due to differences between the studies rather than true differences between the test brands.
|Figure 3. Study results of Type 1 RDTs plotted in ROC space (by RDT brand)|
Type 2 tests
There were eight evaluations of Type 2 RDTs verified with microscopy (based on data from 3397 individuals in eight cohorts described in seven publications); seven were conducted in Asia and one in South America. The median sample size was 347 (range 113 to 896), and the median prevalence of falciparum malaria parasitaemia was 21% (range 6% to 46%). None of the evaluations were undertaken exclusively in children under the age of five. Two different RDT brands were evaluated: ICT Malaria Pf/Pv (6) and NOW ICT Malaria (2). The earliest study was published in 1999, with the majority published between 2000 and 2005.
Sensitivities of the tests ranged from 86% to 100%, specificities from 74% to 100% (Figure 4). The meta-analytical average sensitivity and specificity (95% CI) were 96.0% (94.0% to 97.3%) and 95.3% (87.3% to 98.3%), respectively. Comparing the two RDT brands in an analysis of the eight evaluations showed no statistically significant differences (P = 1.0) ( Table 5, see Appendix 4 for extra figures).
|Figure 4. Forest plot of study results of Type 2, 3 and 5 RDTs (by RDT brand)|
Type 3 tests
There were five evaluations of Type 3 RDTs verified with microscopy (based on data from 958 individuals in five cohorts described in five publications); three were conducted in Africa and two in Asia. The median sample size was 194 (range 30 to 291), and the median prevalence of falciparum malaria parasitaemia was 37% (range 25% to 57%). One of the evaluations was undertaken exclusively in children under the age of five. Three different RDT brands were evaluated: SD Malaria Antigen Bioline (2), Parascreen (2), and First Response Malaria (1). The earliest study was published in 2004.
Sensitivities of the tests ranged from 86% to 100%, specificities from 65% to 100% (Figure 4). The meta-analytical average sensitivity and specificity (95% CI) were 99.5% (71.0% to 100%) and 90.6% (80.5% to 95.7%), respectively. There were inadequate data on each RDT brand to make formal statistical comparisons (see Appendix 4 for extra figures).
Type 6 tests
No studies assessed the accuracy of Type 6 RDTs verified with microscopy.
All HRP-2 antibody based tests
There were 84 evaluations of HRP-2 tests verified with microscopy (based on data from 43,307 individuals in 75 cohorts described in 64 publications); forty-two cohorts were conducted in Africa, 31 in Asia and two in South America. The median sample size was 291 (range 30 to 7000), and the median prevalence of falciparum malaria parasitaemia was 26% (range 1% to 84%). Nine of the evaluations were undertaken exclusively in children under the age of five. Sensitivities of the tests ranged from 42% to 100%, and specificities ranged from 65% to 100%. The meta-analytical average sensitivity and specificity (95% CI) were 95.0% (93.5% to 96.2%) and 95.2% (93.4% to 99.4%), respectively.
pLDH antibody based tests
Type 4 tests
There were 17 evaluations of Type 4 RDTs verified with microscopy (based on data from 13,010 individuals in 16 cohorts described in 14 publications); eight were conducted in Africa, eight in Asia and one in South America. The median sample size was 305 (range 75 to 7000), and the median prevalence of falciparum malaria parasitaemia was 32% (range 2% to 61%). Only four of the 17 evaluations were undertaken exclusively in children under the age of five. Four different brands were assessed: OptiMAL (10), OptiMAL-IT (3), Parabank (2) and Carestart Malaria Pf/Pan (2). The earliest study was published in 1999, with the majority published between 2003 and 2007.
Sensitivities of the tests ranged from 80% to 100%, specificities from 90% to 100% (Figure 5). The meta-analytical average sensitivity and specificity (95% CI) were 91.5% (84.7% to 95.3%) and 98.7% (96.9% to 99.5%), respectively. Comparing the four RDT brands in an analysis of the 17 evaluations revealed statistically significant differences (P = 0.009) ( Table 5). Carestart Malaria Pf/Pan was observed to have a higher sensitivity and lower specificity than either OptiMAL, OptiMAL-IT or Parabank (sensitivity of 97.8% compared with 90.1%, 87.4% and 87.9%, respectively; specificity of 92.2% compared with 99.3%, 97.0% and 98.8%, respectively). See Appendix 4 for extra figures. These differences are based on between-study comparisons, so may have been due to differences between the studies rather than true differences between test brands.
|Figure 5. Study results of Type 4 RDTs plotted in ROC space (by RDT brand)|
Type 5 tests
There were three evaluations of Type 5 RDTs verified with microscopy (based on data from 1777 individuals in three cohorts described in three publications); two were conducted in Africa, one in Asia and none in South America. The median sample size was 668 (range 240 to 869), and the median prevalence of falciparum malaria parasitaemia was 23% (range 20% to 25%). None of the evaluations were undertaken exclusively in children under the age of five. Two different RDT brands were evaluated: Carestart Pf/Pv (2), and ParaSight Pf/Pv (1). The earliest study was published in 2003.
Sensitivities of the tests ranged from 96% to 99%, specificities from 93% to 100% (Figure 4). The meta-analytical average sensitivity and specificity (95% CI) were 98.4% (95.1% to 99.5%) and 97.5% (93.5% to 99.1%), respectively. There were inadequate data on each RDT brand to make formal statistical comparisons. See Appendix 4 for extra figures.
All pLDH antibody based tests
There were 20 evaluations of pLDH antibody-based tests verified with microscopy (based on data from 14,787 individuals in 19 cohorts described in 17 publications); nine cohorts were conducted in Africa, nine in Asia and one in South America. The median sample size was 343 (range 75 to 7000) and the median prevalence of falciparum malaria parasitaemia was 28% (range 2% to 58%). Four of the evaluations were undertaken exclusively in children under the age of five.
Sensitivities of the tests ranged from 80% to 100% and specificities ranged from 90% to 100%. The meta-analytical average sensitivity and specificity (95% CI) were 93.2% (88.0% to 96.2%) and 98.5% (96.7% to 99.4%), respectively.
Comparisons between RDT types
Statistical comparisons could only be made between Type 1 and Type 4 tests, as the number of studies evaluating other test types was inadequate to provide stable estimates of comparisons in the meta-analytical models. Models were fitted allowing for different degrees of heterogeneity for the two test types: results for Type 1 were more heterogeneous than Type 4. Significant differences in test accuracy (P = 0.009) were noted between Type 1 and Type 4 RDTs: Type 4 tests tended to have slightly lower sensitivity (P = 0.34) but significantly higher specificity (P < 0.001) than Type 1 tests in the comparisons based on all data (shown graphically in Figure 6). When the analysis was restricted to the seven studies with direct comparisons, the same patterns were evident, but none were statistically significant ( Table 6). Based on estimates from all studies, Type 1 tests detect on average three more cases out of every 100 people with malaria than Type 4 tests (P = 0.20), but give on average three more false positive diagnoses for every 100 people without malaria (P < 0.001).
|Figure 6. Summary ROC Plot comparing different RDT types verified with microscopy (points are meta-analytical estimates, regions are 95% confidence regions, no regions could be computed for Type 2 and 5 due to small numbers of studies)|
Four further studies provided direct comparisons between tests (Appendix 6). One study showed Type 2 to have higher sensitivity than Type 4, but lower specificity than both Type 4 and Type 1; another study showed that Type 3 tests had higher sensitivity than Type 1. The remaining studies showed no significant differences between types. As these comparisons are based on single small studies, their results should be interpreted with caution.
Comparisons between HRP-2 and pLDH antibody based RDT Types
RDT types 1 to 3 are all based on HRP-2 antibodies, while types 4 and 5 are related to detection of pLDH antigen. The process of grouping types based on this antibody classification is dominated by the results of the Type 1 tests (which constitute 65 out of 75 of the included HRP-2 antibody-based test studies) and Type 4 tests (which constitute 16 out of 19 of the included pLDH antibody-based test studies). Nine studies provide direct within-participant comparisons of HRP-2 and pLDH test types, eight of which are comparisons of a Type 1 test with a Type 4 test. As for Type 1 and Type 4 tests, it was necessary to allow for different heterogeneity between the test types in the meta-analytical model.
On average, HRP-2 antibody-based tests tend to have slightly higher sensitivity (P = 0.34) but significantly lower specificity (P<0.001) than pLDH antibody-based tests, based on analysis of all data ( Table 6; Figure 7). Differences based on direct comparisons showed the same pattern, but none of the differences were statistically significant. For every 100 malaria cases, around two more are detected with HRP-2 antibody-based tests than pLDH antibody-based tests (P = 0.34 in analysis based on all data, P=0.60 in analysis based on within-study comparisons), but this is at the cost of four false positives for every 100 people without malaria (P < 0.001 in analysis based on all data, P = 0.22 in analysis based on within-study comparisons).
|Figure 7. Summary ROC Plot comparing HRP-2-based and pLDH-based RDTs across all studies verified with microscopy (points are meta-analytical estimates, regions are 95% confidence regions)|
Investigations of heterogeneity
Heterogeneity investigations were undertaken to test for differences in RDT performance related to age, endemicity, geographical location and the use of an adequate reference standard. Analyses were restricted to the 65 test cohorts in which RDTs of Type 1 were evaluated. Results are presented in Table 7.
Nine study cohorts only recruited children aged five years or under, 28 recruited mixed age groups, and in 27 age distributions were not described. No difference in test accuracy was noted by age category (P = 0.41).
Fifty-one study cohorts were in low endemicity areas, 10 in high areas, and three were categorized as being in areas of mixed endemicity. Although specificity appeared to be lower in high endemicity areas, the differences were not statistically significant (P = 0.22).
Significant differences were seen by continent, with lower sensitivity (by 2.7%) and specificity (by 3.7%) in Africa than Asia (P = 0.01). Results from the South American studies showed very high specificity (99.4%) and low sensitivity (88.7%), but should be judged with caution due to only two studies being available.
Fifteen of the Type 1 study cohorts used inadequate reference standards and in 32 the reference standard was unclear, but their results did not differ significantly from the 17 with adequate reference standards (P = 0.34).
For all the above analyses, a sensitivity analysis was undertaken by including the one study in the review (Hopkins 2008b) that used PCR-adjusted microscopy as the reference standard; its inclusion made no difference to any of the findings.
Use of PCR as a reference standard
Five study cohorts (from four studies) used PCR as a reference standard: two Type 1 RDTs (ParaSight-F and ParaHIT-F), one Type 3 RDT (SD Malaria Antigen Bioline), one Type 4 RDT (OptiMAL-IT) and one Type 6 RDT (PALUTOP). Comparisons were made with the corresponding microscopy evaluations for the first four of these tests (Appendix 7). Use of PCR as a reference standard reduced estimates of the sensitivity of the RDTs but increased estimates of specificity compared with the microscopy-based reference standard for three of the four studies in which comparison was possible.
For two studies, results are available separately using microscopy and PCR reference standards (Banchongaksorn 1996b; Nicastri 2009b). In one study (Banchongaksorn 1996a; Banchongaksorn 1996b), both sensitivity and specificity for PCR and microscopy were within 1% of each other. In the other (Nicastri 2009a; Nicastri 2009b), specificities were 99% when verified by microscopy and 100% when verified by PCR; sensitivity verified with microscopy was 47% (95% CI 29% to 65%) compared with 72% (95% CI, 51% to 88%) for PCR. In this study with 336 participants, 26 were positive for malaria by PCR, 32 by microscopy and 18 by RDT, suggesting a relatively high rate of false positives for microscopy in the context of a low prevalence.
Five of the included studies presented data, in addition to the comparisons included in the review, on the accuracy of their microscopy reference standard against PCR (excluding one study with only two microscopy positive cases). In three studies, where the quality of the microscopy was unclear (Gaye 1999; Mens 2007b; Nicastri 2009b), sensitivity of microscopy against PCR varied between 69% and 89%; in two studies with adequate quality microscopy (Banchongaksorn 1996b; Rakotonirina 2008), sensitivity varied between 90% and 96%. Specificity of microscopy against PCR was high in all five studies, varying between 96% and 100%.
Comparing the accuracy of RDTs and local standard microscopy
In addition to the comparison of RDT against the 'gold standard' microscopy, seven of the included studies presented a comparison of local microscopy against reference standard microscopy. These studies reported widely differing results: one study showed local microscopy services to be slightly more accurate than RDTs (Kolaczinski 2004); three studies showing local microscopy to be extremely inaccurate, with very low specificities of 0% (A-Elgayoum 2009) to 25% (Tagbo 2007), or a sensitivity so low that only around half of cases were detected (De Oliveira 2009); and the others were intermediate but favouring RDTs. The findings of the two studies with an adequate reference standard are presented in Appendix 8.
Summary of results
Malaria diagnosis and treatment policies have shifted rapidly over the past few years. In 2006, in its guidelines on malaria treatment, the WHO abandoned presumptive treatment with ineffective or only partly effective treatments for the new ACTs. Now, in the second (2010) edition of these guidelines, parasitological diagnosis is expected (WHO 2010): "prompt parasitological confirmation by microscopy or alternatively by RDTs is recommended in all patients suspected of malaria before treatment is started". In primary care in most developing countries, prompt, accurate results from microscopy can't be delivered efficiently, and so demonstrating the sensitivity and specificity of these tests helps reassure policy makers pushing investment in and purchase of this technology.
For P. falciparum malaria, targeting treatment will help to reduce unnecessary drug use and thus help to avoid the development of drug resistance. The test will also help health workers exclude malaria as a cause of fever and thus improve the diagnosis and treatment of other infections. In addition, as malaria control improves as a result of all the new approaches including use of ACTs (Sinclair 2009) and other preventive measures such as impregnated mosquito nets (Lengeler 2004), transmission will drop, immunity will drop and thus prompt detection and treatment becomes even more important for reducing severe illness.
Thus, the current policy question is: how well do RDTs perform in diagnosing symptomatic patients compared to the previous standard of microscopy? There are subsidiary questions about how well the various types and individual commercial tests perform against microscopy and against each other. This information will help to inform choice, although factors such as price, product consistency, stability, and shelf life will also influence those decisions. In addition, areas vary in relation to malaria species not detected by these tests (P. vivax and other non-falciparum malaria species). Thus the choice of commercial product will also depend on whether it is important for clinicians to detect these species. For example, as malaria eradication proceeds and endemicity of malaria falls, being able to detect P. vivax is likely to become more important. In these circumstances, the sensitivity and specificity of the commercial product to P. vivax may be a factor in the choice of product. This is the subject of a forthcoming Cochrane review.
Summary of main results
The main results are summarized in the Summary of Results table (Summary of results).
- There is a large volume of research on the accuracy of RDTs in malaria endemic countries that required meta-analysis.
- In diagnosing P. falciparum malaria, all tests performed reasonably well. Most studies identified for the current review were carried out on Type 1 (HRP-2) and Type 4 (pLDH) test, with fewer reports available on the other HRP-2 tests (Types 2 and 3) and the other pLDH tests (Type 5).
- There is a trade-off between sensitivity and specificity for Type 1 and Type 4 tests. Type 1 tests were falsely negative in about 5% of P. falciparum cases and were falsely positive in about 5% of people without P. falciparum. Type 4 tests were falsely negative in 8% to 9% of cases but were falsely positive only in about 1% of non-malaria cases. The results are mirrored by the available direct within-study comparisons between tests (although results were not statistically significant). There were only two brands of Type 1 and Type 4 tests that failed to follow these patterns. These findings support the results of laboratory-based testing undertaken by WHO (WHO 2010a), and probably reflect the different antigens used by different test types. The lower specificity of Type 1 tests may be due to the use of HRP-2 antibodies, which can give a false positive result in cases where a person has recently been successfully treated for P. falciparum malaria, due to persistent antigenaemia. Analysis of all HRP-2 antibody-based tests and all pLDH antibody-based tests was undertaken and gave similar results, but was dominated by Type 1 and Type 4 tests.
- The sensitivities and specificities of Type 2, Type 3 and Type 5 tests were similar to those of Type 1 and Type 4 tests, but these three types have not been evaluated widely and robust comparisons are not possible.
- Studies of Type 1 tests conducted in Africa reported slightly lower estimates of sensitivity and specificity than those conducted in Asia. The reasons for this are unclear, and may relate to the relative quality of the studies conducted in different locations, but are most likely due to higher rates of transmission and persistent antigenaemia in Africa.
- Reporting of studies is variable: 40% reported an adequate reference standard, 40% did not provide enough information to assess the quality of the reference standard and 20% reported an inadequate reference standard. Other published studies were excluded from the review due to inadequate reporting. It would be helpful in the future for diagnostic test accuracy studies to be more carefully reported on, using the STARD (Bossuyt 2003) criteria, to ensure their inclusion in meta-analyses.
Application of meta-analysis to hypothetical cohort
Table 1 (Summary of results) summarizes the findings of the review and applies them to two hypothetical cohorts of 1000 symptomatic patients. In one of the cohorts, the prevalence of P. falciparum malaria parasitaemia is 30%, while in the other cohort it is 50%.
Falciparum malaria prevalence at 30%: on average, a Type 1 test would miss 16 P. falciparum cases, while a Type 4 test would miss 26 cases. In contrast, a Type 1 test would wrongly identify 34 non-cases as having falciparum malaria, whereas a Type 4 test would only wrongly identify nine non-cases as falciparum malaria.
Falciparum malaria prevalence at 50%: on average, a Type 1 test would miss 26 cases of falciparum malaria, while a Type 4 test would miss 43 cases. In contrast, a Type 1 test would wrongly identify 24 non-cases as having falciparum malaria, whereas a Type 4 test would only wrongly identify seven non-cases as falciparum malaria.
At very low and very high falciparum malaria prevalence: the sensitivity advantage of Type 1 tests, in terms of cases not missed, is less. For example, where prevalence is 10%, Type 1 tests would result in five cases being missed and 43 non-cases incorrectly identified as falciparum malaria. At higher prevalence, the greater sensitivity of Type 1 tests makes a greater difference; at 80% prevalence, Type 1 tests would result in 42 missed cases compared with 68 missed cases with Type 4 tests.
The numbers of false positives presented should be viewed with caution, as some RDTs may be more sensitive than microscopy.
Strengths and weaknesses of the review
The results of this review are based on strict and careful searching, study inclusion, and data extraction. The strength of this review is that it allows an assessment to be made between types and brands of test, and also provides an accurate assessment of the trade-offs.
Completeness of evidence
This is a reasonably complete data set. We excluded 18 potentially eligible studies not published in English, 17 studies that did not provide enough information to accurately assess whether they met our inclusion criteria, and 12 studies that gave only calculated values where imputation was not possible. However, it is known that studies of diagnostic test accuracy tend to be poorly indexed (Whiting 2009), and we may therefore have missed some studies despite the comprehensive search; in fact, two of the included studies were identified only in an earlier, scoping search.
Accuracy of the reference standards used:
Microscopy is regarded as the gold standard for malaria, and hence is the primary comparison, although PCR may be more sensitive. Comparisons of microscopy and PCR showed that microscopy was highly accurate when the microscopy methods were classified as 'adequate', but less accurate when the microscopy methods were of poorer quality. However, the quality of the microscopy did not explain any heterogeneity in the meta-analysis of Type 1 tests and therefore is unlikely to be an important factor in the interpretation of the study findings.
Quality and quality of reporting of the included studies:
Many of the included studies was not well reported. For example, reference standards were often not well described, there was often insufficient methodological detail, and numbers sometimes did not add up.
Only 40% of the included studies reported an adequate reference standard and 20% reported an inadequate reference standard. In Type 1 test studies, which were generally older and of lower quality, only 25% reported an adequate reference standard. As the quality of the reference standard did not explain heterogeneity in this analysis, it seems unlikely that including studies with an unclear or inadequate reference standard caused any kind of bias. In addition, only half of the included studies were explicit about patient recruitment involving a consecutive or random series of patients. Blinding of the index and reference tests was reported in the majority of studies (65% and 70%, respectively). Only 60% of studies explained withdrawals or stated that there were none. Sampling did not seem to be a significant problem, as the tests were taken at the same time, and few lost or uninterpretable test results were reported.
Interpretability of subgroup analyses:
The subgroup analysis is interpreted in relation to the antigen type, test type, and brand, and appears to make sense, although a confounding effect of quality over time cannot be excluded with the newer tests. The differences in specificity observed between HRP-2 and pLDH antibody-based tests are significant and replicate those found in systematic laboratory-based in vitro studies (WHO 2010a).
Completeness and relevance of the review:
This review covers P. falciparum malaria only, and stands alone as relevant to areas where P. falciparum malaria predominates. A further Cochrane diagnostic review in this series will cover P. vivax and other non-falciparum malaria species.
Applicability of findings to clinical practice and policy
We found no important differences in accuracy between different RDT brands within the same type. Where significant differences between tests were found, these differences were small, and were based on weaker between-study comparisons. For some types, there were insufficient data to analyse differences between brands.
We found Type 1 RDTs to be more sensitive than Type 4, and HRP-2 antibody-based tests to be more sensitive than pLDH antibody-based tests, although the differences were not statistically significant. The direction of this finding corresponds closely with a similar analysis in a diagnostic test accuracy review of RDTs for travellers with fever returning from malaria endemic areas to non-endemic areas (Marx 2005). It also corresponds with laboratory-based testing undertaken by WHO (WHO 2010a), where Type 1 tests had a lower threshold for detection of parasitaemia than Type 4 tests. However, Type 4 tests and pLDH antibody-based tests tended to be more specific, and this difference was significant.
This research assesses sensitivity and specificity in applied research settings. In the field, the quality of microscopy is likely to be lower and the RDTs may not be read so accurately (Hawkes 2009). Further research is required on effective implementation of RDTs, as they can only influence clinical practice if the results are believed and acted upon. There may be a reluctance on the part of both health providers and patients to believe negative RDT results, leading to unnecessary prescribing of antimalarials for negative cases (Tavrow 2000). Trials in this area are in the process of being summarized (Odaga 2011).
The consequences of a false positive are that someone may be treated for malaria when they are not infected. The consequences of a false negative in an endemic area, particularly when related to low parasitaemia, means the patient is unlikely to die. The infection may clear by itself, as people living in endemic areas have partial immunity; if it does not, the illness will recur and they would seek care again.
Implications for practice
The high sensitivity and specificity of RDTs means they can replace or augment microscopy for diagnosing P. falciparum malaria.
The performance of RDT types varied but the differences were not large. HRP-2-based tests tended to be more sensitive and were significantly less specific than pLDH-based tests. Choice will depend on prevalence of malaria, and we provide data in this review to assist these decisions, although policy makers will also take into account other factors relating to cost and test stability.
Implications for research
Future studies should include comparisons between new RDTs and commonly-used Type 1 and/or Type 4 RDTs in the same patients.
Studies should be reported according to the STARD guidelines (Bossuyt 2003), which will also facilitate incorporation into meta-analysis.
Further research on effective implementation of RDTs within routine clinical practice is needed.
This research was funded through a grant from the UK Department for International Development (DFID) for the benefit of developing countries.
- Top of page
- Authors' conclusions
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Presented below are all the data for all of the tests entered into the review.
Appendix 1. Appendix 1. Search strategy
Appendix 2. Data extraction: characteristic of included studies
Appendix 3. Data extraction and criteria for judgement: methodological quality
Appendix 4. Extra figures
Estimates of average sensitivity and specificity for Type 1 RDT brands (Figure 8)
|Figure 8. Summary estimates of Type 1 RDTs plotted in ROC space (by RDT brand)|
Estimates of average sensitivity and specificity for Type 4 RDT brands (Figure 9)
|Figure 9. Summary estimates of Type 4 RDTs plotted in ROC space (by RDT brand)|
ROC plot of study results for Type 2, 3 and 5 RDT brands (Figure 10)
|Figure 10. Study results of Type 2, 3 and 5 RDTs plotted in ROC space (by RDT brand)|
Study results and estimates of average sensitivity and specificity for Type 2 RDTs (Figure 11)
|Figure 11. Summary estimates of Type 2 RDTs and study results plotted in ROC space (by RDT brand)|
Study results and estimates of average sensitivity and specificity for Type 3 RDTs (Figure 12)
|Figure 12. Summary estimates of Type 3 RDTs and study results plotted in ROC space|
Study results and estimates of average sensitivity and specificity for Type 5 RDTs (Figure 13)
|Figure 13. Summary estimates of Type 5 RDTs and study results plotted in ROC space|
ROC plot of paired results which compare Type 1 and Type 4 RDT brands (Figure 14)
|Figure 14. Paired comparison of Type 1 and Type 4 RDTs. Connecting lines link the direct comparison of pairs of tests in each study.|
ROC plot of paired results which compare HRP-2-based tests and pLDH-based tests(Figure 15)
|Figure 15. Paired comparison of HRP-2-based tests and pLDH-based tests. Connecting lines link the direct comparison of pairs of tests in each study.|
Appendix 5. Type 1 RDT brands evaluated in more than 1000 participants
Appendix 6. Additional direct comparisons between test types
Appendix 7. Summary of results by RDT type and reference standard
Appendix 8. Comparison of local microscopy and RDTs verified with good quality microscopy
Duplication of test, 3 August 2011
In the background section, under the heading Alternative tests, the same information is repeated as in the previous section Reference tests.
The duplicated text has been removed and the problem resolved. Many thanks for pointing this out.
Last assessed as up-to-date: 14 January 2010.
Protocol first published: Issue 4, 2009
Review first published: Issue 7, 2011
Contributions of authors
The Cochrane Editorial Team identified this review as a priority topic for a Cochrane review. The protocol was developed jointly by the authors. Katharine Abba, Sally Jackson, and Cho-Min Naing applied inclusion criteria, extracted data and entered the data, with guidance from Paul Garner, Piero Olliaro, and Jon Deeks. Statistical analysis was carried out by Yemsi Takwoingi, Sarah Donegan and Jon Deeks. Katharine Abba wrote the first draft of the review. All authors contributed to the final manuscript.
Declarations of interest
There are no known conflicts of interest.
Sources of support
- International Medical University, Malaysia.Research grant ID 134/2007
- Liverpool School of Tropical Medicine, UK.
- Department for International Development, UK.Research Programme Grant
- NIHR Cochrane Diagnostic Test Accuracy Support Unit, Not specified.
Differences between protocol and review
We intended to consider RDTs for detecting all species of malaria in a single review. We subsequently decided to split the review into two to make it more readable.
We had intended to handsearch reference lists of included articles, contact test manufacturers for any unpublished studies, handsearch conference proceedings, and contact authors and other experts for information on ongoing and unpublished studies. However, due to the number of citations returned by our search (over 4000), these activities were not required.
We added four further exclusion criteria: studies that used active case detection to recruit participants; studies that did not present absolute numbers; studies not published in English; and studies not presenting sufficient information to enable a full assessment of their eligibility.
The CIDG editors responsible for editing this review were Dr Hasifa Bukirwa and Dr Hellen Gelband.
Medical Subject Headings (MeSH)
*Endemic Diseases; *Plasmodium falciparum [enzymology; immunology]; Antigens, Protozoan [analysis]; Biological Markers [analysis]; Diagnostic Tests, Routine [*methods]; L-Lactate Dehydrogenase [analysis]; Malaria, Falciparum [*diagnosis; epidemiology; immunology]; Parasitemia [*diagnosis; epidemiology; immunology]; Protozoan Proteins [analysis]; Sensitivity and Specificity
MeSH check words
* Indicates the major publication for the study