Sensitivity of the HS-RDT against HRP2 concentration
The performance of the HS-RDT on clinical samples based on HRP2 concentration was assessed by calculating the sensitivity amongst samples with different levels of HRP2. In samples with HRP2 concentrations > 1,000 pg/ml, the HS-RDT had a mean sensitivity of 99.2% across the high- and low-transmission settings (Fig. 2A and 2B, respectively) and co-RDT had a sensitivity of > 89.0% in these same samples. The sensitivity of the HS-RDT remained high in samples with HRP2 concentrations 100–1,000 pg/ml, with values of 97.1% in Uganda and 82.1% in Myanmar, whereas the sensitivities of the co-RDT were 50.0% and 10.3%, respectively. The sensitivity of the HS-RDT greatly decreased in Myanmar for samples with concentrations less than 100 pg/ml (sensitivity = 15.8%) and in Uganda in samples with concentrations less than 10 pg/ml (sensitivity = 4.9%). For both settings, the HS-RDT is significantly more sensitive than the co-RDT in samples with HRP2 concentrations that are between 10–1,000 pg/ml. Figure 2C and 2D show the probability of positivity either by HRP2 concentration or by parasite density, respectively. The difference in profiles between the Myanmar and Uganda relationships to HRP2 concentration may suggest operational differences in the study, as these would be anticipated to behave more similarly. The correlation between parasite density by qPCR and HRP2 concentration for both studies is shown in Supplementary Fig. 1.
Cross-sectional Prevalence Estimates And Sensitivity Of The Hs-rdt
Comparing the prevalence estimates obtained using an HS-RDT and PCR in cross-sectional prevalence surveys (Fig. 3), we see that in all but four of the studies the prevalence falls below the diagonal x = y line, indicating that generally HS-RDT prevalence is lower than PCR prevalence. In the very low-transmission settings, all data points fall close to this line (Fig. 3B), showing good concordance between the two diagnostics. A fitted line from a previously published meta-analysis looking at PCR versus co-RDT prevalence estimates  was added to the plots. All the HS-RDT prevalence estimates apart from one lie above line, suggesting the HS-RDT is a more accurate tool than a co-RDT for measuring prevalence across a broad range of transmission settings. In all studies, there is some discordance between the individuals that test positive for PCR and those that test positive by HS-RDT due to the tests measuring two different analytes.
One study allows observation of the impact of the PCR method used on the comparative prevalence between the HS-RDT and PCR . In Fig. 3, the red and mauve data points are from the same samples from Papua New Guinea (PNG); however, the mauve point shows prevalence as estimated using a standard PCR assay whereas the red point shows prevalence obtained by assuming any one of three PCR assays being positive means the sample is positive, including an ultra-sensitive technique that they found to be 10x more sensitive than their standard assay . Another study in Myanmar enabled comparison of the performance of the HS-RDT under controlled laboratory settings (orange) versus field conditions (turquoise) . While the prevalence estimate is lower for the HS-RDT run in field conditions versus laboratory conditions, the same drop was also observed with the co-RDT .
The sensitivity of the HS-RDT and the co-RDT in all the studies (where available) was plotted against PCR prevalence in Fig. 4. A binomial regression model was used to explore whether there was a relationship between PCR prevalence and sensitivity for each RDT. For both the HS-RDT and the co-RDT, there is a statistically significant positive relationship between PCR prevalence and sensitivity (p < 0.05). The weighted mean sensitivity of the HS-RDT across all studies was estimated to be 56.1% (95% confidence interval [CI] from weighted t-test = 46.9–65.4%) compared to 44.3% (95% CI = 32.6–56.0%) for the co-RDT. The vertical segments join the sensitivity values from the same study—the length of this segment indicates the percentage point increase in sensitivity of the HS-RDT compared to the co-RDT.
To explore the increase in HS-RDT prevalence estimates compared to those from using a co-RDT, we plotted the ratio of these values in all studies where both tests were used (Fig. 5). The weighted mean estimated ratio of HS-RDT prevalence to co-RDT prevalence is 1.42 (95% CI 1.31–1.53). This means that prevalence estimated using an HS-RDT will be on average 42% higher than if using a co-RDT.
Use For Clinical Case Management
Conventional RDTs are considered to be effective for P. falciparum clinical case management, with high sensitivity against clinical infection [33, 34] that is associated with high parasite densities and HRP2 concentrations. Here we review evidence on the diagnostic performance of the HS-RDT in a clinical setting.
One study has retrospectively evaluated the usefulness of an HS-RDT for clinical diagnosis and fever management compared to a co-RDT . Frozen blood samples from 3,000 children and 515 adults presenting with fever to an outpatient clinic in Dar es Salaam, Tanzania, were tested using a co-RDT, an HS-RDT, by ultra-sensitive qPCR to estimate parasite density, and by enzyme-linked immunosorbent assay (ELISA) to estimate HRP2 concentration. Out of 309 children testing positive by qPCR, 226 (73.1%) and 230 (74.4%) were also positive by co-RDT and HS-RDT, respectively, and out of 48 adults testing positive by qPCR, these values were 35 (72.9%) and 37 (77.1%). Four children and 0 adults were co-RDT positive and qPCR negative, and 9 children and 0 adults were HS-RDT positive and qPCR negative, out of 3,000 children and 515 adults tested. Of individuals positive for HRP2 by ELISA, 83.1% were positive by a co-RDT and 86.5% were positive by the HS-RDT. The two tests performed similarly well for samples with higher HRP2 concentrations (> 3,000 pg/ml), but the HS-RDT was more sensitive than the co-RDT on samples 100–300 pg/ml.
This study was conducted in a very low-transmission setting; test positivity rate among febrile individuals (by co-RDT) was only 7.7% in children and 6.8% in adults, which would suggest an even lower population-level prevalence. Therefore, few individuals presenting with non-malarial fevers would be expected to be co- or HS-RDT positive due to having asymptomatic infections or having been recently treated for a malaria infection (and having residual HRP2).
A second study was carried out in the same location  and tested 2,801 febrile paediatric outpatients. Of the 274 that were PCR positive, 198 and 201 were detected by a co-RDT and an HS-RDT respectively, giving sensitivity values of 72.2% and 73.4%. Furthermore, this study compared the health outcomes up to 28 days after presenting at the health facility with fever. There was no evidence that individuals with PCR-positive, co-RDT-negative infections had worse outcomes compared to PCR-negative individuals, indicating limited benefits of clearing infections in individuals with low-density infections.
A study based in Angola measured the HRP2 concentration of outpatients attending a health facility using the multiplex bead assay .The impact of a hypothetical higher-sensitivity RDT was then estimated by calculating the proportion of individuals that would have HRP2 concentrations above 200 pg/ml. They found that 81% of febrile individuals with detectable HRP2 (> 200 pg/ml) were detected by a co-RDT and an additional 10–20% would have been identified using this hypothetical HS-RDT. In addition, 52 and 77% of HRP2-positive afebrile individuals were detected using a co-RDT in two separate sites, and an additional 50–60% of individuals would have been detected with the hypothetical HS-RDT.
In Uganda, a study was conducted where the HS-RDT was used alongside a conventional RDT and microscopy for testing febrile children under the age of 5 . During 475 clinic visits over the course of a year, positivity by the HS-RDT was 55.2%, by a co-RDT it was 53.5%, and by microscopy it was 40.6%. The HS-RDT yielded only marginally higher positivity leading to the detection of an additional eight more individuals compared to a co-RDT. The co-RDT and HS-RDT each detected 61 and 69 more cases than microscopy.
Use In Screening Pregnant Women
Four published studies were identified that looked at the performance of the HS-RDT in pregnant women. A study in Benin  tested 942 samples in 327 women in the first and third trimesters and at delivery. They found that the overall positivity of the HS-RDT was 16.2% compared to 11.6% by a co-RDT and 18.3% by PCR. Based on 172 PCR-positive samples across all stages of pregnancy, the sensitivity of the HS-RDT was 60.5% compared to 44.2% by a co-RDT. The difference was even more pronounced during the first trimester, where sensitivity by HS-RDT was 57.0% compared to 38.3% by co-RDT.
The authors investigated the clinical impact of treating women that are positive by HS-RDT but negative by co-RDT (i.e., individuals that would only be detected if an HS-RDT was used) by conducting a multivariate analysis to assess the impact of diagnostic status on maternal and birth outcomes. Individuals in this category (HS-RDT+, co-RDT–) have a 3.4 times higher risk of maternal anaemia during pregnancy compared to uninfected (PCR-negative) women . Co-RDT-positive women had a 2 times higher risk of anaemia compared to uninfected women. Both these effects were statistically significant, but the difference between them was not. There was a 5.3 times higher risk of low birthweight in the HS-RDT+, co-RDT– group and a 2.3 times higher risk in the co-RDT + group compared to the uninfected group, but both effects were nonsignificant. These results indicate that the additional infections detected by an HS-RDT may have clinical significance for the mother and child.
A study in Colombia  tested 737 peripheral and placental samples using the HS-RDT as well as light microscopy, nested PCR, a Pf-only RDT and a Pf/Pv RDT. Among all samples, the HS-RDT performed comparably to the best performing co-RDT (sensitivity of 85.7% compared to 82.8%). The authors also disaggregated the data by whether each woman was symptomatic at the time the sample was taken (defined as fever with an axillary temperature ≥ 37.5°C or history of fever within the last 3 days). The sensitivity was high and exactly the same using all diagnostics among symptomatic women (85.7%, n = 61). Among asymptomatic women (n = 649), sensitivity using the HS-RDT was 71.4% compared to 64.3% with the best performing co-RDT and 50% by light microscopy, although none of the differences were statistically significant.
A second study in Colombia tested 858 pregnant women attending an ANC clinic  with an HS-RDT, a co-RDT, microscopy, loop-mediated isothermal amplification (LAMP), and both nested PCR and quantitative reverse transcription PCR (qRT-PCR). The overall prevalence of P. falciparum infection among the participants was 4.5% by the most sensitive diagnostic, qRT-PCR. Using this as the standard reference, the sensitivities of the HS-RDT, co-RDT, microscopy, and LAMP were 64.1%, 53.8%, 59.0%, and 89.7%, respectively. There were four women that were positive by HS-RDT and negative by co-RDT—they all had parasite densities < 100 parasites/µL.
Finally, the performance of the HS-RDT was retrospectively tested against reconstituted stored samples from pregnant women in Indonesia . Based on 158 samples positive for Pf by PCR, the sensitivity of the HS-RDT was 19.6% compared to 22.8% using a co-RDT, showing that in this population there was no improvement in sensitivity.
Active Case Detection
One study in Cambodia used the HS-RDT for both reactive and proactive case detection . In the reactive case detection, the households, high-risk neighbours, and co-travellers of passively detected symptomatic ‘index cases’ were screened with an HS-RDT, a co-RDT, and qPCR. In proactive case detection, all high-risk individuals in villages with a high total number of malaria cases were proactively screened. High-risk individuals were defined as those reporting fever in past 48 hours or having slept in the forest in the past month. A total of 678 individuals were tested as part of both reactive and proactive case detection. Only 26 individuals were PCR positive, of which 12 were detected by a co-RDT and 14 by the HS-RDT. Furthermore, 12 and 21 PCR-negative individuals were co-RDT and HS-RDT positive, respectively. Amongst PCR-positive samples, of the four samples that were co-RDT negative and HS-RDT positive, two had parasite densities > 300 parasites/µL, indicating that the HS-RDT detected additional potentially clinically relevant infections, although the number of samples in this category is too small to make any conclusions. A second study in Cambodia used the HS-RDT to conduct active monthly screening of asymptomatic high-risk forest and plantation workers in 11 villages . In this very low-transmission setting, the HS-RDT had a sensitivity of 66.6% with PCR as the gold standard. The maximum parasite density amongst individuals with false negative HS-RDT results was less than 400 parasites/µL and the median parasite density was only 2 parasites/µL, indicating the test missed mostly low-density infections.
In Zambia, the HS-RDT was used in a reactive focal test and treat study where households of index cases detected at a facility were tested with both a co-RDT and HS-RDT  (HS-RDT data unpublished). Prevalence was very low in this trial, with only 4/205 individuals testing positive by PCR. The HS-RDT detected all four of these individuals and the co-RDT only detected three.
In another very low transmission setting in Laos (34), the HS-RDT was used in active case detection (HS-RDT data unpublished). Of 11,771 individuals tested by PCR and HS-RDT there were only 63 and 66 positives, respectively. Of the 63 PCR positive samples, 43 of them were also positive by the HS-RDT.