Sensitivity and Specificity of a Urine Circulating Anodic Antigen Test for the Diagnosis of Schistosoma haematobium in Low Endemic Settings

Background Elimination of schistosomiasis as a public health problem and interruption of transmission in selected areas are key goals of the World Health Organization for 2025. Conventional parasitological methods are insensitive for the detection of light-intensity infections. Techniques with high sensitivity and specificity are required for an accurate diagnosis in low-transmission settings and verification of elimination. We determined the accuracy of a urine-based up-converting phosphor-lateral flow circulating anodic antigen (UCP-LF CAA) assay for Schistosoma haematobium diagnosis in low-prevalence settings in Zanzibar, Tanzania. Methodology A total of 1,740 urine samples were collected in 2013 from children on Pemba Island, from schools where the S. haematobium prevalence was <2%, 2–5%, and 5–10%, based on a single urine filtration. On the day of collection, all samples were tested for microhematuria with reagent strips and for the presence of S. haematobium eggs with microscopy. Eight months later, 1.5 ml of urine from each of 1,200 samples stored at -20°C were analyzed by UCP-LF CAA assay, while urine filtration slides were subjected to quality control (QCUF). In the absence of a true ‘gold’ standard, the diagnostic performance was calculated using latent class analyses (LCA). Principal Findings The ‘empirical’ S. haematobium prevalence revealed by UCP-LF CAA, QCUF, and reagent strips was 14%, 5%, and 4%, respectively. LCA revealed a sensitivity of the UCP-LF CAA, QCUF, and reagent strips of 97% (95% confidence interval (CI): 91–100%), 86% (95% CI: 72–99%), and 67% (95% CI: 52–81%), respectively. Test specificities were consistently above 90%. Conclusions/Significance The UCP-LF CAA assay shows high sensitivity for the diagnosis of S. haematobium in low-endemicity settings. Empirically, it detects a considerably higher number of infections than microscopy. Hence, the UCP-LF CAA employed in combination with QCUF, is a promising tool for monitoring and surveillance of urogenital schistosomiasis in low-transmission settings targeted for elimination.


INTRODUCTION
2 State the research questions or study aims, such as estimating diagnostic accuracy or comparing accuracy between tests or across participant groups.
8 Here, we assess the accuracy of the UCAA2000 for the diagnosis of S. haematobium in three lowprevalence scenarios (<2%, 2-5%, and 5-10%), as determined with a single urine filtration.In the absence of a true "gold" standard, sensitivity and specificity were determined empirically and by means of latent class analysis (LCA).METHODS Participants 3 The study population: The inclusion and exclusion criteria, setting and locations where data were collected.
10 The urine samples used for the diagnostic investigations presented here were collected from children aged 9-12 years visiting primary schools in 16 shehias on Pemba Island between March and May 2013.
4 Participant recruitment: Was recruitment based on presenting symptoms, results from previous tests, or the fact that the participants had received the index tests or the reference standard? 13,14 The selection of shehias with S. haematobium prevalences of <2%, 2-5%, and 5-10% for inclusion into the present study was based on results of the initial urine filtration examination performed on the day of sample collection and including all children with written informed consent, microhematuria, and urine filtration results.For assessing diagnostic accuracy, however, we only included data from individuals with complete diagnostic results on (i) reagent strip testing; (ii) urine filtration reading; (iii) UCAA2000 testing (considering indecisive results either as positive (UCAA2000+) or as negative (UCAA2000-) or as missing); and (iv) QCUF reading into the final analysis.While urine samples stored for UCP-LF CAA examination were not selected at full random (i.e., only urine samples of sufficient amount of the first 100 among 130 collected samples per school were stored), we yet considered this approach as valid and assumed complete randomness of missing samples (and that missing values are unrelated to the status of S. haematobium infection), since the overall percentage of positive individuals detected by the initial urine filtration did not differ between the initially sampled group (3.3%;Table 1) and the group included into the final analysis (3.4%; Table 2).
5 Participant sampling: Was the study population a consecutive series of participants defined by the selection criteria in item 3 and 4? If not, specify how participants were further selected.
15, Figure 1 To meet the prevalence thresholds and sample size for the study, we selected eight primary schools with a prevalence of S. haematobium <2%, four schools with a prevalence of 2-5%, and four schools with a prevalence of 5-10% based on single urine filtration readings per child.Overall, from the 16 selected schools, 2,067 children were randomly selected to participate in the annual parasitological survey in 2013.Among them, 298 did not provide written informed consent from their parents and were therefore not asked to submit a urine sample (Figure 1).An additional 29 children did not submit a urine sample of sufficient amount to perform reagent strip and urine filtration examinations.Hence, the initial S. haematobium prevalence at the unit of the school was calculated from urine filtration results of 1,740 children.

Test methods
7 The reference standard and its rationale.
Diagnostic accuracy parameters including 95% confidence intervals (CIs) were assessed using two different approaches.In the first approach, we considered the combined results of QCUF and UCAA2000+ as imperfect "gold" standard and calculated the sensitivity of each test by comparing its performance against the imperfect "gold" standard.Assuming a specificity of 100%, the sensitivity of all diagnostic tests was calculated for (i) combined data from all individuals included into the final analysis and (ii) stratified data according to the originally selected different prevalence levels (<2%, 2-5%, and 5-10%).To assess a correlation between CAA pg/ml levels and the number of eggs detected in 10 ml urines or microhematuria grading identified with reagent strips, we applied the non-parametric Spearman's rank correlation test.
In the second approach, in the absence of a true "gold" standard, we used LCA to estimate the sensitivity, specificity, and model estimated prevalences for reagent strip, QCUF, and  1) and the group included into the final analysis (3.4%; Table 2).
From this subsample, we calculated 'empirical' prevalences obtained by each diagnostic method assuming a 100% test specificity.Diagnostic accuracy parameters including 95% confidence intervals (CIs) were assessed using two different approaches.In the first approach, we considered the combined results of QCUF and UCAA2000+ as imperfect "gold" standard and calculated the sensitivity of each test by comparing its performance against the imperfect "gold" standard.Assuming a specificity of 100%, the sensitivity of all diagnostic tests was calculated for (i) combined data from all individuals included into the final analysis and (ii) stratified data according to the originally selected different prevalence levels (<2%, 2-5%, and 5-10%).To assess a correlation between CAA pg/ml levels and the number of eggs detected in 10 ml urines or microhematuria grading identified with reagent strips, we applied the non-parametric Spearman's rank correlation test.
In the second approach, in the absence of a true "gold" standard, we used LCA to estimate the sensitivity, specificity, and model estimated prevalences for reagent strip, QCUF, and UCAA2000 [35][36][37].Four LCA models were applied and validated.The exact procedure is presented in supplementary file 1 (S1) and model details have been described by Ibironke and colleagues ( 2012) [36].The four LCA models were fitted using MPlus V7 [34] with full information maximum likelihood estimation and assuming that data were missing at random.We included the indecisive results of the UCAA2000 in all LCA reproducibility, if done.

Participants
14 When study was performed, including beginning and end dates of recruitment.
11,12 At the day of collection, between March and May 2013, all urine samples of sufficient amount (at least 10 ml) were examined by trained laboratory technicians for microhematuria using reagent strips (Hemastix; Siemens Healthcare Diagnostics GmbH, Eschborn, Germany), and for the presence and number of eggs detected under a microscope using the urine filtration method with polycarbonate filters (Sterlitech, Kent, WA, USA).
The frozen samples from children from the 16 shehias selected for this study were examined with the UCAA2000 or UCAA250 assays in November 2013 at PHL-IdC.
The stored urine filtration slides from all individuals, whose urines were examined with a UCP-LF CAA test, were retrospectively re-read between November 2013 and January 2014 by a post-doctoral fellow (CIC) blinded to the reagent strip, initial urine filtration, and UCP-LF CAA results.
15 Clinical and demographic characteristics of the study population (at least information on age, gender, spectrum of presenting symptoms).
Table 1 Please see Table 1 16 The number of participants satisfying the criteria for inclusion who did or did not undergo the index tests and/or the reference standard; describe why participants failed to undergo either test (a flow diagram is strongly recommended).
Figure 1 Please see Figure 1 Test results 17 Time-interval between the index tests and the reference standard, and any treatment administered in between.
11 At the day of collection, between March and May 2013, all urine samples of sufficient amount (at least 10 ml) were examined by trained laboratory technicians for microhematuria using reagent strips (Hemastix; Siemens Healthcare Diagnostics GmbH, Eschborn, Germany), and for the presence and number of eggs detected under a microscope using the urine filtration method with polycarbonate filters (Sterlitech, Kent, WA, USA).
At the day of collection, before reagent strip and urine filtration were performed, an amount of 1.8 ml urine was frozen and stored at -20°C from children with IDs 1-100 from each shehia for future examinations.The frozen samples from children from the 16 shehias selected for this study were examined with the UCAA2000 or UCAA250 assays in November 2013 at PHL-IdC.
All tests were done from the same urine samples, hence no treatment was given between sample collection 18 Distribution of severity of disease (define criteria) in those with the target condition; other diagnoses in participants without the target condition.
16, Table 4 Noteworthy, the geometric mean egg count decreased significantly from highest to lowest prevalence settings from 0.22 eggs/10 ml urine to 0.05 eggs/10 ml urine.19 A cross tabulation of the results of the index tests (including indeterminate and missing results) by the results of the reference standard; for continuous results, the distribution of the test results by the results of the reference standard.
Table 3 Please see Table 3 20 Any adverse events from performing the index tests or the reference standard.
NA The test was performed on urine; no adverse events occur from urine collection.accuracy and measures of statistical uncertainty (e.g.95% confidence intervals).

16, 17
Table 4, Applying a combination of the QCUF and UCAA2000+ as imperfect diagnostic "gold" standard, the UCAA2000+ had the highest overall sensitivity of 95.2%, followed by the UCAA2000-with a sensitivity of 69.4% (Table 4).The QCUF and reagent strips showed very low sensitivities (24.9% and 16.6%, respectively).