Diagnostic tests for human Schistosoma mansoni and Schistosoma haematobium infection: a systematic review and meta-analysis

Abstract

Background: Accurate diagnosis is pivotal for implementing strategies for surveillance, control, and elimination of schistosomiasis. Despite their low sensitivity in low-endemicity areas, microscopy-based urine filtration and the Kato-Katz technique are considered as reference diagnostic tests for Schistosoma haematobium and Schistosoma mansoni infections, respectively. We aimed to collate all available evidence on the accuracy of other proposed diagnostic techniques.

Methods: In this systematic review and meta-analysis, we searched PubMed, Embase, the Cochrane Library, and LILACS for studies published from database inception to Dec 31, 2022, investigating the sensitivity and specificity of diagnostic tests for S haematobium and S mansoni infections against Kato-Katz thick smears or urine microscopy (reference tests) involving adults (aged ≥18 years), school-aged children (aged 7 to 18 years), or preschool-aged children (aged 1 month to 7 years). We extracted raw data on true positives, true negatives, false positives, and false negatives for the diagnostic tests and data on the number of participants, study authors, publication year, journal, study design, participants' age and sex, prevalence of Schistosoma infection, and treatment status. To account for imperfect reference tests, we used a hierarchical Bayesian latent class meta-analysis to model test accuracy.

Findings: Overall, we included 121 studies, assessing 28 different diagnostic techniques. Most studies (103 [85%] of 121) were done in Africa, 14 (12%) in South America, one (1%) in Asia, and one (1%) in an unknown country. Compared with the reference test, Kato-Katz thick smears, circulating cathodic antigen urine cassette assay version 1 (CCA1, 36 test comparisons) had excellent sensitivity (95% [95% credible interval 88-99]) and reasonable specificity (74% [63-83]) for S mansoni. ELISA-based tests had a performance comparable to circulating cathodic antigen, but there were few available test comparisons. For S haematobium, proteinuria (42 test comparisons, sensitivity 73% [62-82]; specificity 94% [89-98]) and haematuria (75 test comparisons, sensitivity 85% [80-90]; specificity 96% [92-99]) reagent strips showed high specificity, with haematuria reagent strips having better sensitivity. Despite limited data, nucleic acid amplification tests (NAATs; eg, PCR or loop-mediated isothermal amplification [LAMP]) showed promising results with sensitivity estimates above 90%. We found an unclear risk of bias of about 70% in the use of the reference or index tests and of 50% in patient selection. All analyses showed substantial heterogeneity (I²>80%).

Interpretation: Although NAATs and immunological diagnostics show promise, the limited information available precludes drawing definitive conclusions. Additional research on diagnostic accuracy and cost-effectiveness is needed before the replacement of conventional tests can be considered.

Funding: WHO and Luxembourg Institute of Health.

Figure 1

Study selection LCA=latent-class analysis.

Figure 2

HSROC plot with summary points (A) CCA1 tests versus single, double, quadruple, sextuple, or 16 Kato-Katz thick smears. (B) CCA1 tests versus urine filtration. Each unfilled symbol represents the individual study estimates pair (sensitivity, specificity), and larger symbols reflect a larger sample size. The filled circles are the meta-analytic estimates across the studies included in the meta-analysis. The solid lines are the summary HSROC curves and the dashed curves in panel A (green, blue, red) are the 95% prediction regions for a future study. The 95% prediction region in panel B is the whole plot area due to heterogeneity and few number of studies. CCA1=circulating cathodic antigen urine cassette assay version 1. HSROC=hierarchical summary receiver operating characteristic.

Figure 3

HSROC plot with summary points (A) Proteinuria reagent strips versus urine filtration. B) Haematuria reagent strips versus urine filtration. Each unfilled circle represents the individual study estimates pair (sensitivity, specificity), and larger circles reflect a larger sample size. The filled circle is the meta-analytic estimates across the studies included in the meta-analysis, the solid line is the summary HSROC curve, and the dashed curve is the 95% prediction region for a future study. HSROC=hierarchical summary receiver operating characteristic.