Meta-analysis of urine heme dipstick diagnosis of Schistosoma haematobium infection, including low-prevalence and previously-treated populations

Abstract

Background: Urogenital schistosomiasis remains highly endemic in Africa. Current control is based on drug administration, targeted either to school-age children or to high-risk communities at-large. Urine dipsticks for detection of microhematuria offer an inexpensive means for estimating infection prevalence. However, their diagnostic performance has not been systematically evaluated after community treatment, or in areas with continuing low prevalence. The objective of the present study was to perform meta-analysis of dipstick accuracy for S. haematobium infection in endemic regions, with special attention to performance where infection intensity or prevalence was low.

Methodology/principal findings: This review was registered at inception with PROSPERO (CRD42012002165). Included studies were identified by computerized search of online databases and hand search of bibliographies and existing study archives. Eligible studies included published or unpublished population surveys irrespective of date, location, or language that compared dipstick diagnosis of S. haematobium infection to standard egg-count parasitology. For 95 included surveys, variation in dipstick sensitivity and specificity were evaluated according to study size, age- and sex-specific participation, region, local prevalence, treatment status, and other factors potentially affecting test performance. Independent of prevalence, accuracy was greater in surveys of school-age children (vs. adults), whereas performance was less good in North Africa, as compared to other regions. By hierarchical ROC analysis, overall dipstick sensitivity and specificity for detection of egg-positive urine were estimated at 81% and 89%, respectively. Sensitivity was lower among treated populations (72%) and in population subgroups having lower intensity infection (65%). When the insensitivity of egg count testing was considered (and diagnosis inferred instead from combined hematuria and egg-count findings), overall dipstick sensitivity/specificity were 82%/97%, with significantly better sensitivity (92%) in high prevalence settings.

Conclusions/significance: This analysis suggests that dipsticks will continue to serve as very useful adjuncts for monitoring community prevalence following implementation of population-based control of urogenital schistosomiasis.

Figure 1. Flow chart of study search and selection strategy.

Figure 2. Forest plot of dipstick sensitivity and specificity according to prevalence of egg-positive urines in study.

All included studies are listed in descending order of study prevalence of egg-positive urines, starting with their brief citation, then individual data on observed true-positives, false-positives, false-negatives, true-negatives, then study population prevalence. Center columns indicate calculated sensitivity and specificity for each study, with their respective 95% confidence intervals. The forest plots along the right-hand side graphically indicate observed dipstick sensitivity and specificity for each study (blue squares), while the horizontal black lines indicate the 95% confidence interval for each value.

Figure 3. Receiver-operating characteristics (ROC) for dipstick performance in detection of S. haematobium egg-positive urine.

Sensitivity/specificity of dipstick performance among sub-populations having light intensity infection (solid red diamonds), as compared to dipstick test performance reported for the total population in that study (open black ellipses). The symbol for each study is proportional to study size. Lines with long dashes link the data points for each study pair (N = 25 pairs). The short-dashed diagonal line indicates the line of test non-discrimination (i.e., worst performance). The two smooth curves indicate the weighted summary ROC curves for each group—total population in black, and light-intensity subgroup in red.

Figure 4. Performance of dipsticks for detection of S. haematobium egg-positive urine in pre- and post-treatment populations.

ROC plot of sensitivity/specificity of dipstick performance among untreated populations, as compared to populations having had previous anti-schistosomal treatment. Untreated populations are indicated by open black rectangles, while previously treated populations are indicated by solid red triangles. Symbols for each study are proportional to study size. Curved lines indicate the summary performance curves estimated by HSROC for each group—for untreated populations in black (N = 81) and for treated populations in red (N = 13).

Figure 5. Paired shifts in dipstick performance from before treatment to after treatment in the same population.

Untreated populations are indicated by open black rectangles, while treated groups are indicated by solid red triangles. The symbol for each study is proportional to study size. Lines with long dashes link the data points for each study pair (N = 8 pairs).

Figure 6. Dipstick performance before treatment and after one or two rounds of therapy.

Shown are the dipstick diagnostic characteristics in the three studies that provided data on performance after multiple rounds of school age chemotherapy. Sensitivity, as described for the individual studies , , , is shown in the left panel and specificity in the right panel, with treatment status indicated as follows: solid black bars, untreated children; open bars, children previously treated one time; hatched bars, children previously treated twice.