Available via license: CC BY 4.0
Content may be subject to copyright.
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
https://doi.org/10.1186/s40249-025-01292-x
RESEARCH ARTICLE Open Access
© The Author(s) 2025. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecom‑
mons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Infectious Diseases of Poverty
Detecting Schistosoma infections inendemic
countries: adiagnostic accuracy study inrural
Madagascar
Eva Lorenz1,2,3* , Ravo Razafindrakoto4 , Pia Rausche2,5 , Zaraniaina Tahiry Rasolojaona4,
Nantenaina Matthieu Razafindralava4, Alexandre Zerbo1 , Yannick Höppner6 , Heidrun von Thien6,
Njary Rakotozandrindrainy7, Cheick Oumar Doumbia8 , Philipp Klein1, Jean‑Marc Kutz2,5,
Paul L. A. M. Corstjens9 , Claudia de Dood9 , Pytsje T. Hoekstra10 , Govert J. van Dam10 , Anna Jaeger1,2,
Norbert Georg Schwarz1,2, Egbert Tannich11 , Mala Rakoto Andrianarivelo4 , Raphael Rakotozandrindrainy7,
Rivo Andry Rakotoarivelo12 , Jürgen May1,2,13 , Tahinamandranto Rasamoelina4 and Daniela Fusco2,5*
Abstract
Background Schistosoma haematobium and S. mansoni are endemic in Madagascar, but reliable diagnostic tools are
often lacking, contributing to exacerbate transmission and morbidity. This study evaluated the diagnostic accuracy
of three tests for schistosome infection in Malagasy adults from areas of medium to high endemicity.
Methods This cross‑sectional study enrolled adults from three primary health care centres in Madagascar. Urine
and blood samples were tested for schistosome infection using polymerase chain reaction (PCR), up‑converting
reporter particle lateral flow for the circulating anodic antigen (UCP‑LF CAA), and point‑of‑care circulating cathodic
antigen (POC‑CCA) tests. Bayesian latent class models were used to assess diagnostic accuracies and disease
prevalence.
Results Of 1339 participants, 461 were from S. haematobium and 878 from S. mansoni endemic areas. Test detection
rates were 52% (POC‑CCA), 60% (UCP‑LF CAA), and 66% (PCR) in the S. haematobium area, and 54%, 55%, and 59%
respectively in the S. mansoni area. For S. haematobium, PCR and UCP‑LF CAA showed high sensitivity (Se, median
95.2% and 87.8%) but moderate specificity (Sp, 60.3% and 66.2%), while POC‑CCA performed moderately (Se: 64.5%;
Sp: 59.6%). For S. mansoni, PCR and POC‑CCA demonstrated high diagnostic accuracy (Se > 90%, Sp > 80%), while UCP‑
LF CAA showed good sensitivity (79.9%) but moderate specificity (69.7%).
Conclusions While population‑level prevalence estimates were similar across tests, individual‑level agreement
was only low to moderate. Our findings suggest that optimal diagnostic strategies should be tailored to specific
endemic settings, continued development of accurate diagnostics suitable for highly endemic settings remains
a priority.
Keywords Schistosomiasis, Diagnostics, Imperfect reference standards, Bayesian latent class models, Prevalence,
Madagascar
*Correspondence:
Eva Lorenz
eva.lorenz@bnitm.de
Daniela Fusco
fusco@bnitm.de
Full list of author information is available at the end of the article
Page 2 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
Background
Accurate diagnosis is crucial for the control and elimi-
nation of schistosomiasis, yet available diagnostic tools
often fail to meet the needs in endemic settings [1–3].
In line with the neglected tropical diseases (NTD) Road-
map 2021–2030, schistosomiasis is targeted for elimina-
tion as a public health problem, requiring more sensitive
diagnostics for both population screening and individual
case management [4]. Schistosomiasis ranks third among
NTDs in terms of disability-adjusted life years (DALYs),
with up to 2.5 million reported cases [5, 6], and is a major
public health and socio-economic challenge in many sub-
Saharan African (SSA) countries. Over 700 million peo-
ple live in endemic areas and are at risk of infection [7,
8] with an estimated 251.4 million requiring preventive
treatment in 2021, with approximately 90% of those living
in SSA [9, 10]. e disease is caused by Schistosoma spp.,
transmitted to humans through prolonged contact with
contaminated fresh water [11]. S. mansoni causes the
intestinal form of the disease, and S. haematobium causes
urogenital schistosomiasis [7, 12, 13]. Both can become
chronic, leading to serious health conditions such as liver
fibrosis, genital schistosomiasis and cancer.
Treatment with praziquantel (PZQ) is safe and effica-
cious against all schistosome [12, 14, 15]. Lower cure
rates have been reported in numerous areas [16, 17], and
understanding PZQ resistance is hampered by the lack of
sensitive and accurate Schistosoma detection tools. While
mass drug administration (MDA) is widely used to con-
trol schistosomiasis in endemic areas [14, 18], the pre-
cise extent of disease burden remains unclear [7, 19]. e
global burden of schistosomiasis is estimated at 1.9 mil-
lion DALYs [6, 20–22]. MDA programmes alone are not
enough to eliminate schistosomiasis. WHO has therefore
provided guidance and advocacy on non-medical inter-
ventions to improve control [23, 24].
Current diagnosis of schistosomiasis is based on eggs in
stools or urine, depending on the species involved. WHO
recommends microscopy of three urine or stool samples
collected over three consecutive days as reference stand-
ard [23]. While microscopy is cost-effective [25–28], its
sensitivity can be low, especially for low-intensity infec-
tions. In addition, the collection of multiple samples is
logistically challenging. is study evaluates the three
diagnostic approaches, i.e., point-of-care circulating
cathodic antigen (POC-CCA), up-converting reporter
particle lateral flow circulating anodic antigen (UCP-LF
CAA), and real-time polymerase chain reaction (PCR).
POC-CCA has been validated for S. mansoni diagnosis,
with sensitivity (Se) of 60% to 90% depending on infec-
tion intensity and endemicity [29]. WHO recommends it
for population-based prevalence assessment in S. man-
soni endemic areas [23], though its performance for S.
haematobium remains variable [30]. UCP-LF CAA [31],
has demonstrated promising results across multiple set-
tings [32–34], with specificities (Sp) exceeding 95% [35].
PCR-based diagnosis, with serum-based detection offers
high Se for both species [36]. Practical implementation
in endemic settings faces infrastructure and resource
challenges. Systematic evaluation of these methods in
endemic settings is essential to reduce morbidity and
progress towards elimination and enabling fit-for pur-
pose application of available tools [37, 38].
Schistosomiasis in Madagascar affects over 50% of the
population. S. haematobium occurs in the western and
northern half of the island, while S. mansoni in the east-
ern and southern parts and the central highlands [39, 40].
Co-infections occur in the north-central and south-west-
ern regions [41]. e national program is mostly based on
MDA. Even though MDAs are repeated frequently with a
good territorial coverage, the country is lagging behind
schedule for the adaptation of the national guidelines
to those recently released by the WHO [42]. e heavy
centralization of the health system so as the difficulties
of accessing health services by the Malagasy population,
creates strong barriers for the adaptation of schistoso-
miasis related services. Over 60% of the population lives
more than five kilometres from the nearest health care
centre [43] and in 2021, it was reported that Madagascar
had a healthcare worker density of less than 0.5 per 1000
inhabitants [44].
is study investigates the diagnostic accuracy of three
candidate diagnostic tests (Table 1, characteristics of
UCP-LF CAA and POC-CCA adapted from [37]) for use
in endemic settings. We selected the POC-CCA, UCP-LF
CAA and PCR for the detection of Schistosoma infec-
tion in individuals attending primary health care facilities
in Madagascar. ese tests could fill diagnostic gaps for
Schistosoma detection at different levels of the healthcare
system. We used Bayesian latent class models (BLCM) to
assess diagnostic accuracy [45] estimating prevalence, Se,
and Sp for each tests.
Methods
Primary dataset details: study design, setting
andparticipants
e cross-sectional diagnostic accuracy study enrolled
adult patients at primary health care centres (PHCC)
in Madagascar between March 2020 and January 2021
based on consecutive sampling. Specifically, the sam-
ple collection took place in Andina (20°30′58.8ʺS,
47°09′05.2ʺE) and Tsiroanomandidy (18°46′21ʺS,
46°02′57ʺE) in the centre, Ankazomborona (16°06′50ʺS,
46°45′24ʺE) in the north western part of the island. e
centres were selected based on their location in areas
suspected to be endemic for S. mansoni (Andina and
Page 3 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
Tsiroanomandidy) or S. haematobium (Ankazomborona)
respectively, and their accessibility for study implemen-
tation [19, 46]. Participants were eligible for inclusion
if they were at least 18years of age in absence of fever,
history of transfusion or congenital anaemia and epi-
leptic or convulsive episodes and after written informed
consent was obtained. Participants with a transfusion
or congenital anaemia were excluded to limit interfer-
ence with co-infections and/or genetic factors that might
alter the test performances, participants with epileptic or
convulsive episodes were excluded to limit risk of treat-
ment with PZQ due to possible co-infections with Tenia
solium. Participants’ basic sociodemographic and clinical
characteristics (date of admission, sex, age, area of resi-
dence and presenting symptoms) were recorded in RED-
Cap. We excluded individuals treated with PZQ within
12months prior to enrolment to rule out that PCR could
detect cured non-active infections. All participants were
tested using the three tests under evaluation. Conse-
quently, the diagnostic data represented a sample drawn
from this target population that informed the estimation
of prevalence and Se and Sp of the tests. Participants who
were scored as trace or positive by POC-CCA received
immediate treatment with PZQ.
Sample collection andprocessing
POC-CCA analysis was performed by a trained labora-
tory technician on a single urine sample taken on the day
of recruitment. Tests were scored using the G-scores and
grouped into three categories: negative, trace, and posi-
tive. For the UCP-LF CAA analysis, a sample of urine (a
maximum of 25 ml) was taken from each participant
and stored locally at room temperature for a maximum
of seven days until transport to the central laboratory to
reduce sample deterioration. On receipt at the central
laboratory (Charles Mérieuxinfectiologycentre, Antana-
narivo), samples were aliquoted for long-term storage at
–80°C. For the PCR analyses, a sample of 9ml of venous
blood was collected in serum separating tubes from
each participant. e blood samples were centrifuged at
1600 × g for 10 min and two aliquots of 1ml serum each
were prepared by qualified laboratory technicians. e
samples were stored at –20°C according to the required
quality standards and transferred to long-term storage at
−80°C in Madagascar. One of the serum aliquots was
shipped to Hamburg on dry ice and stored at − 80°C
until UCP-LF CAA testing.
Diagnostic test methods
Target condition
e schistosome infection status was determined by
three laboratory tests (i.e., POC-CCA, UCP-LF-CAA,
PCR). In other words, we considered the observed data of
the three tests as indicators of an underlying, not directly
observable variable (i.e., S. haematobium or S. mansoni
active infection). Each participant had one result per test.
Point‑of‑care circulating cathodic antigen (POC‑CCA)
A commercial immunochromatographic lateral flow
POC test (Rapid Medical Diagnostics, South Africa—
batch number: 200326039) was used to test urine for the
presence of CCA of Schistosomaspp. and was performed
using the G-scores [47]. e highest CCA concentra-
tions are detected in S. mansoni infections, which makes
the test particularly suitable for the diagnosis of intesti-
nal schistosomiasis. e concentrations in urogenital
schistosomiasis (S. haematobium) are variable, and also
appear to differ from region to region. In general, there
is a link between the Se of the test and the intensity of
the S. haematobium infection. Some moderate to severe
infection with S. haematobium can be diagnosed with
the CCA urine test strip though the test lacks Se for
mild S. haematobium infections. Briefly, the assay uses
only two drops of urine into a cassette. e test was read
after 20min and the band density was recorded accord-
ing to a semi-quantitative scoring system on a scale of
1–10 (G-scores) and divided into three categories: nega-
tive, trace and positive [47–49]. Samples were considered
positive if the CCA score was equal or higher than the
cut-off of G4. An internal quality control was included
to validate the cut-off of the POC-CCA (G4). No clinical
Table 1 Description of the three laboratory tests assessed
BNITM Bernhard Nocht Institute for Tropical Medicine, POC-CCA point- of-care circulating cathodic antigen, PCR polymerase chain reaction, PHCC primary health care
centre, UCP-LF CAA up-converting reporter particle lateral ow circulating anodic antigen
Diagnostic Sample type Ease of use Cost of test Where assessed Metric Denition of positive
POC‑CCA Urine Field‑applicable, rapid Low PHCCs in Madagascar Intensity of band
against metric G4–G10
UCP‑LF CAA Urine Minimal laboratory
facilities Moderate to high Laboratory in Mada‑
gascar Antigen concentrations
in pg/ml Cut‑off of ≥ 2 pg/ml
PCR Serum Sophisticated laboratory
facilities High Laboratory at BNITM
Hamburg Cycle threshold values Standard curve
Page 4 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
information was provided to the performers or readers of
the test.
Up‑converting reporter particle lateral ow circulating
anodic antigen (UCP‑LF CAA) assay
e UCP-LF-CAA, a non-commercially available assay,
detects a Schistosoma genus-specific antigen originat-
ing from the gut of the adult worm. e urine samples
were pre-treated with tri-chloroacetic acid and after cen-
trifugation, the clear supernatants were concentrated by
ultracentrifugation. e samples were then incubated
with the monoclonal UCP antibody conjugate, then LF
strips were added to the solution and left to run and
dry overnight. UCP-LF CAA analysis was performed
on a urine sample on the day. For quantitative measure-
ment of bound CAA, the strips were scanned by an UCP
strip reader and results analysed with a special software
[31]. e materials used in the UCAAhT417 format of
the UCP-LF CAA diagnostic test have a lower QC limit
of 2pg/ml when performing a single test. Samples were
considered positive if the CAA concentration exceeded
the pre-specified cut-off of 2pg/ml implying high clini-
cal Sp [31]. No clinical information was provided to the
performers or readers of the test.
Polymerase chain reaction (PCR)
Semi-quantitative standardised PCR was performed on
serum to detect a species-specific cell free circulating
deoxyribonucleic acid (DNA) derived from viable schis-
tosomula, adult parasites or disintegrated eggs [50]. e
analyses were performed in the laboratory of the Bern-
hard Nocht Institute for Tropical Medicine, in Hamburg,
Germany. e PCR analysis was based on the previously
published protocol of Frickmann et al. [36], targeting
Dra1 sequence of the S. haematobium complex and the
Sm1-7 sequence of the S. mansoni complex. Specifically,
the primers used for the amplification were: S. man-
soni—Forward Primer: 5′ CAA CCG TTC TAT GAA
AAT CGT TGT 3′, S. mansoni—Reverse Primer: 5′ CCA
CGC TCT CGC AAA TAA TCT 3′, S. mansoni—Probe:
‘Fam-TCC GAA ACC ACT GGA CGG ATT TTT ATG
AT-BHQ1’, S. haematobium—Forward Primer: 5′ GAT
CTC ACC TAT CAG ACG AAA C 3′, S. haematobium—
Reverse Primer: 5′TCA CAA CGA TAC GAC CAA C 3′,
S. haematobium—Probe: 5′ Joe-TGT TGG AAG ZGC
CTG TTT CGC AA-BHQ1 3′ all synthesized by Biom-
ers.net, Ulm, Germany. Primers and a probe were added
for the detection of Phocid herpesvirus (PhHV) DNA
as internal positive control (PhHV—Forward Primer:
5′ GGG CGA ATC ACA GAT TGA ATC 3′, PhHV—
Reverse Primer: 5′ GCG GTT CCA AAC GTA CCA A
3′, PhHV—Probe: 5′ Cy5.5-TTT TTA TGT GTC CGC
CAC CA-BBQ 3′). All results with a clean sigmoid curve
within the PCR cycles were considered positive against a
pre-specified threshold [36, 51]. No clinical information
was provided to the performers or readers of the test.
Latent class model specication
Assessing the accuracy of tests for schistosomiasis infec-
tions is challenging as there is no gold standard. Tra-
ditional approaches that assume a perfect reference
standard can lead to biased accuracy estimates. While
Composite Reference Standards (CRS) are valuable tools
when appropriately implemented, they too have limita-
tions when treated as perfect references. To address these
challenges, we employed Latent Class Models (LCM),
a statistical framework that explicitly accounts for the
imperfect nature of all tests under evaluation [52]. Our
Bayesian implementation of these models (BLCM) offers
additional advantages by allowing the incorporation of
prior knowledge from previous studies and appropriately
handling the complex uncertainty inherent in evaluat-
ing multiple imperfect tests simultaneously. While this
approach can provide less biased estimates of prevalence
and test accuracy than conventional methods, it must be
acknowledged that these methods are considered meth-
odologically complex and difficult to validate [45].
Diagrammatic representation
We first created a heuristic diagram for schistosome
infection to illustrate our assumptions about the relation-
ships between the observed test results and schistosome
infection (Fig.1). is diagram identifies the measurand
of each test, i.e., the quantity it measures. POC-CAA,
UCP-LF CAA, and PCR use different techniques to meas-
ure (i) antigen of active worms in urine samples (POC-
CCA and UCP-LF-CAA) and (ii) the presence of parasitic
worms in cell free circulating DNA from serum (PCR).
While all three tests detect products of active worm
infection, they differ in their biological mechanisms: PCR
detects DNA released from worms into circulation, while
Fig. 1 Heuristic model. The model shows the assumed relationships
between latent class (oval) and diagnostic test results (rectangles). All
measures are worm‑based. Abbreviations: DNA deoxyribonucleic acid,
POC-CCA point‑of care circulating cathodic antigen, PCR polymerase
chain reaction, PZQ praziquantel, UCP-LF CAA up‑converting reporter
particle lateral flow circulating anodic antigen
Page 5 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
POC-CCA and UCP-LF CAA detect specific antigens
excreted by worms that appear in urine. POC-CCA and
UCP-LF CAA are both antigen tests that target different
antigens and were performed on the same urine sample.
e underlying semi-quantitative POC-CCA values and
the UCP-LF CAA antigen concentrations may be posi-
tively correlated and are both influenced by the intensity
of the infection (worm load/burden). While the relation-
ship between worm burden and the amount of detectable
circulating DNA is not fully understood, all three tests
might produce false negative results particularly in indi-
viduals with low infection intensity, potentially leading to
correlated errors in test outcomes.Based on our meth-
odological considerations and biological understand-
ing of the tests’ mechanisms,four probable correlations
between diagnostic tests were explored hypothesising
that underlying worm burden may lead to a conditional
dependence between either two or all three test results,
even if the tests are based on different mechanisms.
Statistical model
e observed diagnostic test results were assumed to be
results from the two underlying latent classes; (1) schis-
tosome infection measure-positive, and (2) schistosome
infection measure-negative. In other words, we consid-
ered the observed data of the three diagnostic tests as
indicators of an underlying, not directly observable vari-
able (i.e., S. haematobium or S. mansoni infection). e
unknown parameters of the model were the prevalence
of the two latent classes and each test’s Se and Sp with
respect to its measure.
Bayesian model estimation
We used a Bayesian approach to fit the LCM (BLCM)
to estimate prevalence, Se, and Sp [53]. All parameters
were estimated with 95% credible intervals using Stan
version 2.32.2 (Stan Development Team, New York, NY,
USA) and R version 4.4.1 (R Foundation for Statistical
Computing, Vienna, Austria). As the posterior distribu-
tions of the parameters of interest (Se, Sp, prevalence)
could not be computed analytically, we sampled from
the posterior distributions using a Markov Chain Monte
Carlo approach with the rstan package through Rstudio
(Version 2023.09.0 + 463, R Version 4.3.1). All point esti-
mates reported are posterior medians with correspond-
ing 95% credible intervals, unless otherwise specified.
For the main models, non-informative priors were used
for all models with truncated prior distributions for the
non-specific tests’ Se and Sp to contain them above 40%
and avoid label switching (mirror solutions). Model esti-
mates from alternative models using informative priors
for UCP-LF CAA and PCR Sp and their specification can
be found in section SI2.1 and SI2.3 (see Supplementary
Information SI). e Supplement contains details of
model specifications, sampling, and model checking. e
BLCM models varied from a model assuming conditional
independence between all tests (Model 0) to those con-
sidering conditional dependence between a single pair
of antigen tests using fixed effects (Model 1), those that
use random effects to represent dependence between
all tests within an infection class (Models 2 and 3) and
finally a model that simultaneously accounted for condi-
tional dependence between all three tests among those
individuals truly infected and those individuals truly not
infected using random effects (Model 4). A class of fixed
and random effect models described by Dendukuri and
Joseph were used to take account for conditional depend-
ence between tests [54, 55]. Individuals with invalid or
missing test results were assumed to be missing at ran-
dom though excluded from analyses [invalid and miss-
ing results were 35 out of 1374 (2.5%)] i.e., complete case
analyses were performed. Bayesian leave one out estimate
of the expected log pointwise predictive density between
two models were used to compare the models.
Additional considerations
e number and percentage of patients along with
descriptive statistics [i.e., median, interquartile range
(IQR), minimum and maximum] are reported for con-
tinuous non-normally distributed data. Categorical
variables (i.e., gender, age groups, and treatment infor-
mation) have the number and percentage of patients
reported. PCR cycle threshold (Ct) values and UCP-LF
CAA antigen concentrations are displayed and com-
pared using boxplots with descriptive statistics strati-
fied by three POC-CCA categories (i.e., negative, traces,
positive). Proportions of positive test results were cal-
culated as percentages of positive results among all
valid tests, with exact binomial 95% confidence inter-
vals (CIs). Agreement between tests is assessed using
Cohen’s and Fleiss’s kappa statistics, percentage agree-
ments and Prevalence-Adjusted and Bias-Adjusted
Kappa (PABAK). To categorise the diagnostic accuracy
of the laboratory tests, we defined sensitivities and spe-
cificities ≥ 90% as high, as commonly used in clinical
research to ensure a high level of reliability and clinical
utility [56]. For assessing agreement between tests, we
used the following kappa (κ) interpretation scale: κ < 0:
poor agreement; 0 < κ < 0.20: slight; 0.20 < κ < 0.40: f air;
0.40 < κ < 0.60: moderate; 0.60 < κ < 0.80: substantial.
Spearman’s rank correlation coefficients (ρ) with 95%
CIs were calculated to assess the relationships between
diagnostic test results. Correlations were calculated
excluding extreme CAA values (> 1000 pg/ml) and
negative values. Negative UCP-LF CAA concentrations
were imputed by a fixed value of 0.5 × cut-off of 2pg/
Page 6 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
ml to facilitate visualisation in boxplots on a log-scale
[57]. For PCR Ct values, correlations were inversed to
reflect that lower Ct values indicate higher parasite
loads. Confidence intervals were calculated using Fish-
er’s Z-transformation. All correlations were computed
separately for samples from S. haematobium and S.
mansoni endemic regions, including only samples with
positive PCR results and detectable CAA levels (> 2
– < 1000pg/ml).
Sample size
e sample size estimation was informed by two key
considerations: (1) the precision required for diagnos-
tic accuracy parameters and (2) the requirements for
stable Bayesian Latent Class Analysis (BLCA) model
convergence. For diagnostic accuracy estimation, we
targeted a precision of ± 5% for Se and Sp, assuming
expected performances of 85% for S. haematobium
and 90% for S. mansoni tests at an anticipated preva-
lence of 50%. is indicated minimum required sample
sizes of 400 participants for S. haematobium and 550
for S. mansoni sites. For the BLCA modelling, we fol-
lowed recommendations suggesting that stable param-
eter estimation requires larger sample sizes, typically
n > 450 per site when analysing three diagnostic tests
simultaneously [58]. e final achieved sample sizes
satisfied both requirements.
Results
Participants
Of the 1500 participants recruited, 161 were excluded
because of PZQ treatment within 12 months prior to
survey and sampling (n = 126), or because at least one of
the diagnostic tests results was missing (n = 35). A total
of 1339 participants were eligible for analyses, of whom
461 participants were from a S. haematobium (urogeni-
tal) and 878 from a S. mansoni (intestinal) endemic area
(Fig.2). Samples and diagnostic test results from all three
tests under consideration were available for these partici-
pants. In the S. haematobium endemic area, median age
was 28years (IQR: 21 to 40) and 255 (55%) of 461 par-
ticipants were female. In the S. mansoni endemic area,
median age was 38years (IQR: 26 to 50) and 473 (54%) of
878 participants were female (Supplementary Table SI1.1;
for characteristics of the sample).
Diagnostic test results
Detection rates ranged from 52% (POC-CCA) over
60% (UCP-LF CAA) to 66% (PCR) in the S. haemato-
bium endemic area. In the S. mansoni endemic area,
detection rates ranged from 54% (POC-CCA) over 55%
(UCP-LF-CAA) to 59% (PCR) (Fig.3, for details see Sup-
plementary Information I Table SI1.2.).
Test result patterns
Figure 4 shows the distribution of PCR Ct values and
UCP-LF CAA concentrations across POC-CCA catego-
ries in S. haematobium and S. mansoni endemic areas.
In both settings, trace results showed distributions more
similar to negative than positive cases. Based on this
observation, trace results were classified as negative for
subsequent dichotomous analyses.
e relationship between diagnostic test results
showed distinct patterns across endemic settings. In S.
mansoni endemic sites, a clear trend was observed with
higher POC-CCA grades (yellower colors) clustering in
the upper left quadrant, indicating stronger correlation
between lower PCR Ct values (higher parasite loads)
and higher CAA concentrations (Fig. 5). is pattern
was particularly evident for samples with high frequency
(larger dots). In contrast, the S. haematobium endemic
site showed a less pronounced relationship, with more
scattered distribution of POC-CCA grades across the Ct
value range. In both settings, CAA concentrations gen-
erally increased (shown on the log-scale y-axis) as PCR
Ct values decreased, though this inverse correlation
appeared stronger in the S. mansoni endemic sites. e
size of data points reveals that the majority of observa-
tions fell within intermediate frequency ranges (20–59
observations per point).
Moderate positive correlations were observed for
pairwise comparisons of the three diagnostic methods
in the S. haematobium endemic setting, while moder-
ate to strong positive correlations were observed in the
S. mansoni endemic setting (details in Supplementary
Fig. 2 Participant recruitment and selection for diagnostic
accuracy evaluation of schistosomiasis tests in Madagascar. From
1500 initially assessed participants, 1339 were eligible for analysis
after excluding those with recent praziquantel (PZQ) treatment
or missing test results. The final sample comprised 461 participants
from S. haematobium and 878 from S. mansoni endemic areas. PZQ
praziquantel
Page 7 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
Information I). In the S. haematobium endemic region,
the strongest correlation was found between UCP-LF
CAA and PCR (ρ = 0.48, 95% CI: 0.36–0.58), followed by
POC-CCA and PCR (ρ = 0.41, 95% CI: 0.28–0.52), and
POC-CCA and UCP-LF CAA (ρ = 0.40, 95% CI: 0.28–
0.52). In the S. mansoni endemic region, the strongest
correlation was found between POC-CCA and PCR
(ρ = 0.60, 95% CI: 0.53–0.66), followed by POC-CCA and
UCP-LF AG (ρ = 0.52, 95% CI: 0.44–0.59), and UCP-LF
CAA and PCR (ρ = 0.44, 95% CI: 0.35–0.52).
Table2 presents the frequency of diagnostic test result
combinations across three different methods (POC-CCA,
Fig. 3 Proportion of positive test results expressed as percentages by each of the three diagnostic tests among the assessed individuals in S.
haematobium (left) and S. mansoni (right) endemic areas. Lighter shades indicate trace results for POC‑CCA and species‑specific results for PCR. Error
bars represent 95% confidence intervals
Fig. 4 Distribution of PCR Ct values and UCP‑LF CAA concentrations measured in pg/ml displayed by three POC‑CCA categories (i.e., negative,
traces, positive). Subfigures show (a) PCR Ct values by POC‑CCA, S. haematobium endemic, b PCR Ct values by POC‑CCA, S. mansoni endemic,
c UCP‑LF CAA concentrations by POC‑CCA, S. haematobium endemic, d UCP‑LF CAA concentrations by POC‑CCA, S. mansoni endemic. The
colour gradient indicates POC‑CCA G score grading. Ct cycle threshold, PCR polymerase chain reaction, pg/ml picograms per millilitre, POC-CCA
point‑of‑care circulating cathodic antigen, UCP-LF CAA up‑converting reporter particle lateral flow circulating anodic antigen
Page 8 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
UCP-LF-CAA, and PCR) in S. haematobium and S. man-
soni endemic areas. Complete agreement of all three tests
was observed in 41% of cases in the S. haematobium area
(29% all positive, 12% all negative) and 57% in the S. man-
soni area (35% all positive, 22% all negative). e remain-
ing cases showed various patterns of discordant results
between the three diagnostic methods. Further details on
the pairwise and overall agreement are provided in Sup-
plementary Table SI1.2 and SI1.3.
Bayesian latent class analysis: estimated diagnostic
accuracies andprevalence
Figure6 presents Se and Sp estimates from five BLCM
for the three diagnostic tests in S. haematobium and S.
mansoni endemic areas [median values and 95% credible
intervals (CrI)]. In both settings, PCR showed the high-
est sensitivity (95.2%, 95% CrI: 83.1–99.8 for S. haema-
tobium; 95.7%, 95% CrI: 91.0–99.5 for S. mansoni) with
narrow CrI, followed by UCP-LF CAA (87.8%, 95% CrI:
73.2–99.1) and POC-CCA in S. mansoni areas (90.6%,
95% CrI: 85.0–96.3). POC-CCA demonstrated lower sen-
sitivity in S. haematobium (64.5%, 95% CrI: 55.5–74.5)
compared to S. mansoni. Specificity was consistently high
for POC-CCA (81.5%, 95% CrI: 75.9–87.1), while both
PCR and UCP-LF CAA showed moderate specificity
(76.7%, 95% CrI: 70.2–83.2 and 69.7%, 95% CrI: 64.2–75.2
respectively). Notably, the different assumptions about
conditional dependence between tests (Models 0–4)
yielded similar estimates with overlapping credible inter-
vals, suggesting robust results regardless of the depend-
ency structure assumed. Prevalence of S. mansoni, and S.
haematobium were estimated as 49.5% (95% CrI: 43.7–
55.4), and 48.1% (95% CrI: 34.4–67.8), respectively. Note
that these are not population-level estimates, but rather
prevalence estimates among participants recruited at the
PHCCs. Details on Se, Sp, negative and positive predic-
tive values and prevalence can be found in Supplemen-
tary Tables SI2.1–SI2.2 along with model comparisons
(Table SI2.3). Sensitivity analyses using informative pri-
ors for test specificities and alternative trace result inter-
pretation (Supplement 2.3, 2.4) confirmed the robustness
of our findings regarding test performance characteris-
tics. While prevalence estimates were sensitive to trace
result interpretation, particularly in S. mansoni endemic
areas (increasing from ~ 50% to ~ 77% when interpret-
ing traces as positive), the relative performance patterns
of the diagnostic tests remained stable across all model
specifications.
Test performance intheS. haematobium endemic region
For POC-CCA, Se and Sp were low (64.5%, 95% CrI:
55.5–74.5) and Sp (59.6%, 95% CrI: 52.1–68.1 respec-
tively), in line with the caveat that the use of POC-CCA
for detecting S. haematobium infections performs poorly.
Imperfect Se prevented infection rule-out in combination
with an imperfect Sp preventing infection rule-in, leaving
an unacceptably high uncertainty in the validity of results.
UCP-LF CAA and PCR Se were substantial (87.8%, 95%
CrI: 73.2–99.1) and high (95.2%, 95% CrI: 83.1–99.8),
respectively, but imperfect Sp prevented disease rule-in,
Fig. 5 Relationship between diagnostic test results for S. haematobium and S. mansoni infections. Scatter plots show the relationship
between PCR cycle threshold (Ct) values, UCP‑LF‑CAA concentrations (pg/ml), and POC‑CCA grades. UCP‑LF CAA antigen concentrations
that were either imputed by a fixed value of 0.5 × cut‑off of 2 pg/ml or the maximum values of 1000 were excluded from this visualisation to ease
interpretation. Point sizes indicate the frequency of observations (< 20, 20–39, 40–59, ≥ 60 observations). POC‑CCA grades range from 1 (negative)
to 10 (strongly positive), represented by the color gradient. Data are shown separately for samples from S. haematobium endemic regions (left)
and S. mansoni endemic regions (right). Note the logarithmic scale on the y‑axis
Page 9 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
and consequently misclassified a considerable proportion
of non-infected participants as infected.
Test performance intheS. mansoni endemic region
POC-CCA Se and Sp were high (90.6%, 95% CrI:
85.0–96.3) and substantial (81.5%, 95% CrI: 75.9–87.1),
respectively, enabling acceptable infection classifica-
tion. UCP-LF CAA Se was substantial (79.9%, 95% CrI:
75.2–84.6), but imperfect Sp prevented infection rule-in,
and consequently misclassified a considerable proportion
of non-infected participants as infected. PCR Se and Sp
were high (95.7%, 95% CrI: 91.0–99.5) and substantial
(76.7%, 95% CrI: 70.2–83.2), enabling acceptable infec-
tion classification.
Discussion
Producing correct estimates of disease prevalence and
diagnostic test accuracy is challenging without a perfect
reference standard. We used BLCM to estimate multiple
tests’ accuracies for the detection of two species of Schis-
tosoma, along with disease prevalence.
Summary ofndings
e main implication from our study is that no single
laboratory test is universally optimal for detecting Schis-
tosoma infection. Our findings revealed distinct patterns
for S. haematobium and S. mansoni endemic areas. In
an S. haematobium area, for which effective diagnos-
tics other than microscopy are scarce, PCR and UCP-LF
CAA showed good Se but limited Sp, while POC-CCA
performed poorly overall. In contrast, for S. mansoni,
both PCR and POC-CCA demonstrated high diagnostic
accuracy, while UCP-LF CAA showed good sensitivity
but limited specificity. e context-dependency of test
performance extends beyond simple accuracy metrics
to include practical considerations such as laboratory
infrastructure requirements, technical expertise, and
cost-effectiveness. While high sensitivity and specificity
(≥ 90%) might be crucial in low-prevalence or elimina-
tion settings, lower accuracy thresholds could be accept-
able in highly endemic areas focusing on MDA needs.
However, the practical application of PCR in field set-
tings is hampered by its significant infrastructure and
human resource requirements. is study created the
opportunity of strengthening local laboratory and diag-
nostic capacities showing the added value in the imple-
mentation of studies in endemic settings.
Comparison withresults fromtheliterature
Our study extends the existing literature on the perfor-
mance of laboratory tests for detecting schistosome
infections. In the S. mansoni endemic region, our findings
that POC-CCA has high Se and Sp, thereby facilitating
acceptable infection classification, corroborate previous
studies that reported variable performance of POC-CCA
depending on endemicity and reference standards used
[30]. In medium to high endemic areas where parasite
burden is likely higher, POC-CCA has shown promis-
ing results, although persistence of very low reactivity
(trace) in the absence of egg shedding has been observed
[59, 60]. is is consistent with reports that the accuracy
of POC-CCA is more robust in higher endemicity set-
tings, although problems with trace results remain [30].
Our findings regarding UCP-LF CAA performance dif-
fer from previous studies that used composite reference
standards or microscopy as reference, which typically
reported higher specificity [32–34, 61]. However, our
results align with other LCA, suggesting that the choice
of reference standard significantly impacts performance
Table 2 Observed frequency of each response profile with overall percentage agreement shaded in gray
Zeros (0) indicate negative test results and Ones (1) indicate a positive test result. Observed frequency shown corresponds to three test results from 461 participants
from S. haematobium endemic area and 878 participants from S. mansoni endemic area
Response prole Observed frequency % Observed frequency %
POC-CCA UCP-LF-CAA PCR S. haematobium endemic area S. mansoni endemic area
0 0 0 54 12 195 22
1 0 0 44 10 47 5
0 1 0 37 8 85 10
1 1 0 21 5 32 4
0 0 1 48 10 66 8
1 0 1 40 9 88 10
0 1 1 82 18 56 6
1 1 1 135 29 309 35
Page 10 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
estimates [35, 62, 63]. While UCP-LF CAA demonstrated
high sensitivity in our study, the lower-than-expected
specificity warrants further investigation. is finding is
particularly relevant as the UCP-LF CAA test is mainly
used for research purposes and it does not yet exist in a
commercial format. Clinical trials are currently under-
way to evaluate the performances of an RDT format
[64, 65]. PCR demonstrated high Se and Sp, consistent
with its reported high performance in other studies [32,
36]. Importantly, PCR offers the additional advantage of
being able to distinguish between S. mansoni and S. hae-
matobium infections. Whileits application is limited by
infrastructure and resource requirements, recent devel-
opments in molecular diagnostics for schistosomiasis,
including loop-mediated isothermal amplification and
recombinase polymerase amplification techniques, show
promise for more field-applicable solutions [66]. In the
S. haematobium endemic area, our results showed that
POC-CCA had low Se and Sp (~ 60%), confirming the
known caveats of its use for the detection of S. haema-
tobium infections [67]. Furthermore, although UCP-LF
CAA and PCR showed high Se, their imperfect Sp also
compromised their diagnostic accuracy, highlighting the
need for more reliable diagnostic methods for S. haema-
tobium. Our results echo the recommendation of Coelho
etal. to optimise POC-CCA in low worm burden samples
[68], and highlight the ongoing debate on the interpreta-
tion of trace results. As shown by Prada etal. [69], trace
results may still indicate active infection, highlighting
the complexity of accurately diagnosing schistosomiasis
Fig. 6 Diagnostic accuracy of the three tests under investigation based on Bayesian Latent Class Analysis in (a) S. haematobium and (b) S.
mansoni endemic regions. CI conditional independence, CDP conditional dependence in positives, CDN conditional dependence in negatives,
CDPN conditional dependence in positives and negatives, PCR polymerase chain reaction, POC-CCA point‑of‑care circulating cathodic antigen,
UCP-LF-CAA up‑converting reporter particle lateral flow circulating anodic antigen
Page 11 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
post-treatment. e relationship between our diagnostic
tests shows both similarities and differences to previous
findings. e stronger correlations observed in S. man-
soni endemic areas, particularly between POC-CCA and
PCR (ρ = 0.60), align with previous studies that reported
similar relationships between antigen levels and molec-
ular markers. e relatively weaker correlations in S.
haematobium areas (maximum ρ = 0.48) suggest more
complex detection dynamics, possibly due to varying
antigen excretion patterns or differential test sensitivities
at different infection intensities.
Strengths andLimitations
Strengths
First, we estimated the prevalence of two species of Schis-
tosoma; understanding prevalence in a given health-
care setting is critical for planning and policy making
[70–72]. Using BLCM, we have made the best possible
use of data by including results from all available tests to
determine Se and Sp, while accounting for the possibil-
ity of between-test dependence [52, 73, 74]. Second, our
comprehensive dataset includes both qualitative diag-
nostic outcomes and quantitative measures of infection
intensity through PCR Ct values, UCP-LF CAA concen-
trations, and semi-quantitative POC-CCA G-scores. is
multi-dimensional assessment provides deeper insights
into the relationship between test performance and infec-
tion intensity, which is crucial for understanding diagnos-
tic accuracy across different endemicity settings. ird,
data on diagnostic accuracy are often scarce for endemic
populations, which have a different profile of infection
intensity, host immunity, co-infections and environmen-
tal factors than the population in which they are initially
validated. e collection of data from endemic regions
is essential because diagnostic tests need to reflect real-
world conditions in order to provide accurate and effec-
tive disease detection. Our study provides important
insights into the diagnostic accuracy of different tests in
rural Madagascar. Finally, a major strength is the compre-
hensive dataset from a large and diverse sample, which
increases the reliability and applicability of our findings.
Limitations
First, as with any statistical model, it cannot be proven
that the BLCM we fit are the true models. It is worth
emphasizing that estimating diagnostic test accuracy is
different from making a clinical decision. Here we con-
structed a model to estimate test performance and tried
to be transparent about the unknowns, assumptions, and
subjective choices, while other parameterisations are
certainly possible. However, our models were reason-
ably well specified, as evidenced by the good agreement
between observed and expected test result patterns and
the low residual correlation between test results. While
there is no ideal way to validate the results of an LCM
as there is no perfect reference test, we attempted to
validate our findings through sensitivity analyses using
different prior specifications and trace result interpre-
tations.However, due to the lack of additional external
validation datasuch as the proportion of patients with a
particular test pattern who were treated,our validation
options were limited. Second, LCM has been described
as a “black box” [75], and cautions have been raised that
model misspecification is difficult to detect [76]. Cer-
tainly, its mechanisms are less intuitive to understand
than Boolean decision rules as often used in defining
CRS, but the theory underlying LCM is clearly defined
and transparent [54, 77]. We have tried to be clear by
providing a heuristic model that illustrate our assump-
tions about the relationships within the model. While
our BLCM approach using various dependency struc-
tures appropriately captured the diagnostic uncertainty,
other statistical methods could provide complementary
insights. For instance, frequentist latent class models or
Bayesian models with different prior specifications might
offer additional perspectives on the test performance
variations between endemic settings. However, given
the consistent pattern of wider credible intervals for S.
haematobium across our different model specifications,
this uncertainty likely reflects true diagnostic challenges
rather than methodological limitations. In addition, our
diagnostic approach was developed with the input from
a multidisciplinary team of experts to address potential
model misspecification, as traditional goodness-of-fit
metrics alone may not fully capture these issues. A final,
general limitation related to the statistical model is that
the parameter estimates depend on the available data
which is true for any model estimate.
Due to the constraints of our study setting, which only
allowed a single sample to be collected per participant,
we were unable to include microscopy as a reference
standard, despite its usual role as a key diagnostic tool
to validate results. is limitation is relevant as micros-
copy’s sensitivity is known to improve with multiple
samples, and its inclusion could have provided valuable
comparative data. PCR’s ability to identify active infec-
tions may be limited by biomarker persistence, though
we mitigated this by excluding participants with PZQ
treatment within 12months [36]. Clinical variables that
could influence diagnostic outcomes, such as infection
intensity stage and disease stage, were unavailable. Our
quantitative and semi-quantitative measures of infection
intensity (PCR Ct values, UCP-LF CAA concentrations,
and POC-CCA G-scores) have inherent limitations.
POC-CCA is formally recommended only as a qualitative
test with potential operator and batch variability, which
Page 12 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
we addressed through standardisation. e relationship
between our measurements (PCR Ct values, UCP-LF
CAA concentrations, POC-CCA G-scores) and actual
worm burden remains incompletely validated. Finally,
our use of consecutive sampling limits the generalisabil-
ity of our findings to the broader population. While this
sampling method was practical for our study setting, we
recommend future studies employ probabilistic sampling
methods to enable broader population-level inferences.
Conclusions
Achieving the NTD roadmap target [4] to eliminate
schistosomiasis as a public health problem in endemic
countries hinges on the deployment of precise and sensi-
tive diagnostic tests [42]. Our comprehensive evaluation
of diagnostic tests in rural Madagascar, involving a large
sample size across different endemic settings, provides
robust evidence for test performance in real-world con-
ditions.In S. mansoni endemic areas, the high sensitivity
of POC-CCA combined with its field applicability makes
it particularly suitable for initial prevalence assessment
and program monitoring. However, in S. haematobium
settings, the lower test agreement suggests that combin-
ing multiple diagnostic approaches might be necessary.
Our results emphasize that it is essential to define the
use case for each test and recommend its use according
to the epidemiology, the context and the purpose. It also
highlights the need to focus on underserved areas where
data are scarce to generate valuable evidence that could
guide better diagnostic practices and health policies in
similar settings worldwide. ese considerations align
with recent calls for context-specific approaches to schis-
tosomiasis diagnostics and highlight the need for flexible,
evidence-based diagnostic strategies that can adapt to
changing epidemiological situations. Advanced methods
to overcome limitations in the assessment of test accu-
racy should be encouraged.
Abbreviations
BLCM Bayesian latent class model
BNITM Bernhard Nocht Institute for Tropical Medicine
CDP Conditional dependence in positives
CDN Conditional dependence in negatives
CDPN Conditional dependence in positives and negatives
CI Conditional independence
CrI Credible interval
CRS Composite reference standard
Ct Cycle threshold
DALYs Disability‑adjusted life years
DNA Deoxyribonucleic acid
IQR Interquartile range
LCM Latent class model
MDA Mass drug administration
NTDs Neglected tropical diseases
PABAK Prevalence‑adjusted and bias‑adjusted kappa
PCR Polymerase chain reaction
PHCC Primary health care centre
POC‑CCA Point‑of‑care circulating cathodic antigen
PZQ Praziquantel
RDTs Rapid diagnostic tests
Se Sensitivity
Sp Specificity
SSA Sub‑Saharan African
UCP‑LF CAA Up‑converting reporter particle lateral flow circulating anodic
antigen
Supplementary Information
The online version contains supplementary material available at https:// doi.
org/ 10. 1186/ s40249‑ 025‑ 01292‑x.
Supplementary material 1
Acknowledgements
The authors would like to thank the German Center for Infection Research
for the generous funds received that made this research possible. Most
importantly, the study would not be possible without the dedicated research
and implementation staff working on the ground to carry the project to frui‑
tion. We also warmly thank our implementation teams from the drivers to the
nurses carrying out the daily work in the cities. Lastly, we want to thank all the
men, women, children and families in Itasy, Bongolava, Amoron’ i Mania and
Boney who generously gave their time and their effort to participate in SCHIS‑
DIMA. We are grateful to Nandini Dendukuri and Suzanne Keddie for sharing
their methodological expertise about the implementation and interpretation
of Bayesian latent class models.
Author contributions
Daniela Fusco, Norbert Georg Schwarz, Raphael Rakotozandrindrainy, Rako‑
toarivelo Rivo Andry, Tahinamandranto Rasamoelina, Mala Rakoto Andrianari‑
velo and Jürgen May conceived the idea for the study. Daniela Fusco, Norbert
Georg Schwarz and Eva Lorenz conceptualised the design of the research.
Daniela Fusco developed the questionnaire and oversaw data collection and
monitoring. Anna Jaeger implemented the database. Pia Rausche, Jean‑Marc
Kutz and Cheick O Doumbia contributed to data management and data
cleaning. Ravo Razafindrakoto, Zaraniaina Tahiry Rasolojaona, Nantenaina Mat‑
thieu Razafindralava, and Alexandre Zerbo performed the laboratory analysis.
Daniela Fusco, Egbert Tannich, Heidrun von Thien, Claudia de Dood, and Pytsje
T Hoekstra supervised the laboratory procedures. Philipp Klein and Yannick
Höppner supported and monitored field activities. Njary Rakotozandrind‑
rainy supported field activities. Eva Lorenz conceptualised and performed
statistical analyses. Eva Lorenz, Daniela Fusco, Paul LAM Corstjens, Claudia
de Dood, Pytsje T Hoekstra and Govert J van Dam reviewed and discussed
data analysis. Daniela Fusco, Norbert G Schwarz, Mala Rakoto Andrianarivelo,
Egbert Tannich and Jürgen May secured the funding for the study. Eva Lorenz
and Daniela Fusco wrote the first draft of the manuscript. The first draft of
the manuscript was written by Eva Lorenz and Daniela Fusco and all authors
commented on versions of the manuscript. All authors read and approved the
final manuscript.
Funding
Open Access funding enabled and organized by Projekt DEAL. This work was
supported by the German Center for Infection Research (Grant number TI
03.907_00).
Availability of data and materials
The datasets used and/or analysed during the current study are available from
the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The SCHISDIMA study protocol was positively reviewed by Hamburg
Ärztekammer (Reference: PV 7019). The study protocol was furthermore
approved by the Malagasy ethical commission (Reference: 023‑MSANP/
CERBM). Any protocol changes are reported to the respective ethical boards
for approval. Written informed consent was obtained from all individual
Page 13 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
participants (or their parent or legal guardian in the case of children under 18)
included in the study.
Consent for publication
The authors affirm that human research participants provided informed
consent for publication.
Competing interests
The authors have no relevant financial or non‑financial interests to disclose.
Author details
1 Department of Infectious Diseases Epidemiology, Bernhard Nocht Institute
for Tropical Medicine (BNITM), Bernhard‑Nocht‑Strasse 74, 20359 Hamburg,
Germany. 2 German Center for Infection Research (DZIF), Hamburg‑Borstel‑
Lübeck‑Riems, Hamburg, Germany. 3 Institute of Medical Biostatistics, Epidemi‑
ology and Informatics, University Medical Centre of the Johannes Gutenberg
University Mainz, Mainz, Germany. 4 Centre d’Infectiologie Charles Mérieux,
University of Antananarivo, 101 Antananarivo, Madagascar. 5 Research Group:
Implementation Research, Bernhard Nocht Institute for Tropical Medicine
(BNITM), Bernhard‑Nocht‑Strasse 74, 20359 Hamburg, Germany. 6 Depart‑
ment of Cellular Parasitology, Bernhard Nocht Institute for Tropical Medicine
(BNITM), Bernhard‑Nocht‑Strasse 74, 20359 Hamburg, Germany. 7 Department
of Microbiology and Parasitology, University of Antananarivo, 101 Antanana‑
rivo, Madagascar. 8 University Clinical Research Center, University of Sciences,
Techniques and Technologies of Bamako, Bamako, Mali. 9 Department of Cell
and Chemical Biology, Leiden University Medical Center, Albinusdreef 2, 2333
ZA Leiden, the Netherlands. 10 Department of Parasitology, Leiden University
Medical Center, Albinusdreef 2, 2333 ZA Leiden, the Netherlands. 11 National
Reference Centre for Tropical Pathogens, Bernhard Nocht Institute for Tropical
Medicine (BNITM), Hamburg, Germany. 12 Department of Infectious Diseases,
University of Fianarantsoa Andrainjato, 301 Fianarantsoa, Madagascar.
13 Department of Tropical Medicine I, University Medical Center Hamburg‑
Eppendorf (UKE), Hamburg, Germany.
Received: 5 January 2025 Accepted: 4 March 2025
References
1. WHO. Neglected Tropical Diseases. 2018. https:// www. who. int/ health‑
topics/ negle cted‑ tropi cal‑ disea ses. Accessed 06 Jan 2025.
2. Hotez PJ, Alvarado M, Basáñez M‑G, Bolliger I, Bourne R, Boussinesq M,
et al. The global burden of disease study 2010: interpretation and implica‑
tions for the neglected tropical diseases. PLoS Negl Trop Dis. 2014;8(7):
e2865.
3. Choi HL, Ducker C, Braniff S, Argaw D, Solomon AW, Borisch B, et al.
Landscape analysis of NTD diagnostics and considerations on the
development of a strategy for regulatory pathways. PLoS Negl Trop Dis.
2022;16(7): e0010597.
4. WHO. Ending the neglect to attain the Sustainable Development Goals: a
road map for neglected tropical diseases 2021–2030. Licence: CC BY‑NC‑
SA 3.0 IGO. 2020.
5. King CH, Dickman K, Tisch DJ. Reassessment of the cost of chronic
helmintic infection: a meta‑analysis of disability‑related outcomes in
endemic schistosomiasis. Lancet. 2005;365(9470):1561–9.
6. Hay SI, Abajobir AA, Abate KH, Abbafati C, Abbas KM, Abd‑Allah F, et al.
Global, regional, and national disability‑adjusted life‑years (DALYs) for 333
diseases and injuries and healthy life expectancy (HALE) for 195 countries
and territories, 1990–2016: a systematic analysis for the Global Burden of
Disease Study 2016. Lancet. 2017;390(10100):1260–344.
7. WHO. Schistosomiasis 2023. https:// www. who. int/ news‑ room/ fact‑
sheets/ detail/ schis tosom iasis. Accessed 06 Jan 2025.
8. Rollinson D, Knopp S, Levitz S, Stothard JR, Tchuem Tchuenté L‑A, Garba
A, et al. Time to set the agenda for schistosomiasis elimination. Acta Trop.
2013;128(2):423–40.
9. McManus DP, Dunne DW, Sacko M, Utzinger J, Vennervald BJ, Zhou XN.
Schistosomiasis. Nat Rev Dis Prim. 2018;4(1):13.
10. Colley DG, Bustinduy AL, Secor WE, King CH. Human schistosomiasis.
Lancet. 2014;383(9936):2253–64.
11. McManus DP, Dunne DW, Sacko M, Utzinger J, Vennervald BJ, Zhou XN.
Schistosomiasis. Nat Rev Dis Primers. 2018;4(1):13.
12. Shen Y, Wiegand RE, Olsen A, King CH, Kittur N, Binder S, et al. Five‑year
impact of different multi‑year mass drug administration strategies on
childhood Schistosoma mansoni‑Associated Morbidity: a combined
analysis from the schistosomiasis consortium for operational research
and evaluation cohort studies in the Lake Victoria Regions of Kenya and
Tanzania. Am J Trop Med Hyg. 2019;101(6):1336–44.
13. CDC. Schistosomiasis: Epidemiology & Risk Factors 2018. https:// www.
cdc. gov/ paras ites/ schis tosom iasis/ epi. html. Accessed 06 Jan 2025.
14. King CH, Binder S, Shen Y, Whalen CC, Campbell CH, Wiegand RE, et al.
SCORE studies on the impact of drug treatment on morbidity due to
Schistosoma mansoni and Schistosoma haematobium infection. Am J Trop
Med Hyg. 2020;103(1_Suppl):30–5.
15. Doenhoff MJ, Cioli D, Utzinger J. Praziquantel: mechanisms of action,
resistance and new derivatives for schistosomiasis. Curr Opin Infect Dis.
2008;21(6):659–67.
16. Melman SD, Steinauer ML, Cunningham C, Kubatko LS, Mwangi IN,
Wynn NB, et al. Reduced susceptibility to praziquantel among naturally
occurring Kenyan isolates of Schistosoma mansoni. PLoS Negl Trop Dis.
2009;3(8):e504.
17. Utzinger J, N’Goran EK, N’Dri A, Lengeler C, Tanner M. Efficacy of prazi‑
quantel against Schistosoma mansoni with particular consideration for
intensity of infection. Trop Med Int Health. 2000;5(11):771–8.
18. King CH, Kittur N, Binder S, Campbell CH, N’Goran EK, Meite A, et al.
Impact of different mass drug administration strategies for gaining and
sustaining control of Schistosoma mansoni and Schistosoma haemato-
bium infection in Africa. Am J Trop Med Hyg. 2020;103(1_Supple):14–23.
19. GAHI. 2024. https:// www. thisw ormyw orld. org/ maps/ find‑a‑ map.
Accessed 06 Jan 2025.
20. Terer CC, Bustinduy AL, Magtanong RV, Muhoho N, Mungai PL, Muchiri
EM, et al. Evaluation of the health‑related quality of life of children in
Schistosoma haematobium‑endemic communities in Kenya: a cross‑
sectional study. PLoS Negl Trop Dis. 2013;7(3): e2106.
21. Nascimento GL, Domingues ALC, Ximenes RAA, Itria A, Cruz LN, Oliveira
MRF. Quality of life and quality‑adjusted life years of chronic schisto‑
somiasis mansoni patients in Brazil in 2015. Trans R Soc Trop Med Hyg.
2018;112(5):238–44.
22. Martins‑Melo FR, Carneiro M, Ramos AN Jr, Heukelbach J, Ribeiro
ALP, Werneck GL. The burden of neglected tropical diseases in Brazil,
1990–2016: a subnational analysis from the global burden of disease
study 2016. PLoS Negl Trop Dis. 2018;12(6): e0006559.
23. WHO. WHO guideline on control and elimination of human schistosomia‑
sis. Geneva: WHO; 2022.
24. Boisson S, Engels D, Gordon BA, Medlicott KO, Neira MP, Montresor A,
et al. Water, sanitation and hygiene for accelerating and sustaining pro‑
gress on neglected tropical diseases: a new Global Strategy 2015–20. Int
Health. 2016;8(1_Suppl):i19‑21.
25. Worrell CM, Bartoces M, Karanja DM, Ochola EA, Matete DO, Mwinzi PN,
et al. Cost analysis of tests for the detection of Schistosoma mansoni infec‑
tion in children in western Kenya. Am J Trop Med Hyg. 2015;92(6):1233–9.
26. Turner HC, Bettis AA, Dunn JC, Whitton JM, Hollingsworth TD, Fleming
FM, et al. Economic considerations for moving beyond the Kato‑Katz
technique for diagnosing intestinal parasites as we move towards elimi‑
nation. Trends Parasitol. 2017;33(6):435–43.
27. Speich B, Knopp S, Mohammed KA, Khamis IS, Rinaldi L, Cringoli G, et al.
Comparative cost assessment of the Kato‑Katz and FLOTAC techniques
for soil‑transmitted helminth diagnosis in epidemiological surveys. Parasit
Vectors. 2010;3:71.
28. Miswan N, Singham GV, Othman N. Advantages and limitations of
microscopy and molecular detections for diagnosis of soil‑transmitted
helminths: an overview. Helminthologia. 2022;59(4):321–40.
29. Alemu G, Nibret E. Evaluation of the Urine POC‑CCA test accuracy in the
detection of Schistosoma mansoni infection: a systematic review and
meta‑analysis. J Trop Med. 2024;2024(1):5531687.
30. Peralta JM, Cavalcanti MG. Is POC‑CCA a truly reliable test for schistosomi‑
asis diagnosis in low endemic areas? The trace results controversy. PLoS
Negl Trop Dis. 2018;12(11): e0006813.
31. Corstjens P, de Dood CJ, Knopp S, Clements MN, Ortu G, Umulisa I, et al.
Circulating anodic antigen (CAA): a highly sensitive diagnostic biomarker
Page 14 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
to detect active Schistosoma infections‑improvement and use during
SCORE. Am J Trop Med Hyg. 2020;103(1_Suppl):50–7.
32. Vaillant MT, Philippy F, Neven A, Barré J, Bulaev D, Olliaro PL, et al. Diag‑
nostic tests for human Schistosoma mansoni and Schistosoma haemato-
bium infection: a systematic review and meta‑analysis. Lancet Microbe.
2024;5(4):e366–78.
33. Sousa MS, van Dam GJ, Pinheiro MCC, de Dood CJ, Peralta JM, Peralta
RHS, et al. Performance of an ultra‑sensitive assay targeting the circulat‑
ing anodic antigen (CAA) for detection of Schistosoma mansoni infection
in a low endemic area in Brazil. Front Immunol. 2019. https:// doi. org/ 10.
3389/ fimmu. 2019. 00682.
34. Hoekstra PT, Madinga J, Lutumba P, van Grootveld R, Brienen EAT,
Corstjens PLAM, et al. Diagnosis of schistosomiasis without a microscope:
evaluating circulating antigen (CCA, CAA) and DNA detection methods
on banked samples of a community‑based survey from DR Congo. Trop
Med Infect Dis. 2022;7(10):315.
35. Knopp S, Corstjens PLAM, Koukounari A, Cercamondi CI, Ame SM, Ali SM,
et al. Sensitivity and specificity of a urine circulating anodic antigen test
for the diagnosis of Schistosoma haematobium in low endemic settings.
PLoS Negl Trop Dis. 2015;9(5): e0003752.
36. Frickmann H, Lunardon LM, Hahn A, Loderstädt U, Lindner AK, Becker
SL, et al. Evaluation of a duplex real‑time PCR in human serum for
simultaneous detection and differentiation of Schistosoma mansoni and
Schistosoma haematobium infections—cross‑sectional study. Travel Med
Infect Dis. 2021;41:102035.
37. Hoekstra PT, van Dam GJ, van Lieshout L. Context‑specific procedures for
the diagnosis of human schistosomiasis—A mini review. Front Trop Dis.
2021. https:// doi. org/ 10. 3389/ fitd. 2021. 722438.
38. Amoah AS, Hoekstra PT, Casacuberta‑Partal M, Coffeng LE, Corstjens P,
Greco B, et al. Sensitive diagnostic tools and targeted drug administra‑
tion strategies are needed to eliminate schistosomiasis. Lancet Infect Dis.
2020;20(7):e165–72.
39. Madagascar. MdlSPd. Plan Directeur de lutte contre les Maladies Tropi‑
cales Négligées—(MTN): 2016–2020. In French. https:// www. pseau. org/
outils/ biblio/ resume. php?d= 6652. Accessed 06 Jan 2025.
40. ESPEN W. Madagascar: Schistosomiasis. https:// espen. afro. who. int/ count
ries/ madag ascar. Accessed 06 Jan 2025.
41. Deka MA. Predictive risk mapping of schistosomiasis in Madagascar using
ecological niche modeling and precision mapping. Trop Med Infect Dis.
2022;7(2):15.
42. Rasoamanamihaja CF, Rakotoarivelo RA, Edosoa G, Rasamoelina T,
Montresor A, Marchese V, et al. Schistosomiasis elimination in Madagas‑
car: challenges and opportunities for implementing the new WHO guide‑
lines. BMJ Glob Health. 2023;8(8): e012598.
43. Development. USAfI. Global Health: Madagascar. https:// www. usaid. gov/
madag ascar/ global‑ health. Accessed 06 Jan 2025.
44. Africa WROf. The state of the health workforce in the WHO African Region,
2021. Licence: CC BY‑NC‑SA 3.0 IGO. 2021.
45. van Smeden M, Naaktgeboren CA, Reitsma JB, Moons KG, de Groot JA.
Latent class models in diagnostic studies when there is no reference
standard–a systematic review. Am J Epidemiol. 2014;179(4):423–31.
46. Rasoamanamihaja CF, Rahetilahy AM, Ranjatoarivony B, Dhanani N,
Andriamaro L, Andrianarisoa SH, et al. Baseline prevalence and intensity
of schistosomiasis at sentinel sites in Madagascar: informing a national
control strategy. Parasit Vectors. 2016;9(1):50.
47. Casacuberta‑Partal M, Hoekstra PT, Dieuwke K, Lisette L, Govert J. An
innovative and user‑friendly scoring system for standardised quantitative
interpretation of the urine‑based point‑of‑care strip test (POC‑CCA) for
the diagnosis of intestinal schistosomiasis: a proof‑of‑concept study. Acta
Trop. 2019;199:105150.
48. Colley DG, King CH, Kittur N, Ramzy RMR, Secor WE, Fredericks‑James
M, et al. Evaluation, validation, and recognition of the Point‑of‑care
circulating cathodic antigen, urine‑based assay for mapping Schistosoma
mansoni infections. Am J Trop Med Hyg. 2020;103(1_Suppl):42–9.
49. Deelder AM, van Dam GJ, Kornelis D, Fillié YE, van Zeyl RJ. Schistosoma:
analysis of monoclonal antibodies reactive with the circulating antigens
CAA and CCA. Parasitology. 1996;112(Pt 1):21–35.
50. Wichmann D, Panning M, Quack T, Kramme S, Burchard G‑D, Grevelding
C, et al. Diagnosing schistosomiasis by detection of cell‑free parasite DNA
in human plasma. PLoS Negl Trop Dis. 2009;3(4): e422.
51. Gruninger SK, Rasamoelina T, Rakotoarivelo RA, Razafindrakoto AR,
Rasolojaona ZT, Rakotozafy RM, et al. Prevalence and risk distribution of
schistosomiasis among adults in Madagascar: a cross‑sectional study.
Infect Dis Poverty. 2023;12(1):44.
52. Koukounari A, Jamil H, Erosheva E, Shiff C, Moustaki I. Latent class analysis:
insights about design and analysis of schistosomiasis diagnostic studies.
PLoS Negl Trop Dis. 2021;15(2): e0009042.
53. Goodman LA. The Analysis of systems of qualitative variables when some
of the variables are unobservable. Part I‑A modified latent structure
approach. Am J Sociol. 1974;79(5):1179–259.
54. Joseph L, Gyorkos TW, Coupal L. Bayesian estimation of disease preva‑
lence and the parameters of diagnostic tests in the absence of a gold
standard. Am J Epidemiol. 1995;141(3):263–72.
55. Dendukuri N, Joseph L. Bayesian approaches to modeling the con‑
ditional dependence between multiple diagnostic tests. Biometrics.
2001;57(1):158–67.
56. Deeks JJ BP, Leeflang MM, Takwoingi Y. Cochrane Handbook for System‑
atic Reviews of Diagnostic Test Accuracy. Version 2.0. Cochrane; 2023
https:// train ing. cochr ane. org/ handb ook‑ diagn ostic‑ test‑ accur acy/ curre
nt. Accessed 12 July 2023.
57. Polman K, Engels D, Fathers L, Deelder AM, Gryseels B. Day‑to‑day
fluctuation of schistosome circulating antigen levels in serum and urine
of humans infected with Schistosoma mansoni in Burundi. Am J Trop Med
Hyg. 1998;59(1):150–4.
58. Wurpts IC, Geiser C. Is adding more indicators to a latent class analysis
beneficial or detrimental? Results of a monte‑carlo study. Front Psychol.
2014;5:920.
59. Shiff C. Accurate diagnostics for schistosomiasis: a new role for PCR? Rep
Parasitol. 2015. https:// doi. org/ 10. 2147/ RIP. S74319.
60. Oliveira WJ, Magalhães FDC, Elias AMS, de Castro VN, Favero V, Lindholz
CG, et al. Evaluation of diagnostic methods for the detection of intestinal
schistosomiasis in endemic areas with low parasite loads: saline gradient,
Helmintex, Kato‑Katz and rapid urine test. PLoS Negl Trop Dis. 2018;12(2):
e0006232.
61. Assaré RK, Tra‑Bi MI, Coulibaly JT, Corstjens PLAM, Ouattara M, Hürlimann
E, et al. Accuracy of two circulating antigen tests for the diagnosis and
surveillance of Schistosoma mansoni infection in low‑endemicity settings
of Côte d’Ivoire. Am J Trop Medi Hyg. 2021;105(3):677–83.
62. Ruberanziza E, Wittmann U, Mbituyumuremyi A, Mutabazi A, Campbell
CH, Colley DG, et al. Nationwide remapping of Schistosoma mansoni
infection in rwanda using circulating cathodic antigen rapid test: taking
steps toward elimination. Am J Trop Medi Hyg. 2020;103(1):315–24.
63. Clements MN, Corstjens PLAM, Binder S, Campbell CH, de Dood CJ,
Fenwick A, et al. Latent class analysis to evaluate performance of point‑
of‑care CCA for low‑intensity Schistosoma mansoni infections in Burundi.
Parasit Vectors. 2018;11(1):111.
64. GHIT. A schistosomiasis rapid diagnostic test to support control pro‑
grammes in monitoring treatment impact and reassessment mapping
2022. https:// www. ghitf und. org/ inves tment/ portf oliod etail/ detail/ 167/
en. Accessed 06 Jan 2025.
65. FIND. Rapid tests for schistosomiasis control & elimination Foundation
for Innovative New Diagnostics; 2021. https:// www. finddx. org/ what‑
wedo/ proje cts/ rapid‑ tests‑ for‑ schis tosom iasis contr ol‑ and‑ elimi nation/.
Accessed 06 Jan 2025.
66. Mesquita SG, Gadd G, Coelho FS, Cieplinski A, Emery A, Lugli EB, et al.
Laboratory and field validation of the recombinase polymerase amplifica‑
tion assay targeting the Schistosoma mansoni mitochondrial minisatellite
region (SmMIT‑RPA) for snail xenomonitoring for schistosomiasis. Int J
Parasitol. 2024;54(5):247–56.
67. Coulibaly JT, N’Gbesso YK, Knopp S, N’Guessan NA, Silué KD, van Dam GJ,
et al. Accuracy of urine circulating cathodic antigen test for the diagnosis
of Schistosoma mansoni in preschool‑aged children before and after
treatment. PLoS Negl Trop Dis. 2013;7(3): e2109.
68. Coelho PM, Siqueira LM, Grenfell RF, Almeida NB, Katz N, Almeida Á, et al.
Improvement of POC‑CCA interpretation by using lyophilization of urine
from patients with Schistosoma mansoni low worm burden: towards an
elimination of doubts about the concept of trace. PLoS Negl Trop Dis.
2016;10(6): e0004778.
69. Prada JM, Touloupou P, Adriko M, Tukahebwa EM, Lamberton PHL,
Hollingsworth TD. Understanding the relationship between egg‑ and
Page 15 of 15
Lorenzetal. Infectious Diseases of Poverty (2025) 14:20
antigen‑based diagnostics of Schistosoma mansoni infection pre‑ and
post‑treatment in Uganda. Parasit Vectors. 2018;11(1):21.
70. Mazigo HD, Uisso C, Kazyoba P, Mwingira UJ. Primary health care facilities
capacity gaps regarding diagnosis, treatment and knowledge of schisto‑
somiasis among healthcare workers in North‑western Tanzania: a call to
strengthen the horizontal system. BMC Health Serv Res. 2021;21(1):529.
71. Klohe K, Koudou BG, Fenwick A, Fleming F, Garba A, Gouvras A, et al. A
systematic literature review of schistosomiasis in urban and peri‑urban
settings. PLoS Negl Trop Dis. 2021;15(2): e0008995.
72. Janoušková E, Clark J, Kajero O, Alonso S, Lamberton PHL, Betson M, et al.
Public health policy pillars for the sustainable elimination of zoonotic
schistosomiasis. Front Trop Dis. 2022. https:// doi. org/ 10. 3389/ fitd. 2022.
826501.
73. MacLean EL, Kohli M, Köppel L, Schiller I, Sharma SK, Pai M, et al. Bayesian
latent class analysis produced diagnostic accuracy estimates that were
more interpretable than composite reference standards for extrapulmo‑
nary tuberculosis tests. Diagn Progn Res. 2022;6(1):11.
74. Keddie SH, Baerenbold O, Keogh RH, Bradley J. Estimating sensitivity and
specificity of diagnostic tests using latent class models that account for
conditional dependence between tests: a simulation study. BMC Med Res
Methodol. 2023;23(1):58.
75. Pepe MS. The statistical evaluation of medical tests for classification and
prediction. Oxford: Oxford University Press; 2003.
76. Albert PS, Dodd LE. A cautionary note on the robustness of latent class
models for estimating diagnostic error without a gold standard. Biomet‑
rics. 2004;60(2):427–35.
77. Goodman LA. Exploratory latent structure analysis using both identifiable
and unidentifiable models. Biometrika. 1974;61(2):215–31.
