Bayesian Latent Class Models in malaria diagnosis

Latent classification model for censored longitudinal binary outcome

Latent classification model is a class of statistical methods for identifying unobserved class membership among the study samples using some observed data. In this study, we proposed a latent classification model that takes a censored longitudinal binary outcome variable and uses its changing pattern over time to predict individuals' latent class membership. Assuming the time-dependent outcome variables follow a continuous-time Markov chain, the proposed method has two primary goals: (1) estimate the distribution of the latent classes and predict individuals' class membership, and (2) estimate the class-specific transition rates and rate ratios. To assess the model's performance, we conducted a simulation study and verified that our algorithm produces accurate model estimates (ie, small bias) with reasonable confidence intervals (ie, achieving approximately 95% coverage probability). Furthermore, we compared our model to four other existing latent class models and demonstrated that our approach yields higher prediction accuracies for latent classes. We applied our proposed method to analyze the COVID-19 data in Houston, Texas, US collected between January first 2021 and December 31st 2021. Early reports on the COVID-19 pandemic showed that the severity of a SARS-CoV-2 infection tends to vary greatly by cases. We found that while demographic characteristics explain some of the differences in individuals' experience with COVID-19, some unaccounted-for latent variables were associated with the disease.

Evaluation of Kato-Katz and multiplex quantitative polymerase chain reaction performance for clinical helminth infections in Thailand using a latent class analysis

Using an appropriate diagnostic tool is essential to soil-transmitted helminth control and elimination efforts. Kato-Katz (KK) is the most commonly used diagnostic, but recently other tools, such as real-time quantitative polymerase chain reaction (multiplex qPCR), are starting to be employed more. Here, we evaluated the performance of these two diagnostic tools for five helminth species in Thailand. In the absence of a gold standard, diagnostic performance can be evaluated using latent class analysis. Our results suggest that in moderate to high prevalence settings above 2% multiplex qPCR could be more sensitive than KK, this was particularly apparent for Opisthorchis viverrini in the northeastern provinces. However, for low prevalence, both diagnostics suffered from low sensitivity. Specificity of both diagnostics was estimated to be high (above 70%) across all settings. For some specific helminth infection such as O. viverrini, multiplex qPCR is still a preferable choice of diagnostic test. KK performed equally well in detecting Ascaris lumbricoides and Taenia solium when the prevalence is moderate to high (above 2%). Neither test performed well when the prevalence of infection is low (below 2%), and certainly in the case for hookworm and Trichuris trichiura. Combination of two or more diagnostic tests can improve the performance although the cost would be high. Development of new methods for helminth surveillance at the pre-elimination phase is therefore very important. This article is part of the theme issue 'Challenges and opportunities in the fight against neglected tropical diseases: a decade from the London Declaration on NTDs'.

Evaluation of tuberculosis diagnostic test accuracy using Bayesian latent class analysis in the presence of conditional dependence between the diagnostic tests used in a community-based tuberculosis screening study

Diagnostic accuracy studies in pulmonary tuberculosis (PTB) are complicated by the lack of a perfect reference standard. This limitation can be handled using latent class analysis (LCA), assuming independence between diagnostic test results conditional on the true unobserved PTB status. Test results could remain dependent, however, e.g. with diagnostic tests based on a similar biological basis. If ignored, this gives misleading inferences. Our secondary analysis of data collected during the first year (May 2018 -May 2019) of a community-based multi-morbidity screening program conducted in the rural uMkhanyakude district of KwaZulu Natal, South Africa, used Bayesian LCA. Residents of the catchment area, aged ≥15 years and eligible for microbiological testing, were analyzed. Probit regression methods for dependent binary data sequentially regressed each binary test outcome on other observed test results, measured covariates and the true unobserved PTB status. Unknown model parameters were assigned Gaussian priors to evaluate overall PTB prevalence and diagnostic accuracy of 6 tests used to screen for PTB: any TB symptom, radiologist conclusion, Computer Aided Detection for TB version 5 (CAD4TBv5≥53), CAD4TBv6≥53, Xpert Ultra (excluding trace) and culture. Before the application of our proposed model, we evaluated its performance using a previously published childhood pulmonary TB (CPTB) dataset. Standard LCA assuming conditional independence yielded an unrealistic prevalence estimate of 18.6% which was not resolved by accounting for conditional dependence among the true PTB cases only. Allowing, also, for conditional dependence among the true non-PTB cases produced a 1.1% plausible prevalence. After incorporating age, sex, and HIV status in the analysis, we obtained 0.9% (95% CrI: 0.6, 1.3) overall prevalence. Males had higher PTB prevalence compared to females (1.2% vs. 0.8%). Similarly, HIV+ had a higher PTB prevalence compared to HIV- (1.3% vs. 0.8%). The overall sensitivity for Xpert Ultra (excluding trace) and culture were 62.2% (95% CrI: 48.7, 74.4) and 75.9% (95% CrI: 61.9, 89.2), respectively. Any chest X-ray abnormality, CAD4TBv5≥53 and CAD4TBv6≥53 had similar overall sensitivity. Up to 73.3% (95% CrI: 61.4, 83.4) of all true PTB cases did not report TB symptoms. Our flexible modelling approach yields plausible, easy-to-interpret estimates of sensitivity, specificity and PTB prevalence under more realistic assumptions. Failure to fully account for diagnostic test dependence can yield misleading inferences.

The Impact of Emerging Plasmodium knowlesi on Accurate Diagnosis by Light Microscopy: A Systematic Review and Modeling Analysis

The five major Plasmodium spp. that cause human malaria appear similar under light microscopy, which raises the possibility that misdiagnosis could routinely occur in clinical settings. Assessing the extent of misdiagnosis is of particular importance for monitoring P. knowlesi, which cocirculates with the other Plasmodium spp. We performed a systematic review and meta-analysis of studies comparing the performance of microscopy and polymerase chain reaction (PCR) for diagnosing malaria in settings with co-circulation of the five Plasmodium spp. We assessed the extent to which co-circulation of Plasmodium parasites affects diagnostic outcomes. We fit a Bayesian hierarchical latent class model to estimate variation in microscopy sensitivity and specificity measured against PCR as the gold standard. Mean sensitivity of microscopy was low, yet highly variable across Plasmodium spp., ranging from 65.7% (95% confidence interval: 48.1-80.3%) for P. falciparum to 0.525% (95% confidence interval 0.0210-3.11%) for P. ovale. Observed PCR prevalence was positively correlated with estimated microscopic sensitivity and negatively correlated with estimated microscopic specificity, though the strength of the associations varied by species. Our analysis suggests that cocirculation of Plasmodium spp. undermines the accuracy of microscopy. Sensitivity was considerably lower for P. knowlesi, P. malariae, and P. ovale. The negative association between specificity and prevalence imply that less frequently encountered species may be misdiagnosed as more frequently encountered species. Together, these results suggest that the burden of P. knowlesi, P. malariae, and P. ovale may be underappreciated in a clinical setting.

A Bayesian latent class model for predicting gestational age in health administrative data

Health administrative data are oftentimes of limited use in epidemiological study on drug safety in pregnancy, due to lacking information on gestational age at birth (GAB). Although several studies have proposed algorithms to estimate GAB using claims database, failing to incorporate the unique distributional shape of GAB, can introduce bias in estimates and subsequent modeling. Hence, we develop a Bayesian latent class model to predict GAB. The model employs a mixture of Gaussian distributions with linear covariates within each class. This approach allows modeling heterogeneity in the population by identifying latent subgroups and estimating class-specific regression coefficients. We fit this model in a Bayesian framework conducting posterior computation with Markov Chain Monte Carlo methods. The method is illustrated with a dataset of 10,043 Rhode Island Medicaid mother-child pairs. We found that the three-class and six-class mixture specifications maximized prediction accuracy. Based on our results, Medicaid women were partitioned into three classes, featured by extreme preterm or preterm birth, preterm or" early" term birth, and" late" term birth. Obstetrical complications appeared to pose a significant influence on class-membership. Altogether, compared to traditional linear models our approach shows an advantage in predictive accuracy, because of superior flexibility in modeling a skewed response and population heterogeneity.

Leishmania exposure in dogs from two endemic countries from New and Old Worlds (Brazil and Portugal): evaluation of three serological tests using Bayesian Latent Class Models

BACKGROUND

Zoonotic leishmaniosis caused by Leishmania infantum is endemic in several countries of the Mediterranean Basin, Latin America, and Asia. Dogs are the main hosts and reservoirs of human infection. Thus, from a One Health perspective, early diagnosis of Leishmania infection in dogs is essential to control the dissemination of the parasite among other dogs and to humans. The aim of this study was to estimate the diagnosis accuracy of three serological tests to detect antibodies to Leishmania in dogs from two endemic settings using Bayesian latent class models (BLCM).

METHODS

A total of 378 dogs from two Portuguese and Brazilian endemic areas of leishmaniosis (194 animals from Portugal and 184 from Brazil) were screened. Detection of anti-Leishmania antibodies was performed using two commercial ELISA (L. infantum IgG-ELISA® and EIE-LVC®) and a rapid immunochromatographic test (DPP-LVC®). Bayesian latent class models were used to estimate Leishmania infection prevalence, together with sensitivities and specificities of the three diagnostic tests, in the two dog populations simultaneously. Predictive values were also calculated. Credibility intervals (CI) were obtained, considering different types of prior information.

RESULTS

A posterior median Leishmania seroprevalence of 13.4% (95% CI 9.0-18.7) and of 21.6% (15.0-28.3) was estimated to the Portuguese and Brazilian dog subpopulations, respectively. The Bayesian analysis indicated that all tests were highly specific (specificity above 90%), and that the DPP-LVC® was more sensitive (96.6%; 83.1-99.9) than both ELISAs in the Portuguese subpopulation, while in the Brazilian subpopulation, EIE-LVC® and L. infantum IgG-ELISA®, had the highest sensitivity (88.2%; 73.7-97.0) and specificity (98.7%; 95.1-99.9), respectively.

CONCLUSIONS

In general, the levels of diagnosis accuracy of the three serological tests to detect Leishmania antibodies assessed by BLCM indicate their utility in canine epidemiological studies. The same approach should be used to assess the performance of these techniques in the clinical management of infected and sick dogs using representative samples from the wide spectrum of clinical situations, namely from subclinical infection to manifest disease. The low positive predictive value of the serological tests used in the current protocol of the Brazilian Ministry of Health suggests that they should not be used individually and may not be sufficient to target reservoir-based control interventions.

Accuracy of self-reported HIV-testing history and awareness of HIV-positive status in four sub-Saharan African countries

BACKGROUND

In many countries in sub-Saharan Africa, self-reported HIV testing history and awareness of HIV-positive status from household surveys are used to estimate the percentage of people living with HIV (PLHIV) who know their HIV status. Despite widespread use, there is limited empirical information on the sensitivity of those self-reports, which can be affected by nondisclosure.

METHODS

Bayesian latent class models were used to estimate the sensitivity of self-reported HIV-testing history and awareness of HIV-positive status in four Population-based HIV Impact Assessment surveys in Eswatini, Malawi, Tanzania, and Zambia. Antiretroviral (ARV) metabolite biomarkers were used to identify persons on treatment who did not accurately report their status. For those without ARV biomarkers, we used a pooled estimate of nondisclosure among untreated persons that was 1.48 higher than those on treatment.

RESULTS

Among PLHIV, the model-estimated sensitivity of self-reported HIV-testing history ranged from 96% to 99% across surveys. The model-estimated sensitivity of self-reported awareness of HIV status varied from 91% to 97%. Nondisclosure was generally higher among men and those aged 15-24 years. Adjustments for imperfect sensitivity did not substantially influence estimates of PLHIV ever tested (difference <4%) but the proportion of PLHIV aware of their HIV-positive status was higher than the unadjusted proportion (difference <8%).

CONCLUSION

Self-reported HIV-testing histories in four Eastern and Southern African countries are generally robust although adjustment for nondisclosure increases estimated awareness of status. These findings can contribute to further refinements in methods for monitoring progress along the HIV testing and treatment cascade.

Reevaluating the true diagnostic accuracy of dipstick tests to diagnose urinary tract infection using Bayesian latent class analysis

OBJECTIVE

Previous studies on diagnostic accuracy of dipstick testing for leukocyte esterase (LE) and nitrite to diagnose urinary tract infection (UTI) had used urine culture, which is an imperfect gold standard. Estimates of diagnostic accuracy obtained using the classical gold standard framework might not reflect the true diagnostic accuracy of dipstick tests.

METHODS

We used the dataset from a prospective, observational study conducted in the emergency department of a teaching hospital in southern India. Patients with a clinical suspicion of UTI underwent dipstick testing for LE and nitrite, urine microscopy, and urine culture. Based on the results of urine microscopy and culture, UTI was classified into definite, probable, and possible. Patients with microscopic pyuria and a positive urine culture were adjudicated as definite UTI. Unequivocal imaging evidence of emphysematous pyelonephritis or perinephric collections was also considered definite UTI. We estimated the diagnostic accuracy of LE and nitrite tests using the classical analysis (assuming definite UTI as gold standard) and two different Bayesian latent class models (LCMs; 3-tests in 1-population and 2-tests in 2-populations models).

RESULTS

We studied 149 patients. Overall, 64 (43%) patients had definite, 76 (51%) had probable, and 2 (1.3%) had possible UTI; 7 (4.6%) had alternate diagnoses. In classical analysis, LE was more sensitive than nitrite (87.5% versus 70.5%), while nitrite was more specific (24% versus 58%). The 3-tests in 1-population Bayesian LCM indicated a substantially better sensitivity and specificity for LE (98.1% and 47.6%) and nitrite (88.2% and 97.7%). True sensitivity and specificity of urine culture as estimated by the model was 48.7% and 73.0%. Estimates of the 2-tests in 2-populations model were in agreement with the 3-tests in 1-population model.

CONCLUSIONS

Bayesian LCMs indicate a clinically important improvement in the true diagnostic accuracy of urine dipstick testing for LE and nitrite. Given this, a negative dipstick LE would rule-out UTI, while a positive dipstick nitrite would rule-in UTI in our study setting. True diagnostic accuracy of urine dipstick testing for UTI in various practice settings needs reevaluation using Bayesian LCMs.

Evaluation of LAMP for detection of Shigella from stool samples in children

Background

To assess the diagnostic accuracy of loop-mediated isothermal amplification (LAMP) for the detection of Shigella from stool samples from children.

Methods

Consecutive stool samples from children aged <13 years old who presented with acute watery diarrhoea or dysentery to the Department of Paediatrics were collected and processed in the Department of Microbiology. All the stool samples were subjected to culture, conventional PCR and LAMP. Genomic sequencing was performed for samples that were positive by LAMP but negative by both culture and conventional PCR. The LAMP results were compared to those from culture and to a composite reference standard based on culture and conventional PCR.

Results

Amongst the 374 stool samples tested, 291 samples were positive by LAMP and 213 were positive by the composite reference standard. The sensitivity of LAMP was 100 % (98.3–100 %) and its specificity was 51.6 % (43.6–59.5 %) with a disease prevalence of 57 %. The sensitivity and specificity of LAMP improved to 99.3 % (94.2–100) and 98.2 % (94.5–99.9), respectively, using latent class analysis, while assuming that genomic sequencing has perfect specificity.

Discussion

The authors have standardized the LAMP procedure for direct application to clinical stool samples. LAMP is a sensitive and specific method for the diagnosis of Shigella from stool samples of children as compared to both culture and conventional PCR.

Intestinal schistosomiasis: Can a urine sample decide the infection?

Intestinal schistosomiasis, one of the neglected tropical diseases whose control depends on accurate diagnosis of the disease prevalence. The use of low sensitive Kato Katz (KK) fecal egg detection method as a reference gold standard is not an accurate indication especially in low transmission areas. Latent class analysis frameworks especially the Bayesian could be used instead to compare between different diagnostic tests without the use of a gold standard method as a reference. Thus, this study compared two urine-based tests for the detection of circulating antigen and cell free DNA of Schistosoma mansoni versus KK method using the Bayesian latent class analytical framework and in two models where the trace results of point of contact - assay of circulating cathodic antigen (POC-CCA) were once estimated as positive, and as negative in the other model. The Bayesian framework in the trace CCA positive model showed an estimate of disease prevalence of 26% (95% BCI:0 to 60%). POC-CCA showed the highest sensitivity (74% with BCI: 9 to 91%) and lowest specificity for (20% with BCI: 0% to 37%) and the reverse for KK. For POC-CCA with traces considered negative, it was found that results between the three tests were moderated where the positivity for infection by Schistosoma antigen detection and PCR for cell free DNA approached that estimated by the Bayesian framework (44%), and the specificity for point of contact assay(81%; 95%BCI: 59% to 100%) rose in hand with its sensitivity(77%, 95% BCI:53% to 100%) and with results for PCR test (sensitivity = 80%; 95% BCI: 61% to 100%, specificity = 69%; 95% BIC: 47% to 100%). KK remains with the highest specificity while its sensitivity in the two models never exceeded 22%. Thus, we conclude that the use of a single urine sample could be very sensitive and highly specific in the diagnosis of intestinal schistosomiasis using either the trace negative model of point of contact assay, or conventional PCR, when compared to the fecal egg detection using duplicate KK. However, the use of a single tool restricts the management of the disease in areas of low endemicity.

Common Medical and Statistical Problems: The Dilemma of the Sample Size Calculation for Sensitivity and Specificity Estimation

Sample size calculation in biomedical practice is typically based on the problematic Wald method for a binomial proportion, with potentially dangerous consequences. This work highlights the need of incorporating the concept of conditional probability in sample size determination to avoid reduced sample sizes that lead to inadequate confidence intervals. Therefore, new definitions are proposed for coverage probability and expected length of confidence intervals for conditional probabilities, like sensitivity and specificity. The new definitions were used to assess seven confidence interval estimation methods. In order to determine the sample size, two procedures—an optimal one, based on the new definitions, and an approximation—were developed for each estimation method. Our findings confirm the similarity of the approximated sample sizes to the optimal ones. R code is provided to disseminate these methodological advances and translate them into biomedical practice.

Estimating prevalence and test accuracy in disease ecology: How Bayesian latent class analysis can boost or bias imperfect test results

Obtaining accurate estimates of disease prevalence is crucial for the monitoring and management of wildlife populations but can be difficult if different diagnostic tests yield conflicting results and if the accuracy of each diagnostic test is unknown. Bayesian latent class analysis (BLCA) modeling offers a potential solution, providing estimates of prevalence levels and diagnostic test accuracy under the realistic assumption that no diagnostic test is perfect.In typical applications of this approach, the specificity of one test is fixed at or close to 100%, allowing the model to simultaneously estimate the sensitivity and specificity of all other tests, in addition to infection prevalence. In wildlife systems, a test with near-perfect specificity is not always available, so we simulated data to investigate how decreasing this fixed specificity value affects the accuracy of model estimates.We used simulations to explore how the trade-off between diagnostic test specificity and sensitivity impacts prevalence estimates and found that directional biases depend on pathogen prevalence. Both the precision and accuracy of results depend on the sample size, the diagnostic tests used, and the true infection prevalence, so these factors should be considered when applying BLCA to estimate disease prevalence and diagnostic test accuracy in wildlife systems. A wildlife disease case study, focusing on leptospirosis in California sea lions, demonstrated the potential for Bayesian latent class methods to provide reliable estimates under real-world conditions.We delineate conditions under which BLCA improves upon the results from a single diagnostic across a range of prevalence levels and sample sizes, demonstrating when this method is preferable for disease ecologists working in a wide variety of pathogen systems.

Bayesian evaluation of the performance of three diagnostic tests for Plasmodium falciparum infection in a low-transmission setting in Kilifi County, Kenya

Background: Central to the successful elimination of Plasmodium falciparum malaria, are tests with superior capability of diagnosing low-density parasitaemias. Empirical evidence on the performance of the commonly available diagnostics (light microscopy (LM), rapid diagnostic tests (RDT) and polymerase chain reaction (PCR)) is needed to better inform case management and surveillance activities within primary health care settings where elimination of falciparum malaria is targeted. The objective of this study was to estimate the sensitivity (Se) and specificity (Sp) and predictive values of LM, RDT and PCR tests for P. falciparum infection in children, while evaluating the effect of specific covariates on the accuracy of the tests. Methods: The study enrolled 1,563 children presenting with fever (axillary temperature ≥ 37.5 0C) to the Ngerenya dispensary, Kilifi County between March and December 2014. A Bayesian latent class model (BLCM) was fitted to the participants' diagnostic data obtained from blood samples that were screened for the presence of P. falciparum using the three tests. Results: The PCR assay registered a higher Se (97.6% [92.0; 99.7]) than LM (84.0% [74.8; 91.0]) but similar to RDT (92.2% [84.4; 97.0]). However, the assay showed a similar Sp (98.9% [98.2; 99.4]) to both RDT (99.4% [98.9; 99.7]) and LM (99.5% [99.0; 99.8]). Regarding predictive values, the tests yielded statistically similar estimates of positive and negative predictive values (PPV and NPV). A serial interpretation of the results of RDT and LM raised the PPVs and NPVs to >98%. Conclusions: LM and RDT afford high Se and Sp in symptomatic care-seeking children in this low P. falciparum prevalence setting. A serial combination of the tests assures high PPV and NPV estimates. These elements, coupled with the wide deployment and affordability of the tests, lend the tests useful for guiding clinical care and surveillance activities for P. falciparum within elimination settings.

Bayesian evaluation of the performance of three diagnostic tests for Plasmodium falciparum infection in a low-transmission setting in Kilifi County, Kenya

Background: Central to the successful elimination of Plasmodium falciparum malaria, are tests with superior capability of diagnosing low-density parasitaemias. Empirical evidence on the performance of the commonly available diagnostics (light microscopy (LM), rapid diagnostic tests (RDT) and polymerase chain reaction (PCR)) is needed to better inform case management and surveillance activities within primary health care settings where elimination of falciparum malaria is targeted. The objective of this study was to estimate the sensitivity (Se) and specificity (Sp) and predictive values of LM, RDT and PCR tests for P. falciparum infection in children, while evaluating the effect of specific covariates on the accuracy of the tests. Methods: The study enrolled 1,563 children presenting with fever (axillary temperature ≥ 37.50C) to the Ngerenya dispensary, Kilifi County between March and December 2014. A Bayesian latent class model (BLCM) was fitted to the participants' diagnostic data obtained from blood samples that were screened for the presence of P. falciparum using the three tests. Results: The PCR assay registered a higher Se (97.6% [92.0; 99.7]) than LM (84.0% [74.8; 91.0]) but similar to RDT (92.2% [84.4; 97.0]). However, the assay showed a similar Sp (98.9% [98.2; 99.4]) to both RDT (99.4% [98.9; 99.7]) and LM (99.5% [99.0; 99.8]). Regarding predictive values, the tests yielded statistically similar estimates of positive and negative predictive values (PPV and NPV). A serial interpretation of the results of RDT and LM raised the PPVs and NPVs to >98%. Conclusions: LM and RDT afford high Se and Sp in a low P. falciparum prevalence setting. A serial combination of the tests assures high PPV and NPV estimates. These elements, coupled with the wide deployment and affordability of the tests, lend the tests useful for guiding clinical care and surveillance activities for P. falciparum within elimination settings.

Risk prediction for severe disease and better diagnostic accuracy in early dengue infection; the Colombo dengue study

Background

A major challenge in dengue management in resource limited settings is the confirmation of diagnosis. Clinical features of dengue often overlap with other infections and molecular diagnostic tools are not readily accessible to clinicians at hospitals. In addition, the prediction of plasma leakage in dengue is also difficult. Hematocrit level and ultrasound scans (combined with clinical parameters) are helpful to detect plasma leakage once it has happened, not before.

Methods

Colombo Dengue Study (CDS) is a prospective cohort study of clinically suspected adult dengue patients recruited from the National hospital of Sri Lanka (within the first 3 days of fever) that aimed to a) identify clinical and basic laboratory test parameters to differentiate dengue from non-dengue fever, b) evaluate the comparative efficacy of loop-mediated isothermal amplification (LAMP) for dengue diagnosis (vs. NS1 antigen test and RT-qPCR) and c) identify early associations that are predictive of plasma leakage or severe dengue. The basic laboratory tests considered here included hematological parameters, serum biochemistry and inflammatory markers.

Results

Only 70% of clinically suspected patients were confirmed as having dengue by either the NS1 antigen test or RT-qPCR. On a Bayesian latent class model which assumes no “gold standard”, LAMP performed equally or better than RT-qPCR and NS1 antigen test respectively. When confirmed dengue patients were compared with others, the earlier group had significantly lower lymphocyte counts and higher aspartate aminotransferase levels (AST) within the first 3 days of fever. Confirmed dengue patients with plasma leakage had a lower mean age and a higher median baseline AST level compared to those without plasma leakage (p < 0.05).

Conclusion

Clinical suspicion overestimates the true number of dengue patients. RT-LAMP is a potentially useful low-cost diagnostic tool for dengue diagnosis. Confirmed dengue patients had significantly higher AST levels and lower lymphocyte counts in early disease compared to others. In confirmed dengue patients, younger age and a higher AST level in early infection were associated with subsequent plasma leakage.

Electronic supplementary material

The online version of this article (10.1186/s12879-019-4304-9) contains supplementary material, which is available to authorized users.

Bayesian evaluation of the performance of three diagnostic tests for Plasmodium falciparum infection in a low-transmission setting in Kilifi County, Kenya

Background: Central to the successful elimination of Plasmodium falciparum malaria, are tests with superior capability of diagnosing low-density parasitaemias. Empirical evidence on the performance of the commonly available diagnostics (light microscopy (LM), rapid diagnostic tests (RDT) and polymerase chain reaction (PCR)) is needed to better inform case management and surveillance activities within primary health care settings where elimination of falciparum malaria is targeted. The objective of this study was to estimate the sensitivity (Se) and specificity (Sp) and predictive values of LM, RDT and PCR tests for P. falciparum infection, while evaluating the effect of specific covariates on the accuracy of the tests. Methods: The study enrolled 1,563 children via a cross-sectional survey for asymptomatic malaria and those presenting with symptomatic malaria to the Ngerenya dispensary, Kilifi County between March and December 2014. A Bayesian latent class model (BLCM) was fitted to the participants' diagnostic data obtained from blood samples that were screened for the presence of P. falciparum using the three tests. Results: The PCR assay registered a higher Se (97.6% [92.0; 99.7]) than LM (84.0% [74.8; 91.0]) but similar to RDT (92.2% [84.4; 97.0]). However, the assay showed a similar Sp (98.9% [98.2; 99.4]) to both RDT (99.4% [98.9; 99.7]) and LM (99.5% [99.0; 99.8]). Regarding predictive values, the tests yielded statistically similar estimates of Positive and negative predictive values (PPV and NPV). A serial interpretation of the results of RDT and LM raised the PPVs and NPVs to >98%. Conclusions: LM and RDT tests afford high Se and Sp in a low P. falciparum prevalence setting. A serial combination of the tests assures high PPV and NPV estimates. These elements, coupled with the wide deployment and affordability of the tests, lend the tests useful for guiding clinical care and surveillance activities for P. falciparum within elimination settings.

Prior Precision, Prior Accuracy, and the Estimation of Disease Prevalence Using Imperfect Diagnostic Tests

Estimates of disease prevalence in any host population are complicated by uncertainty in the outcome of diagnostic tests on individuals. In the absence of gold standard diagnostics (tests that give neither false positives nor false negatives), Bayesian latent class inference can be applied to batteries of diagnostic tests, providing posterior estimates of the sensitivity and specificity of each test, alongside posterior estimates of disease prevalence. Here we explore the influence of precision and accuracy of prior information on the precision and accuracy of posterior estimates of these key parameters. Our simulations use three diagnostic tests, yielding eight possible diagnostic outcomes per individual. Seven degrees of freedom allow the estimation of seven parameters: sensitivity and specificity of each test, and disease prevalence. We show that prior precision begets posterior precision but only when priors are accurate. We also show that analyses without gold standard can use imprecise priors as long as they are initialised with accuracy. Imprecise priors risk the divergence of MCMC chains towards inaccurate posterior estimates, if inaccurate initial values are used. We note that inaccurate priors can yield inaccurate and imprecise inference. Bounded priors should certainly not be used unless their accuracy is well established. Inaccurate estimates of sensitivity or specificity can yield wildly inaccurate estimates of disease prevalence. Our analyses are motivated by studies of bovine tuberculosis in a wild badger population.

Diagnostic accuracy of blood sucrose as a screening test for equine gastric ulcer syndrome (EGUS) in weanling foals

Background

Equine gastric ulcer syndrome is an important cause of morbidity in weanling foals. Many foals are asymptomatic, and the development of an inexpensive screening test to ensure an early diagnosis is desirable. The objective of this study was to determine the diagnostic accuracy of blood sucrose for diagnosis of EGUS in weanling foals.

Results

45 foals were studied 7 days before and 14 days after weaning. The diagnostic accuracy of blood sucrose for diagnosis of gastric lesions (GL); glandular lesions (GDL); squamous lesions (SQL) and clinically significant gastric lesions (CSL) at 45 and 90 min after administration of 1 g/kg of sucrose via nasogastric intubation was assessed using ROC curves and calculating the AUC. For each lesion type, sucrose concentration in blood was compared to gastroscopy; and sensitivities (Se) and specificities (Sp) were calculated across a range of sucrose concentrations. Cut-off values were selected manually to optimize Se. Because of concerns over the validity of the gold standard, additional Se, Sp, and lesion prevalence data were subsequently estimated and compared using Bayesian latent class analysis. Using the frequentist approach, the prevalence of GL; GDL; SQL and CSL before weaning was 21; 9; 7 and 8% respectively; and increased to 98; 59; 97 and 82% respectively after weaning. At the selected cut-off, Se ranged from 84 to 95% and Sp ranged from 47 to 71%, depending upon the lesion type and time of sampling. In comparison, estimates of Se and Sp were consistently higher when using a Bayesian approach, with Se ranging from 81 to 97%; and Sp ranging from 77 to 97%, depending upon the lesion type and time of sampling.

Conclusions

Blood sucrose is a sensitive test for detecting EGUS in weanling foals. Due to its poor specificity, it is not expected that the sucrose blood test will replace gastroscopy, however it may represent a clinically useful screening test to identify foals that may benefit from gastroscopy. Bayesian latent class analysis represents an alternative method to evaluate the diagnostic accuracy of the blood sucrose test in an attempt to avoid bias associated with the assumption that gastroscopy is a perfect test.

Latent class analysis to evaluate performance of point-of-care CCA for low-intensity Schistosoma mansoni infections in Burundi

BACKGROUND

Kato-Katz examination of stool smears is the field-standard method for detecting Schistosoma mansoni infection. However, Kato-Katz misses many active infections, especially of light intensity. Point-of-care circulating cathodic antigen (CCA) is an alternative field diagnostic that is more sensitive than Kato-Katz when intensity is low, but interpretation of CCA-trace results is unclear. To evaluate trace results, we tested urine and stool specimens from 398 pupils from eight schools in Burundi using four approaches: two in Burundi and two in a laboratory in Leiden, the Netherlands. In Burundi, we used Kato-Katz and point-of-care CCA (CCAB). In Leiden, we repeated the CCA (CCAL) and also used Up-Converting Phosphor Circulating Anodic Antigen (CAA).

METHODS

We applied Bayesian latent class analyses (LCA), first considering CCA traces as negative and then as positive. We used the LCA output to estimate validity of the prevalence estimates of each test in comparison to the population-level infection prevalence and estimated the proportion of trace results that were likely true positives.

RESULTS

Kato-Katz yielded the lowest prevalence (6.8%), and CCAB with trace considered positive yielded the highest (53.5%). There were many more trace results recorded by CCA in Burundi (32.4%) than in Leiden (2.3%). Estimated prevalence with CAA was 46.5%. LCA indicated that Kato-Katz had the lowest sensitivity: 15.9% [Bayesian Credible Interval (BCI): 9.2-23.5%] with CCA-trace considered negative and 15.0% with trace as positive (BCI: 9.6-21.4%), implying that Kato-Katz missed approximately 85% of infections. CCAB underestimated disease prevalence when trace was considered negative and overestimated disease prevalence when trace was considered positive, by approximately 12 percentage points each way, and CAA overestimated prevalence in both models. Our results suggest that approximately 52.2% (BCI: 37.8-5.8%) of the CCAB trace readings were true infections.

CONCLUSIONS

Whether measured in the laboratory or the field, CCA outperformed Kato-Katz at the low infection intensities in Burundi. CCA with trace as negative likely missed many infections, whereas CCA with trace as positive overestimated prevalence. In the absence of a field-friendly gold standard diagnostic, the use of a variety of diagnostics with differing properties will become increasingly important as programs move towards elimination of schistosomiasis. It is clear that CCA is a valuable tool for the detection and mapping of S. mansoni infection in the field and CAA may be a valuable field tool in the future.

True malaria prevalence in children under five: Bayesian estimation using data of malaria household surveys from three sub-Saharan countries

BACKGROUND

Malaria is one of the major causes of childhood death in sub-Saharan countries. A reliable estimation of malaria prevalence is important to guide and monitor progress toward control and elimination. The aim of the study was to estimate the true prevalence of malaria in children under five in the Democratic Republic of the Congo, Uganda and Kenya, using a Bayesian modelling framework that combined in a novel way malaria data from national household surveys with external information about the sensitivity and specificity of the malaria diagnostic methods used in those surveys-i.e., rapid diagnostic tests and light microscopy.

METHODS

Data were used from the Demographic and Health Surveys (DHS) and Malaria Indicator Surveys (MIS) conducted in the Democratic Republic of the Congo (DHS 2013-2014), Uganda (MIS 2014-2015) and Kenya (MIS 2015), where information on infection status using rapid diagnostic tests and/or light microscopy was available for 13,573 children. True prevalence was estimated using a Bayesian model that accounted for the conditional dependence between the two diagnostic methods, and the uncertainty of their sensitivities and specificities obtained from expert opinion.

RESULTS

The estimated true malaria prevalence was 20% (95% uncertainty interval [UI] 17%-23%) in the Democratic Republic of the Congo, 22% (95% UI 9-32%) in Uganda and 1% (95% UI 0-3%) in Kenya. According to the model estimations, rapid diagnostic tests had a satisfactory sensitivity and specificity, and light microscopy had a variable sensitivity, but a satisfactory specificity. Adding reported history of fever in the previous 14 days as a third diagnostic method to the model did not affect model estimates, highlighting the poor performance of this indicator as a malaria diagnostic.

CONCLUSIONS

In the absence of a gold standard test, Bayesian models can assist in the optimal estimation of the malaria burden, using individual results from several tests and expert opinion about the performance of those tests.

Interpreting ambiguous 'trace' results in Schistosoma mansoni CCA Tests: Estimating sensitivity and specificity of ambiguous results with no gold standard

BACKGROUND

The development of new diagnostics is an important tool in the fight against disease. Latent Class Analysis (LCA) is used to estimate the sensitivity and specificity of tests in the absence of a gold standard. The main field diagnostic for Schistosoma mansoni infection, Kato-Katz (KK), is not very sensitive at low infection intensities. A point-of-care circulating cathodic antigen (CCA) test has been shown to be more sensitive than KK. However, CCA can return an ambiguous 'trace' result between 'positive' and 'negative', and much debate has focused on interpretation of traces results.

METHODOLOGY/PRINCIPLE FINDINGS

We show how LCA can be extended to include ambiguous trace results and analyse S. mansoni studies from both Côte d'Ivoire (CdI) and Uganda. We compare the diagnostic performance of KK and CCA and the observed results by each test to the estimated infection prevalence in the population. Prevalence by KK was higher in CdI (13.4%) than in Uganda (6.1%), but prevalence by CCA was similar between countries, both when trace was assumed to be negative (CCAtn: 11.7% in CdI and 9.7% in Uganda) and positive (CCAtp: 20.1% in CdI and 22.5% in Uganda). The estimated sensitivity of CCA was more consistent between countries than the estimated sensitivity of KK, and estimated infection prevalence did not significantly differ between CdI (20.5%) and Uganda (19.1%). The prevalence by CCA with trace as positive did not differ significantly from estimates of infection prevalence in either country, whereas both KK and CCA with trace as negative significantly underestimated infection prevalence in both countries.

CONCLUSIONS

Incorporation of ambiguous results into an LCA enables the effect of different treatment thresholds to be directly assessed and is applicable in many fields. Our results showed that CCA with trace as positive most accurately estimated infection prevalence.

Malaria and Chikungunya Detected Using Molecular Diagnostics Among Febrile Kenyan Children

BACKGROUND

In sub-Saharan Africa, malaria is frequently overdiagnosed as the cause of an undifferentiated febrile illness, whereas arboviral illnesses are presumed to be underdiagnosed.

METHODS

Sera from 385 febrile Kenyan children, who presented to 1 of 4 clinical sites, were tested using microscopy and real-time molecular assays for dengue virus (DENV), chikungunya virus (CHIKV), malaria, and Leptospira.

RESULTS

Malaria was the primary clinical diagnosis for 254 patients, and an arboviral infection (DENV or CHIKV) was the primary diagnosis for 93 patients. In total, 158 patients (41.0%) had malaria and 32 patients (8.3%) had CHIKV infections. Compared with real-time polymerase chain reaction, microscopy demonstrated a percent positive agreement of 49.7%. The percentage of malaria cases detected by microscopy varied significantly between clinical sites. Arboviral infections were the clinical diagnosis for patients on the Indian Ocean coast (91 of 238, 38.2%) significantly more often than patients in the Lake Victoria region (2 of 145, 1.4%; P < .001). However, detection of CHIKV infections was significantly higher in the Lake Victoria region (19 of 145 [13.1%] vs 13 of 239 [5.4%]; P = .012).

CONCLUSIONS

The clinical diagnosis of patients with an acute febrile illness, even when aided by microscopy, remains inaccurate in malaria-endemic areas, contributing to inappropriate management decisions.

Comparison of diagnostics for the detection of asymptomatic Plasmodium falciparum infections to inform control and elimination strategies

The global burden of malaria has been substantially reduced over the past two decades. Future efforts to reduce malaria further will require moving beyond the treatment of clinical infections to targeting malaria transmission more broadly in the community. As such, the accurate identification of asymptomatic human infections, which can sustain a large proportion of transmission, is becoming a vital component of control and elimination programmes. We determined the relationship across common diagnostics used to measure malaria prevalence - polymerase chain reaction (PCR), rapid diagnostic test and microscopy - for the detection of Plasmodium falciparum infections in endemic populations based on a pooled analysis of cross-sectional data. We included data from more than 170,000 individuals comparing the detection by rapid diagnostic test and microscopy, and 30,000 for detection by rapid diagnostic test and PCR. The analysis showed that, on average, rapid diagnostic tests detected 41% (95% confidence interval = 26-66%) of PCR-positive infections. Data for the comparison of rapid diagnostic test to PCR detection at high transmission intensity and in adults were sparse. Prevalence measured by rapid diagnostic test and microscopy was comparable, although rapid diagnostic test detected slightly more infections than microscopy. On average, microscopy captured 87% (95% confidence interval = 74-102%) of rapid diagnostic test-positive infections. The extent to which higher rapid diagnostic test detection reflects increased sensitivity, lack of specificity or both, is unclear. Once the contribution of asymptomatic individuals to the infectious reservoir is better defined, future analyses should ideally establish optimal detection limits of new diagnostics for use in control and elimination strategies.

Molecular Testing for Plasmodium falciparum by Use of Serum or Plasma and Comparison with Microscopy and Rapid Diagnostic Testing in Febrile Nigerian Patients

Plasmodium nucleic acids have been detected in serum and plasma, but there is little published data describing the diagnostic performance of malaria nucleic acid amplification tests (NAATs) using these specimen types. Previously, our group described a multiplex NAAT for the detection of dengue virus, Leptospira, and Plasmodium species with a callout for P. falciparum (the DLM assay) that demonstrated sensitive detection of P. falciparum from plasma samples during initial evaluation. In this study, we evaluated the sensitivity and specificity of P. falciparum detection in febrile Nigerian patients using the DLM assay, microscopy, and a rapid diagnostic test (BinaxNOW Malaria). Assay performances were compared using a composite reference, which was considered positive if malaria was detected by two or more methods. Serum (n = 182) or plasma (n = 148) from 317 patients was tested; the average sample volume was 70 μl (range, 5 to 300 μl). The sensitivity and specificity of the DLM assay were 97.1% and 93.5%, respectively. The sensitivity of the malaria rapid diagnostic test (98.1%) was similar to that of the DLM assay, and both proved significantly more sensitive than microscopy (79%; P < 0.0001). When analysis was limited to samples with ≥75 μl of serum or plasma, the sensitivity of the DLM assay improved to 99% and specificity was 97.5%. For P. falciparum cases, cycle threshold values in the DLM assay correlated with the parasite density detected by microscopy (Spearman's rank correlation coefficient, P < 0.0001). In conclusion, malaria detection using the DLM assay on serum or plasma is more sensitive than and equal in specificity to microscopy in patients with P. falciparum malaria.

Measuring obesity in the absence of a gold standard

Reliable measures of body composition are essential to develop effective policies to tackle obesity. The lack of an acceptable gold-standard for measuring fatness has made it difficult to evaluate alternative measures of obesity. We use latent class analysis to characterise existing diagnostics. Using data on US adults we show that measures based on body mass index and bioelectrical impedance analysis misclassify large numbers of individuals. For example, 45% of obese White women are misclassified as non-obese using body mass index, while over 50% of non-obese White women are misclassified as being obese using bioelectrical impedance analysis. In contrast the misclassification rates are low when waist circumference is used to measure obesity. These results have important implications for our understanding of differences in obesity rates across time and groups, as well as posing challenges for the econometric analysis of obesity.

Pooled PCR testing strategy and prevalence estimation of submicroscopic infections using Bayesian latent class models in pregnant women receiving intermittent preventive treatment at Machinga District Hospital, Malawi, 2010

BACKGROUND

Low malaria parasite densities in pregnancy are a diagnostic challenge. PCR provides high sensitivity and specificity in detecting low density of parasites, but cost and technical requirements limit its application in resources-limited settings. Pooling samples for PCR detection was explored to estimate prevalence of submicroscopic malaria infection in pregnant women at delivery. Previous work uses gold-standard based methods to calculate sensitivity and specificity of tests, creating a challenge when newer methodologies are substantially more sensitive than the gold standard. Thus prevalence was estimated using Bayesian latent class models (LCMs) in this study.

METHODS

Nested PCR (nPCR) for the 18S rRNA gene subunit of Plasmodium falciparum was conducted to detect malaria infection in microscopy-negative Malawian women on IPTp. Two-step sample pooling used dried blood spot samples (DBSs) collected from placenta or periphery at delivery. Results from nPCR and histology as well as previously published data were used to construct LCMs to estimate assay sensitivity and specificity. Theoretical confidence intervals for prevalence of infection were calculated for two-step and one-step pooling strategies.

RESULTS

Of 617 microscopy-negative Malawian women, 39 (6.3%) were identified as actively infected by histology while 52 (8.4%) were positive by nPCR. One hundred forty (22.7%) individuals had past infection assessed by histology. With histology as a reference, 72% of women in the active infection group, 7.1% in the past infection group and 3.2% in histology-negative group were nPCR positive. Using latent class models without a gold standard, histology had a median sensitivity of 49.7% and specificity of 97.6% for active infection while PCR had a median sensitivity of 96.0% and specificity of 99.1%. The true prevalence of active infection was estimated at 8.0% (CI: 5.8-10.5%) from PCR. PCR also had similar sensitivity for detecting either peripheral or placental malaria for submicroscopic infections. One-step pooling would give similar confidence intervals for pool sizes less than 20 while reducing the number of tests performed.

CONCLUSIONS

Pooled nPCR testing was a sensitive and resource-efficient strategy and LCMs provided precise prevalence estimates of submicroscopic infections. Compared to two-step pooling, one-step pooling could provide similar prevalence estimates at population levels with many fewer tests required.

Sensitivity of diagnostic tests for human soil-transmitted helminth infections: a meta-analysis in the absence of a true gold standard

•

A Bayesian latent class meta-analysis of diagnostic tests for soil-transmitted helminths was performed.

•

Overall sensitivity of evaluated diagnostic tests was low.

•

Test performance was strongly influenced by intensity of infection.

•

FLOTAC method sensitivity was highest overall and in both intensity groups.

•

The performance of the Kato-Katz method in high intensity settings was acceptable.

Reliable, sensitive and practical diagnostic tests are an essential tool in disease control programmes for mapping, impact evaluation and surveillance. To provide a robust global assessment of the relative performance of available diagnostic tools for the detection of soil-transmitted helminths, we conducted a meta-analysis comparing the sensitivities and the quantitative performance of the most commonly used copro-microscopic diagnostic methods for soil-transmitted helminths, namely Kato-Katz, direct microscopy, formol-ether concentration, McMaster, FLOTAC and Mini-FLOTAC. In the absence of a perfect reference standard, we employed a Bayesian latent class analysis to estimate the true, unobserved sensitivity of compared diagnostic tests for each of the soil-transmitted helminth species Ascaris lumbricoides, Trichuris trichiura and the hookworms. To investigate the influence of varying transmission settings we subsequently stratified the analysis by intensity of infection. Overall, sensitivity estimates varied between the different methods, ranging from 42.8% for direct microscopy to 92.7% for FLOTAC. The widely used double slide Kato-Katz method had a sensitivity of 74–95% for the three soil-transmitted helminth species at high infection intensity, however sensitivity dropped to 53–80% in low intensity settings, being lowest for hookworm and A. lumbricoides. The highest sensitivity, overall and in both intensity groups, was observed for the FLOTAC method, whereas the sensitivity of the Mini-FLOTAC method was comparable with the Kato-Katz method. FLOTAC average egg count estimates were significantly lower compared with Kato-Katz, while the compared McMaster counts varied. In conclusion, we demonstrate that the Kato-Katz and Mini-FLOTAC methods had comparable sensitivities. We further show that test sensitivity of the Kato-Katz method is reduced in low transmission settings.

Estimation of diagnostic test accuracy without full verification: a review of latent class methods

The performance of a diagnostic test is best evaluated against a reference test that is without error. For many diseases, this is not possible, and an imperfect reference test must be used. However, diagnostic accuracy estimates may be biased if inaccurately verified status is used as the truth. Statistical models have been developed to handle this situation by treating disease as a latent variable. In this paper, we conduct a systematized review of statistical methods using latent class models for estimating test accuracy and disease prevalence in the absence of complete verification.

Multiplex nucleic acid amplification test for diagnosis of dengue fever, malaria, and leptospirosis

Dengue, leptospirosis, and malaria are among the most common etiologies of systemic undifferentiated febrile illness (UFI) among travelers to the developing world, and these pathogens all have the potential to cause life-threatening illness in returned travelers. The current study describes the development of an internally controlled multiplex nucleic acid amplification test for the detection of dengue virus (DENV) and Leptospira and Plasmodium species, with a specific callout for Plasmodium falciparum (referred to as the UFI assay). During analytical evaluation, the UFI assay displayed a wide dynamic range and a sensitive limit of detection for each target, including all four DENV serotypes. In a clinical evaluation including 210 previously tested samples, the sensitivities of the UFI assay were 98% for DENV (58/59 samples detected) and 100% for Leptospira and malaria (65/65 and 20/20 samples, respectively). Malaria samples included all five Plasmodium species known to cause human disease. The specificity of the UFI assay was 100% when evaluated with a panel of 66 negative clinical samples. Furthermore, no amplification was observed when extracted nucleic acids from related pathogens were tested. Compared with whole-blood samples, the UFI assay remained positive for Plasmodium in 11 plasma samples from patients with malaria (parasitemia levels of 0.0037 to 3.4%). The syndrome-based design of the UFI assay, combined with the sensitivities of the component tests, represents a significant improvement over the individual diagnostic tests available for these pathogens.

Diagnostic accuracy and cost-effectiveness of alternative methods for detection of soil-transmitted helminths in a post-treatment setting in western Kenya

OBJECTIVES

This study evaluates the diagnostic accuracy and cost-effectiveness of the Kato-Katz and Mini-FLOTAC methods for detection of soil-transmitted helminths (STH) in a post-treatment setting in western Kenya. A cost analysis also explores the cost implications of collecting samples during school surveys when compared to household surveys.

METHODS

Stool samples were collected from children (n = 652) attending 18 schools in Bungoma County and diagnosed by the Kato-Katz and Mini-FLOTAC coprological methods. Sensitivity and additional diagnostic performance measures were analyzed using Bayesian latent class modeling. Financial and economic costs were calculated for all survey and diagnostic activities, and cost per child tested, cost per case detected and cost per STH infection correctly classified were estimated. A sensitivity analysis was conducted to assess the impact of various survey parameters on cost estimates.

RESULTS

Both diagnostic methods exhibited comparable sensitivity for detection of any STH species over single and consecutive day sampling: 52.0% for single day Kato-Katz; 49.1% for single-day Mini-FLOTAC; 76.9% for consecutive day Kato-Katz; and 74.1% for consecutive day Mini-FLOTAC. Diagnostic performance did not differ significantly between methods for the different STH species. Use of Kato-Katz with school-based sampling was the lowest cost scenario for cost per child tested ($10.14) and cost per case correctly classified ($12.84). Cost per case detected was lowest for Kato-Katz used in community-based sampling ($128.24). Sensitivity analysis revealed the cost of case detection for any STH decreased non-linearly as prevalence rates increased and was influenced by the number of samples collected.

CONCLUSIONS

The Kato-Katz method was comparable in diagnostic sensitivity to the Mini-FLOTAC method, but afforded greater cost-effectiveness. Future work is required to evaluate the cost-effectiveness of STH surveillance in different settings.

Potential biomarkers and their applications for rapid and reliable detection of malaria

Malaria has been responsible for the highest mortality in most malaria endemic countries. Even after decades of malaria control campaigns, it still persists as a disease of high mortality due to improper diagnosis and rapidly evolving drug resistant malarial parasites. For efficient and economical malaria management, WHO recommends that all malaria suspected patients should receive proper diagnosis before administering drugs. It is thus imperative to develop fast, economical, and accurate techniques for diagnosis of malaria. In this regard an in-depth knowledge on malaria biomarkers is important to identify an appropriate biorecognition element and utilize it prudently to develop a reliable detection technique for diagnosis of the disease. Among the various biomarkers, plasmodial lactate dehydrogenase and histidine-rich protein II (HRP II) have received increasing attention for developing rapid and reliable detection techniques for malaria. The widely used rapid detection tests (RDTs) for malaria succumb to many drawbacks which promotes exploration of more efficient economical detection techniques. This paper provides an overview on the current status of malaria biomarkers, along with their potential utilization for developing different malaria diagnostic techniques and advanced biosensors.

Field evaluation of the photo-induced electron transfer fluorogenic primers (PET) real-time PCR for the detection of Plasmodium falciparum in Tanzania

Background

Accurate diagnosis of malaria infections remains challenging, especially in the identification of submicroscopic infections. New molecular diagnostic tools that are inexpensive, sensitive enough to detect low-level infections and suitable in laboratory settings of resource-limited countries are required for malaria control and elimination programmes. Here the diagnostic potential of a recently developed photo-induced electron transfer fluorogenic primer (PET) real-time polymerase chain reaction (PCR) called PET-PCR was investigated. This study aimed to (i) evaluate the use of this assay as a method for the detection of both Plasmodium falciparum and other Plasmodium species infections in a developing country’s diagnostic laboratory; and, (ii) determine the assay’s sensitivity and specificity compared to a nested 18S rRNA PCR.

Methods

Samples used in this study were obtained from a previous study conducted in the region of Iringa, Tanzania. A total of 303 samples from eight health facilities in Tanzania were utilized for this evaluation. All samples were screened using the multiplex PET-PCR assay designed to detect Plasmodium genus and P. falciparum initially in laboratory in Tanzania and then repeated at a reference laboratory at the CDC in the USA. Microscopy data was available for all the 303 samples. A subset of the samples were tested in a blinded fashion to find the sensitivity and specificity of the PET-PCR compared to the nested 18S rRNA PCR.

Results

Compared to microscopy, the PET-PCR assay was 59% more sensitive in detecting P. falciparum infections. The observed sensitivity and specificity were 100% (95% confidence interval (CI_0.95) = 94-100%) and (CI_0.95 = 96-100%), respectively, for the PET-PCR assay when compared to nested 18S rRNA PCR. When compared to 18S rRNA PCR, microscopy had a low sensitivity of 40% (CI_0.95 = 23-61%) and specificity of 100% (CI_0.95 = 96-100%). The PET-PCR results performed in the field laboratory in Tanzania were in 100% concordance with the results obtained at the reference laboratory in the USA.

Conclusion

The PET-PCR is a new molecular diagnostic tool with similar performance characteristics as commonly used PCR methods that is less expensive, easy to use, and amiable to large scale-surveillance studies in developing country settings.

Impact of intermittent screening and treatment for malaria among school children in Kenya: a cluster randomised trial

BACKGROUND

Improving the health of school-aged children can yield substantial benefits for cognitive development and educational achievement. However, there is limited experimental evidence of the benefits of alternative school-based malaria interventions or how the impacts of interventions vary according to intensity of malaria transmission. We investigated the effect of intermittent screening and treatment (IST) for malaria on the health and education of school children in an area of low to moderate malaria transmission.

METHODS AND FINDINGS

A cluster randomised trial was implemented with 5,233 children in 101 government primary schools on the south coast of Kenya in 2010-2012. The intervention was delivered to children randomly selected from classes 1 and 5 who were followed up for 24 months. Once a school term, children were screened by public health workers using malaria rapid diagnostic tests (RDTs), and children (with or without malaria symptoms) found to be RDT-positive were treated with a six dose regimen of artemether-lumefantrine (AL). Given the nature of the intervention, the trial was not blinded. The primary outcomes were anaemia and sustained attention. Secondary outcomes were malaria parasitaemia and educational achievement. Data were analysed on an intention-to-treat basis. During the intervention period, an average of 88.3% children in intervention schools were screened at each round, of whom 17.5% were RDT-positive. 80.3% of children in the control and 80.2% in the intervention group were followed-up at 24 months. No impact of the malaria IST intervention was observed for prevalence of anaemia at either 12 or 24 months (adjusted risk ratio [Adj.RR]: 1.03, 95% CI 0.93-1.13, p = 0.621 and Adj.RR: 1.00, 95% CI 0.90-1.11, p = 0.953) respectively, or on prevalence of P. falciparum infection or scores of classroom attention. No effect of IST was observed on educational achievement in the older class, but an apparent negative effect was seen on spelling scores in the younger class at 9 and 24 months and on arithmetic scores at 24 months.

CONCLUSION

In this setting in Kenya, IST as implemented in this study is not effective in improving the health or education of school children. Possible reasons for the absence of an impact are the marked geographical heterogeneity in transmission, the rapid rate of reinfection following AL treatment, the variable reliability of RDTs, and the relative contribution of malaria to the aetiology of anaemia in this setting.

TRIAL REGISTRATION

www.ClinicalTrials.gov NCT00878007.

Using a web-based application to define the accuracy of diagnostic tests when the gold standard is imperfect

BACKGROUND

Estimates of the sensitivity and specificity for new diagnostic tests based on evaluation against a known gold standard are imprecise when the accuracy of the gold standard is imperfect. Bayesian latent class models (LCMs) can be helpful under these circumstances, but the necessary analysis requires expertise in computational programming. Here, we describe open-access web-based applications that allow non-experts to apply Bayesian LCMs to their own data sets via a user-friendly interface.

METHODS/PRINCIPAL FINDINGS

Applications for Bayesian LCMs were constructed on a web server using R and WinBUGS programs. The models provided (http://mice.tropmedres.ac) include two Bayesian LCMs: the two-tests in two-population model (Hui and Walter model) and the three-tests in one-population model (Walter and Irwig model). Both models are available with simplified and advanced interfaces. In the former, all settings for Bayesian statistics are fixed as defaults. Users input their data set into a table provided on the webpage. Disease prevalence and accuracy of diagnostic tests are then estimated using the Bayesian LCM, and provided on the web page within a few minutes. With the advanced interfaces, experienced researchers can modify all settings in the models as needed. These settings include correlation among diagnostic test results and prior distributions for all unknown parameters. The web pages provide worked examples with both models using the original data sets presented by Hui and Walter in 1980, and by Walter and Irwig in 1988. We also illustrate the utility of the advanced interface using the Walter and Irwig model on a data set from a recent melioidosis study. The results obtained from the web-based applications were comparable to those published previously.

CONCLUSIONS

The newly developed web-based applications are open-access and provide an important new resource for researchers worldwide to evaluate new diagnostic tests.

A latent class model for defining severe hemorrhage: experience from the PROMMTT study

BACKGROUND

Several predictive models have been developed to identify trauma patients who have had severe hemorrhage (SH) and may need a massive transfusion (MT) protocol. However, almost all these models define SH as the transfusion of 10 or more units of red blood cells (RBCs) within 24 hours of emergency department admission (also known as MT). This definition excludes some patients with SH, especially those who die before a 10th unit of RBCs could be transfused, which calls the validity of these prediction models into question. We show how a latent class model could improve the accuracy of identifying the SH patients.

METHODS

Modeling SH classification as a latent variable, we estimate the posterior probability of a patient in SH based on emergency department admission variables (systolic blood pressure, heart rate, pH, hemoglobin), the 24-hour blood product use (plasma/RBC and platelet/RBC ratios), and 24-hour survival status. We define the SH subgroup as those having a posterior probability of 0.5 or greater. We compare our new classification of SH with that of the traditional MT using data from PROMMTT study.

RESULTS

Of the 1,245 patients, 913 had complete data, which were used in the latent class model. About 25.3% of patients were classified as SH. The overall agreement between the MT and SH classifications was 83.8%. However, among 49 patients who died before receiving the 10th unit of RBCs, 41 (84%) were classified as SH. Seven (87.5%) of the remaining eight patients who were not classified as SH had head injury.

CONCLUSION

Our definition of SH based on the aforementioned latent class model has an advantage of improving on the traditional MT definition by identifying SH patients who die before receiving the 10th unit of RBCs. We recommend further improvements to more accurately classify SH patients, which could replace the traditional definition of MT for use in developing prediction algorithms.

The use of bayesian latent class cluster models to classify patterns of cognitive performance in healthy ageing

The main focus of this study is to illustrate the applicability of latent class analysis in the assessment of cognitive performance profiles during ageing. Principal component analysis (PCA) was used to detect main cognitive dimensions (based on the neurocognitive test variables) and Bayesian latent class analysis (LCA) models (without constraints) were used to explore patterns of cognitive performance among community-dwelling older individuals. Gender, age and number of school years were explored as variables. Three cognitive dimensions were identified: general cognition (MMSE), memory (MEM) and executive (EXEC) function. Based on these, three latent classes of cognitive performance profiles (LC1 to LC3) were identified among the older adults. These classes corresponded to stronger to weaker performance patterns (LC1>LC2>LC3) across all dimensions; each latent class denoted the same hierarchy in the proportion of males, age and number of school years. Bayesian LCA provided a powerful tool to explore cognitive typologies among healthy cognitive agers.

Multiplex 5' nuclease quantitative real-time PCR for clinical diagnosis of malaria and species-level identification and epidemiologic evaluation of malaria-causing parasites, including Plasmodium knowlesi

Molecular diagnosis of malaria offers many potential advantages over microscopy, including identification of malaria to the species level in an era with few experienced microscopists. We developed high-throughput multiplex 5' nuclease quantitative PCR (qPCR) assays, with the potential to support large studies, to specifically identify Plasmodium falciparum, P. vivax, P. ovale, P. malariae, and P. knowlesi. We compared qPCR to microscopy and confirmed discordant results with an alternative target PCR assay. The assays specifically detected 1 to 6 parasites/μl of blood. The clinical sensitivities (95% confidence intervals [CIs]) of the 4-plex assay to detect microscopically confirmed malaria were 95.8% (88.3 to 99.1%) for P. falciparum, 89.5% (75.2 to 97.1%) for P. vivax, 94.1% (71.3 to 99.9%) for P. ovale, and 100% (66.4 to 100%) for P. malariae. The specificities (95% CIs) were 98.6% (92.4 to 100%) for P. falciparum, 99% (84.8 to 100%) for P. vivax, 98.4% (94.4 to 99.8%) for P. ovale, and 99.3% (95.9 to 100%) for P. malariae. The clinical specificity for samples without malaria was 100%. The clinical sensitivity of the 5-plex assay for confirmed P. knowlesi malaria was 100% (95% CI, 69.2 to 100%), and the clinical specificity was 100% (95% CI, 87.2 to 100%). Coded retesting and testing with an alternative target PCR assay showed improved sensitivity and specificity of multiplex qPCR versus microscopy. Additionally, 91.7% (11/12) of the samples with uncertain species by microscopy were identified to the species level identically by both our multiplex qPCR assay and the alternative target PCR assay, including 9 P. falciparum infections. Multiplex qPCR can rapidly and simultaneously identify all 5 Plasmodium species known to cause malaria in humans, and it offers an alternative or adjunct to microscopy for clinical diagnosis as well as a needed high-throughput tool for research.

The mechanistic, diagnostic and prognostic utility of biomarkers in severe malaria

Malaria remains an important global cause of severe illness and mortality. This literature review summarizes available data on how biomarkers might be applied to diagnose, prognosticate and provide mechanistic insights in patients with severe malaria. Of the large number of candidate biomarkers, only PfHRP2 has consistently demonstrated clinical utility and, when incorporated into rapid antigen detection tests, has shown diagnostic sensitivity above 95%, which is at least as good as light microscopy. As a quantitative test, PfHRP2 also shows some promise in differentiating severe malarial from non-malarial disease in areas where asymptomatic carriage of malaria parasites is common, and possibly as a tool to estimate sequestered parasite burden and subsequent mortality. Biomarkers such as pLDH and panmalarial antigen have lower sensitivity for non-falciparum malaria in rapid antigen detection tests. There is an urgent need to discover and validate better biomarkers for incorporation into rapid antigen detection tests in countries where Plasmodium vivax is a common cause of severe disease. A large number of host-derived acute-phase reactants, markers of endothelial dysfunction and immune mediators have been proposed as biomarkers. Although they have provided mechanistic insights into the immunopathology of severe malaria, their roles as clinical tools remain uncertain.

Application of loop-mediated isothermal amplification for malaria diagnosis during a follow-up study in São Tomé

BACKGROUND

A reliable and simple test for the detection of malaria parasite is crucial in providing effective treatment and therapeutic follow-up, especially in malaria elimination programmes. A comparison of four methods, including nested polymerase chain reaction (PCR) and loop-mediated isothermal amplification (LAMP) were used for the malaria diagnosis and treatment follow-up in São Tomé and Príncipe, during a successful pre-elimination campaign.

METHOD

During the period September to November 2009, blood samples from 128 children (five to 14 years old) with temperature ≥38°C (tympanic) in the District of Agua Grande were examined using four different methods, i.e., histidine-rich protein 2 (HRP-2) based rapid diagnostic tests (HRP-2-RDTs), optical microscopy, nested PCR, and LAMP. First-line treatment with artesunate-amodiaquine was given for uncomplicated malaria and intravenous quinine was given for complicated malaria. Children with persistent positivity for malaria by microscopy, or either by nested PCR, or by LAMP on day 7 were given second-line treatment with artemether-lumefantrine. Treatment follow-up was made weekly, for up to four weeks.

RESULTS

On day 0, positive results for HRP-2-RDTs, microscopy, nested PCR, and LAMP, were 68(53%), 47(37%), 64(50%), and 65(51%), respectively. When nested PCR was used as a reference standard, only LAMP was comparable; both HRP-2-RDTs and microscopy had moderate sensitivity; HRP-2-RDTs had poor positive predictive value (PPV) and a moderate negative predictive value (NPV) for the treatment follow-up. Seventy-one children with uncomplicated malaria and eight children with complicated falciparum malaria were diagnosed based on at least one positive result from the four tests as well as clinical criteria. Twelve of the 79 children receiving first-line treatment had positive results by nested PCR on day 7 (nested PCR-corrected day 7 cure rate was 85%). After the second-line treatment, nested PCR/LAMP-corrected day 28 cure rate was 83% for these 12 children.

CONCLUSIONS

HRP-2-RDTs have similar sensitivity as microscopy but less specificity. However, as compared to nested PCR, the poor sensitivity of HRP-2-RDTs indicates that low parasitaemia may not be detected after treatment, as well as the low specificity of HRP-2-RDTs indicates it cannot be applied for treatment follow-up. LAMP has similar sensitivity and specificity to nested PCR. With high PPV and NPV, LAMP is simpler and faster as compared to nested PCR with the advantage of detecting low parasitaemia becoming a potential point-of-care test for treatment follow-up.

Rapid antigen detection tests for malaria diagnosis in severely ill Papua New Guinean children: a comparative study using Bayesian latent class models

Background

Although rapid diagnostic tests (RDTs) have practical advantages over light microscopy (LM) and good sensitivity in severe falciparum malaria in Africa, their utility where severe non-falciparum malaria occurs is unknown. LM, RDTs and polymerase chain reaction (PCR)-based methods have limitations, and thus conventional comparative malaria diagnostic studies employ imperfect gold standards. We assessed whether, using Bayesian latent class models (LCMs) which do not require a reference method, RDTs could safely direct initial anti-infective therapy in severe ill children from an area of hyperendemic transmission of both Plasmodium falciparum and P. vivax.

Methods and Findings

We studied 797 Papua New Guinean children hospitalized with well-characterized severe illness for whom LM, RDT and nested PCR (nPCR) results were available. For any severe malaria, the estimated prevalence was 47.5% with RDTs exhibiting similar sensitivity and negative predictive value (NPV) to nPCR (≥96.0%). LM was the least sensitive test (87.4%) and had the lowest NPV (89.7%), but had the highest specificity (99.1%) and positive predictive value (98.9%). For severe falciparum malaria (prevalence 42.9%), the findings were similar. For non-falciparum severe malaria (prevalence 6.9%), no test had the WHO-recommended sensitivity and specificity of >95% and >90%, respectively. RDTs were the least sensitive (69.6%) and had the lowest NPV (96.7%).

Conclusions

RDTs appear a valuable point-of-care test that is at least equivalent to LM in diagnosing severe falciparum malaria in this epidemiologic situation. None of the tests had the required sensitivity/specificity for severe non-falciparum malaria but the number of false-negative RDTs in this group was small.