https://orcid.org/0000-0002-8336-1949
Figures
Abstract
The absence of an accurate reference test complicates the evaluation of tuberculosis (TB) diagnostic tests among people living with Human Immunodeficiency Virus (PLWHIV). The objective of this study was to estimate (using Bayesian latent class models [BLCM]) the sensitivity (Se), specificity (Sp) and negative and positive predictive values (NPV and PPV) of sputum smear microscopy (SSM), Xpert Ultra and lipoarabinomannan antigen (LAM) tests for TB among PLWHIV in Nairobi, Kenya. This cross-sectional study enrolled a total of 190 patients aged ≥ 18 years with presumptive TB seeking treatment at the Kibra Community Health Center Comprehensive Care Centre (CCC) clinic between September 2022 and March 2023. The diagnostic data obtained from the three tests were analysed using a BLCM framework to derive accuracy estimates of the three diagnostic tests. The Xpert Ultra assay registered a higher Se (85.0; 95% PCI [41.4–99.4]) compared to LAM (26.8; 95% PCI [4.7–67.6]) and SSM (56.7 [16.4–97.4]). However, SSM had the highest Sp (99.6; 95% PCI [97.7–100.0]). The Xpert Ultra assay yielded the highest overall combination of Se and Sp at 80.8% (95% PCI [37.0–96.5]). On predictive values, SSM recorded the highest PPV at 84.5% (95% PCI [38.4–99.4]). Nonetheless, all the tests exhibited noticeably high NPVs (>96%). The Xpert Ultra assay recorded the highest Se of the three tests. Nevertheless, the SSM test registered the highest Sp and correspondingly the highest PPV. On NPVs, the tests had similar estimates. Consequently, a two-test serial testing strategy, entailing an initial screening with the more sensitive Xpert Ultra assay followed by retesting of any positives with the more specific SSM test, could be most optimal in this low-burden TB setting.
Citation: Ochola EA, Mweu MM (2024) Performance of three diagnostic tests for tuberculosis among Human Immunodeficiency Virus (HIV) patients presenting to a health facility in an informal urban settlement in Nairobi, Kenya. PLoS ONE 19(12): e0307727. https://doi.org/10.1371/journal.pone.0307727
Editor: Guocan Yu, Hangzhou Red Cross Hospital, CHINA
Received: July 25, 2024; Accepted: December 9, 2024; Published: December 30, 2024
Copyright: © 2024 Ochola, Mweu. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All underlying data files are available from the Havard dataverse database https://doi.org/10.7910/DVN/PYRPCQ.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Worldwide, tuberculosis (TB) ranks among the main causes of mortality from a single pathogenic agent [1]. Human Immunodeficiency Virus (HIV) infected patients have a 20 to 30 fold higher probability of contracting TB [2]. As of 2017, approximately 9% of the 10 million new TB cases were people living with HIV (PLWHIV), with Africa accounting for a 72% disproportionate share [3]. Kenya ranks amid the high-TB burden nations whose primary cause of death is TB [4].
In Kenya, pulmonary TB is the most prevalent type and 66% of TB cases occur among those aged ≤44 years [5]. In line with the national TB diagnosis and treatment guidelines, the commencement of TB treatment is informed by bacteriological confirmation of Mycobacterium tuberculosis (MTB). However, clinical diagnosis where bacteriological confirmation is not possible may suffice for initiation of treatment [6].
Sputum smear microscopy (SSM) has been the main diagnostic tool for TB especially in developing countries [7]. It is extensively used because it is easy to run, rapid, inexpensive and has high specificity (Sp) for TB diagnosis [8,9]. Nevertheless, it has a limited sensitivity (Se) of approximately 10,000 TB bacilli/ml of sputum. Notably, its poor Se poses a challenge for diagnosis of TB in PLWHIV since they have fewer bacilli in their sputum [10].
Diagnostic tests with improved Se and rapid turnaround time among PLWHIV could lead to improved patient outcomes, reduced mortality and lower progressive transmission of TB [11]. The real time polymerase chain reaction (PCR) assay utilizing the gene Xpert platform such as Xpert Ultra, offers better test Se and Sp with ability to discriminate between tuberculous and non-tuberculous mycobacteria[12]. Moreover, the test may detect at least 15.6 Colony forming units (CFU) of MTB/ml in sputum samples in under two hours [12]. Nonetheless, the technical and operational cost implications of the diagnostic test in resource-limited settings remain a challenge [13,14].
With sputum production among HIV infected individuals being limited [15], the need for alternative biological specimens for the diagnosis of TB arises. A rapid immunochromatographic test detecting the Mycobacterium cell wall antigen lipoarabinomannan (LAM) in urine samples is available as a point-of-care test for TB [16]. The test is easy to run, has no need for specialized equipment or laboratory facilities and results are obtainable within 25 minutes—rendering the test field applicable. Moreover, among PLWHIV, the test displays high Sp, with Se increasing with diminishing CD4 counts [15–18].
Diagnostic studies evaluating the performance of SSM, Xpert Ultra and TB LAM among PLWHIV have previously been conducted globally [19,20]. A major shortcoming of these evaluations is that the index tests were evaluated against imperfect reference standards, thus introducing bias in the accuracy estimates of the evaluated tests[21]. A superior alternative involves the use of latent class models fit using Bayesian methods that allow for concurrent evaluation of Se and Sp of two or more tests without prior knowledge of the unobserved disease status of individuals [22]. Consequently, the objective here was to estimate the Se and Sp as well as the predictive values of SSM, Xpert Ultra and LAM tests for TB among HIV patients within an informal outpatient setting using a Bayesian latent class model (BLCM) framework.
Materials and methods
Study setting, design and population
The study was carried out at the Kibra Community Health Centre; a level three state-run facility located in the southern part of Nairobi. Notably, the facility serves an immediate large informal-settlement population characterised by a dual burden of HIV/AIDS and TB, which are leading causes of mortality in slums [23]. It registers around150-200 out-patients daily, conducts approximately 50–60 deliveries monthly and offers comprehensive care clinic (CCC) services to over 1,500 patients per month. The services offered at the CCC include: HIV counselling and testing, HIV prevention services including prevention of mother—to—child transmission of HIV, care and treatment of HIV and TB, TB screening and prevention, and nutritional counselling. Importantly, all the three tests under evaluation are routinely carried out for diagnosis of TB among PLWHIV in Kenyan outpatient settings.
A facility-based cross-sectional study was employed to evaluate the performance of the three diagnostic tests for TB. The rationale for the choice of this design stems not only from the descriptive nature of the study but also the ease of recruitment of study participants presenting to the CCC clinic for care. The study population comprised all PLWHIV seeking health services at the centre’s CCC who met specific eligibility criteria. Eligibility was limited to individuals aged ≥ 18 years (symptomatic or asymptomatic for TB and who, at the time of recruitment (28th September 2022- 31st March 2023), were visiting the CCC for HIV care and consented to participation in the study. All PLWHIV already under treatment for TB as well as those not able to produce the required diagnostic specimen were excluded from the study.
Sample size and participant selection
The required sample size was derived using the McNemar’s sample size formula for paired proportions[24]:
Where: nper test = sample size required for each test,
(1.96) is the value required for the two-sided 95% confidence level, and Zβ (-0.84) is the value specifying the desired power of 80%. Se1 and Se2 are estimates of Se of the Gene xpert MTB/RIF and SSM respectively from literature i.e. Se1 is 90.3% and Se2 is 67.1% [25].
As per the above mentioned figures, a total sample size of 190 was generated, after an upward adjustment by 5% to account for non-response.
To obtain the required sample, eligible individuals were systematically random sampled in their order of arrival at the facility’s CCC. Importantly, recruitment followed triage by the clinic nurse and examination by a physician.
Ethical considerations
Written permission to conduct this study was obtained from and approved from the Kenyatta National Hospital -University of Nairobi Ethics and Research Committee (Ref. No. KNH/ERC/R/145 and KNH/ ERC/Mod & SAE/347), the Nairobi metropolitan Services (Ref. No. EOP/NMS/HS/222), as well as the National Commission for Technology and Innovation (NACOSTI) (Ref. No. NACOSTI/P/21/14098). Moreover, a written informed consent was secured from individual patients prior to enrolling them into the study. To ensure general understandability, the consent form was captured in both English and Swahili (local language) and administered orally.
Sample collection
Participants were provided with clear instructions on collection of urine and sputum samples. Of note, the LAM and SSM tests were carried out immediately at the facility’s laboratory by a trained laboratorian. Samples for the Xpert Ultra assay were stored in cooler boxes and delivered to a referral laboratory for processing. On enrolment, the participants’ sociodemographic characteristics (age, sex, marital status, level of education, area of residence, occupation and highly active antiretroviral therapy status (HAART) were elicited using an orally-administered semi-structured questionnaire. As with the consent form, the questionnaire was presented in both English and Swahili.
Target condition
The infection (latent) status being targeted by the three tests (Xpert Ultra, SSM and LAM) constitutes a respiratory or urine sample carrying either the live MTB, its DNA or cell wall LAM antigen at any concentration.
Sputum Smear Microscopy (SSM)
This was performed as per standard guidelines[7]. Briefly, a sputum smear covering 2cm by 1 cm on the centre of the slide was prepared immediately after receipt of the sample. The smears were then stained with the primary stain 1% Carbol fuchsin for a period not exceeding 10 minutes. Decolourisation was then performed by flooding the slides with 25% sulphuric acid for 3 minutes and subsequently the counterstain 0.1% methylene blue was added for 1 minute. Slides were rinsed before every staining stage and excess water drained by gently tilting each slide. The slides were thereafter examined under a light microscope at 100x magnification. Mycobacterium quantification was achieved by counting the acid fast bacteria (AFB) as per the International Union Against TB and Lung Disease recommended grading of sputum smear microscopy [7]. For analytical purposes, AFB counts above zero constituted a positive result; otherwise negative.
Gene Xpert Mycobacterium/ Rifampicin Ultra (Xpert Ultra)
This was carried out as described elsewhere [26]. Briefly, a sample reagent (a constitution of sodium hydroxide and Isopropanol) was added into a sputum sample in a ratio of 2:1 for liquefaction and inactivation of the sputum. The mixture was then vigorously shaken and incubated at room temperature for 15 minutes. The liquefied specimen was afterwards pipetted into the test cartridge (Cat. No. 42722, Cepheid). The pre-labelled test cartridge was loaded into the MTB/RIF test platform (Cepheid) where automated testing commenced. The semi-nested real time amplification and detection occurred in an integrated reaction tube and the results were printed. Results were interpreted as positive when MTB was detected.
Lateral Flow Urine Lipoarabinomannan (LAM)
Alere Determine™ TB LAM antigen kit (Cat No. 238237, Abbott Diagnostics Scarborough) was used to test for the presence of TB LAM antigen in collected urine samples. This was executed as per the instructions in the Alere Determine™ TB LAM antigen manual [27]. A total of 60ul of urine was added into the strip sample well. In order to allow for the urine to flow towards both the test and control windows, the strip was incubated at room temperature for 25 minutes. A positive result was denoted by the appearance of two colour bands; one band on the control line being indicative of a negative result.
Statistical analysis
A Bayesian latent class model fitted in OpenBUGS v3.2.2 [28] but called from R software through the ‘BRugs’ package [29] was used to estimate the TB prevalence, Se and Sp of the three tests together with their predictive values. The guidelines for standards for reporting diagnostic accuracy studies that use BLCMs were used to inform the analysis and reporting of the results [30]. The model code is available as supporting information [31].
Three assumptions are necessary when fitting a BLCM [32]. Firstly, the tests should be conditionally independent given the TB infection status. This was met since microscopy targets the live bacterium, Xpert Ultra amplifies and detects the Mycobacterium DNA and TB LAM detects the Mycobacterium cell wall LAM antigen. Hence, the probability of testing either positive or negative on one diagnostic tool did not depend on the outcome of the other tests. Secondly, the Se and Sp of the tests being evaluated should be constant across the subpopulations studied. With a single target population, as was the case here, the tests’ constancy premise was likely to be upheld. Thirdly, the target population should consist of two or more subpopulations with different prevalences. In situations where only a single population exists, at least three tests are necessary to provide sufficient degrees of freedom to achieve model identifiability [30].
A multinomial distribution was assumed for the counts (O) of the different test combinations (e. g. +, +, +) and took the form below:
Where Sei and Spi reflect the specific test characteristics of the individual tests (i = 1, 2, 3), P denotes the prevalence for the singular population, Prob is a vector of probabilities of observing different test combinations and n denotes the sample size of the study population. As an illustration, the probability of an individual testing positive on all three tests is given by:
With three tests, a total of seven degrees of freedom was sufficient to estimate the required seven parameters (the Se and Sp of the three tests together with the single population prevalence). Since there was no reliable prior information for any of the tests, uninformative priors (beta (1, 1)) were used as initial parameters for the model.
The positive predictive value (PPV) and negative predictive value (NPV) of the test (i) were estimated as follows:
Two Markov Chain Monte Carlo (MCMC) chains each having different values were used to initialise the Bayesian model. Convergence of the chains was then evaluated via the trace plots of selected variables and the Gelman-Rubin diagnostic plots. The goodness-of-fit of the Bayesian model was assessed using the posterior predictive p-value. This value is derived from the difference between the deviance of the observed dataset and that of simulated data generated randomly from the fitted model—with values around 0.50 suggestive of an overall good fit [33]. The posterior distributions of the test estimates (Se, Sp and the predictive values) and the population prevalence were reported as the median and the associated 95% posterior credible intervals (PCI). The Youden index which is a measure of a test’s overall diagnostic ability was also computed as: Se +Sp−1 [34].
Results
Of the 190 participants, two were unable to provide all the required samples and were thus excluded from the analysis. Another five were excluded on the basis of their Xpert Ultra results being invalid.
The sociodemographic characteristics and information on HAART status is as shown in Table 1. The median age of the participants was 43 years (Range: 19–82 yrs). The majority of participants were female (64.0%, n = 117), married (60.7%, n = 111), with only 2.2% (n = 4) lacking a formal education. A substantial percentage of the participants was on HAART (97.5%, n = 178).
(n = 183).
The cross-tabulated counts of the tests’ outcomes are displayed in Table 2. Notably, 81.4% (n = 149) of the participants registered negative results on all the tests.
(n = 183).
The estimates of the Se and Sp of the three tests, along with their respective predictive values and the prevalence of MTB infection are shown in Table 3. The Xpert Ultra assay registered a higher Se (85.0; 95% PCI [41.4–99.4]) compared to LAM (26.8; 95% PCI [4.7–67.6]) and SSM (56.7 [16.4–97.4]). However, SSM displayed the highest Sp (99.6; 95% PCI [97.7–100.0]). As per the Youden indices, Xpert Ultra yielded the highest overall combination of Se and Sp at 80.8% (95% PCI [37.0–96.5]). On predictive values, SSM recorded the highest PPV at 84.5% (95% PCI [38.4–99.4]). Nonetheless, all the tests exhibited noticeably high NPVs (>96%). The true prevalence of MTB infection in this HIV-infected outpatient population was 4.1% (95% PCI [1.2–11.2]). A posterior predictive p-value = 0.545 suggested a good model fit to the data. Of note, the trace plots hovered around the median of the variables’ distributions whereas the Gelman-Rubin plots had ratios converging to 1 –suggestive of convergence of the MCMC chains (Figs 1 & 2).
Discussion
This study has utilized Bayesian latent class analysis for the estimation of the accuracy of SSM, Xpert Ultra and LAM tests (together with their corresponding predictive values) for the diagnosis of TB infection among PLWHIV. This is a key strength in this study since BLCM allows for the estimation of index tests’ characteristics devoid of classification errors typically inherent in traditional evaluation studies employing the use of imperfect reference standards [21]. Accordingly, the estimates obtained from this study can be considered generalisable to settings with similar MTB burden among PLWHIV.
On Se, Xpert Ultra assay recorded the highest Se of the three tests. This result is in agreement with previous reports that demonstrated the superiority of Xpert Ultra Se over other tests in TB screening and diagnosis among PLWHIV in outpatient settings. In particular, in a systematic review involving studies conducted in high TB prevalence settings among PLWHIV, the Xpert Ultra test Se ranged between 81%–90% [20]. The superiority of the test’s Se owes to its low detection limit of tubercle bacilli (16 CFU/ ml of sputum) [35] compared to SSM whose threshold is 5,000–10,000 CFU/ ml of sputum [36] and LAM at ~ 2000 CFU / ml [37,38].
The LAM test exhibited the lowest Se in the diagnosis of TB in the present study. This performance is comparable with findings from studies carried out in other outpatient settings. For instance, among existing, ART-naïve and newly diagnosed HIV patients presenting to outpatient facilities in LMICs, LAM’s Se was found to range between 25%–30% [17,37,39,40]. Notably, an inverse relationship has been demonstrated, with LAM’s test Se increasing with reducing CD4 counts among critically ill HIV patients [17,41]. LAM Se was observed to be at its highest (≥65%) among patients with CD4 counts <50 cells/ul and Se of35%-55% for patients with CD4 counts between100 and 200 cells/ul respectively [17,41]. Increased uptake of ARV therapy in the present study setting may have contributed to improved CD4 counts in the patient population resulting in a low Se of the LAM test—underscoring its limited utility in this setting. SSM registered a test Se of 56.7% which falls within the 18%–94.2% range observed in other studies [42,43]. The wide variability in performance is attributable to the number of CFUs available in a sample, with Se increasing with higher CFUs. Moreover, the test Se is also highly dependent on user expertise and other technical and operational factors [43].
As regards Sp, SSM exhibited the highest which is consistent with findings from studies carried out in outpatient settings among PLWHIV [44,45]. The Sp displayed by LAM is also corroborated by findings from other studies carried out in similar low-resource settings where specificities ranging between >90%-100% were registered [37,39,46,47]. Xpert Ultra Sp is in agreement with specificities found in other similar settings where the test Sp was reported to range between 78%- 96% [20,48].
False negative and false positive results compromise Se and Sp estimates respectively. As for LAM and SSM, false negative results are more likely to occur in patients with bacterial load below the tests detection limits [49]. Additionally, LAM is more likely to give false negative results in patients who have already received anti-TB treatment [50]. False positive results by LAM have been demonstrated in samples contaminated with non-Mycobacterial pathogens such as Nocardia and Candida spp. that similarly exhibit LAM-like glycoprotein antigen [47]. SSM is also unable to distinguish MTB from other smear-positive Mycobacteria spp. [51]. False positives by Xpert Ultra could be attributable to detection of MTB DNA in patients recently cured of TB [45].
Notably, in this population, the three tests’ sensitivities displayed wide credible intervals (signifying low precision of the estimates) since there were few diseased individuals. In contrast, the tests’ Sps demonstrated high precision owing to the significant number of truly non-diseased individuals. Overall, in this low-prevalence TB setting, Xpert Ultra affords good promise for informing treatment and surveillance for TB among PLWHIV.
With low MTB prevalence (4.1%), generally low PPVs but high NPVs would be anticipated for the three tests. These estimates signify a stronger confidence in a negative than a positive test result. Consequently, in this low-prevalence setting where the probability of false positives is highly contemplated, a multiple testing strategy with serial interpretation of the test results may be necessary in order to raise the confidence in a positive test result. This could entail an initial screening with the more sensitive Xpert Ultra test, with any resulting positives followed up with the more specific SSM test. This approach should assure that false positive individuals are not unnecessarily subjected to protracted therapy for TB.
Conclusions
The Xpert Ultra assay registered the highest Se compared to SSM and LAM. However, SSM registered the highest Sp and thus PPV estimate. Nevertheless, the three tests recorded similar NPVs. Owing to the low prevalence of TB in the study setting, an optimal testing approach could entail an initial screening with the more sensitive Xpert Ultra assay, with any resultant positives re-tested with the more specific SSM test—a serial testing strategy—to bolster the overall PPV of a TB control programme.
Acknowledgments
The authors sincerely thank the study participants, the administration and staff at Kibra Community Health Centre for making it possible to realise this research work.
References
- 1.
WHO (2020) World Health Organization annual report 2019 WHO Country Office Lebanon: health for all. Cairo: World Health Organization. Regional Office for the Eastern Mediterranean. 9789290223207 9789290223207.
- 2.
WHO (2017) WHO meeting report of a technical expert consultation: non-inferiority analysis of Xpert MTF/RIF Ultra compared to Xpert MTB/RIF. Geneva: World Health Organization. Geneva.
- 3.
World Health O (2018) Global tuberculosis report 2018. Geneva: World Health Organization. 9789241565646 9789241565646.
- 4. Enos M, Sitienei J, Ong’ang’o J, Mungai B, Kamene M, et al. (2018) Kenya tuberculosis prevalence survey 2016: Challenges and opportunities of ending TB in Kenya. PLoS One 13: e0209098. pmid:30586448
- 5.
MOH (2016) Kenya Tuberculosis Prevalence Survey: Final Survey Report. Nairobi, Kenya.
- 6.
MOH (2013) Guidelines for management of Tuberculosis and Leprosy in Kenya. Nairobi, Kenya.
- 7.
Dawson D, Akhtar M, Bretzel G, Boulahbal F, Fattorini L, et al. (2000) Technical guide sputum examination for Tuberculosis by direct microscopy in low income countries.
- 8. Akanbi MO, Achenbach C, Taiwo B, Idoko J, Ani A, et al. (2017) Evaluation of gene xpert for routine diagnosis of HIV-associated tuberculosis in Nigeria: A prospective cohort study. BMC Pulm Med 17: 87. pmid:28558662
- 9. Gelaw B, Shiferaw Y, Alemayehu M, Bashaw AA (2017) Comparison of loop-mediated isothermal amplification assay and smear microscopy with culture for the diagnostic accuracy of tuberculosis. BMC Infect Dis 17: 79. pmid:28095790
- 10. Behr MA, Warren SA, Salamon H, Hopewell PC, de Leon AP, et al. (1999) Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli. The Lancet 353: 444–449. pmid:9989714
- 11.
Steingart KR, Schiller I, Horne DJ, Pai M, Boehme CC, et al. (2014) Xpert(R) MTB/RIF assay for pulmonary tuberculosis and rifampicin resistance in adults. Cochrane Database Syst Rev: CD009593.
- 12. Chakravorty S S A, Rowneki M H P, Cao Y, Ryan J, Banada PP,, Deshpande S S S, et al. (2017) The new Xpert MTB/RIF Ultra: improving detection of Mycobacterium tuberculosis and resistance to rifampin in an assay suitable for point-of-care testing. American Society for Microbiology Volume 8.
- 13.
Steingart KR, Schiller I, Horne DJ, Pai M, Boehme CC, et al. (2014) Xpert® MTB/RIF assay for pulmonary tuberculosis and rifampicin resistance in adults (Review). JohnWiley & Sons, Ltd.
- 14.
Stephen D. Lawn, Sophie V. Brooks, Kranzer K, MPN, Andrew Whitelaw, et al. (2011) Screening for HIV-Associated Tuberculosis and Rifampicin Resistance before Antiretroviral Therapy Using the Xpert MTB/RIF Assay: A Prospective Study. PLOS MEDICINE 8.
- 15. Peter J, Green C, Hoelscher M, Mwaba P, Zumla A, et al. (2010) Urine for the diagnosis of tuberculosis: current approaches, clinical applicability, and new developments. Curr Opin Pulm Med 16: 262–270. pmid:20375787
- 16. Huerga H, Ferlazzo G, Bevilacqua P, Kirubi B, Ardizzoni E, et al. (2017) Incremental Yield of Including Determine-TB LAM Assay in Diagnostic Algorithms for Hospitalized and Ambulatory HIV-Positive Patients in Kenya. PLoS One 12: e0170976. pmid:28125693
- 17. Drain PK, Losina E, Coleman SM, Giddy J, Ross D, et al. (2014) Diagnostic accuracy of a point-of-care urine test for tuberculosis screening among newly-diagnosed HIV-infected adults: a prospective, clinic-based study. BMC Infect Dis 14: 110. pmid:24571362
- 18.
Shah M, Hanrahan C, Wang ZY, Dendukuri N, Lawn SD, et al. (2016) Lateral flow urine lipoarabinomannan assay for detecting active tuberculosis in HIV-positive adults. Cochrane Database Syst Rev: CD011420.
- 19. Cattamanchi A, Dowdy DW, Davis JL, Worodria W, Yoo S, et al. (2009) Sensitivity of direct versus concentrated sputum smear microscopy in HIV-infected patients suspected of having pulmonary tuberculosis. BMC Infect Dis 9: 53. pmid:19419537
- 20.
Zifodya JS, Kreniske JS, Schiller I, Kohli M, Dendukuri N, et al. (2021) Xpert Ultra versus Xpert MTB/RIF for pulmonary tuberculosis and rifampicin resistance in adults with presumptive pulmonary tuberculosis. Cochrane Database Syst Rev 2: CD009593.
- 21. Enøe Claes, Georgiadis Marios P, Johnson WO (2000) Estimation of sensitivity and specificity of diagnostic tests and disease prevalence when the true disease state is unknown, Preventive Veterinary Medicine. 45: 61–81. pmid:10802334
- 22. Joseph L, Gyorkos TW, Coupal L (1995) Bayesian estimation of disease prevalence and the parameters of diagnostic tests in the absence of a gold standard. Am J Epidemiol 141: 263–272. pmid:7840100
- 23. Beguy D, Elung’ata P, Mberu B, Oduor C, Wamukoya M, et al. (2015) Health & Demographic Surveillance System Profile: The Nairobi Urban Health and Demographic Surveillance System (NUHDSS). Int J Epidemiol 44: 462–471.
- 24. Connor RJ (1987) Sample size for testing differences in proportions for the paired-sample design. Biometrics 43: 207–211. pmid:3567305
- 25. Boehme CC, Nicol MP, Nabeta P, Michael JS, Gotuzzo E, et al. (2011) Feasibility, diagnostic accuracy, and effectiveness of decentralised use of the Xpert MTB/RIF test for diagnosis of tuberculosis and multidrug resistance: a multicentre implementation study. The Lancet 377: 1495–1505. pmid:21507477
- 26. Raja S, Ching J, Xi L, Hughes SJ, Chang R, et al. (2005) Technology for automated, rapid, and quantitative PCR or reverse transcription-PCR clinical testing. Clin Chem 51: 882–890. pmid:15746302
- 27.
Abbott (2010) Determine™ TB LAM Ag SAVING LIVES with early TB diagnosis in people with HIV.
- 28. Lunn D, Spiegelhalter D, Thomas A, Best N (2009) The BUGS project: Evolution, critique and future directions. Stat Med 28: 3049–3067. pmid:19630097
- 29.
Thomas A, O’Hara, R, Ligges, U & Sturtz, (2006) ’Making BUGS open’, R News.,.
- 30. Kostoulas P, Nielsen SS, Branscum AJ, Johnson WO, Dendukuri N, et al. (2017) STARD-BLCM: Standards for the Reporting of Diagnostic accuracy studies that use Bayesian Latent Class Models. Prev Vet Med 138: 37–47. pmid:28237234
- 31.
Ochola EA (2023) Replication Data for: Evaluation of the performance of three diagnostic tests for Tuberculosis among HIV patients presenting to a health facility in an informal urban settlement in Nairobi county, Kenya. DRAFT VERSION ed: Harvard Dataverse.
- 32. Hui SL, Walter SD (1980) Estimating the Error Rates of Diagnostic Tests. Biometrics 36: 167–171. pmid:7370371
- 33.
Andrew Gelman JBC, Hal S. Stern, David B. Dunson, Aki Vehtari, Donald B. Rubin (2014) Bayesian Data Analysis.
- 34. Schisterman EF, Perkins NJ, Liu A, Bondell H (2005) Optimal Cut-Point and Its Corresponding Youden Index to Discriminate Individuals Using Pooled Blood Samples. Epidemiology 16: 73–81. pmid:15613948
- 35. Helb D, Jones M, Story E, Boehme C, Wallace E, et al. (2010) Rapid detection of Mycobacterium tuberculosis and rifampin resistance by use of on-demand, near-patient technology. J Clin Microbiol 48: 229–237. pmid:19864480
- 36. Behr MA, Warren SA, Salamon H, Hopewell PC, Ponce de Leon A, et al. (1999) Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli. Lancet 353: 444–449. pmid:9989714
- 37. Nakiyingi L, Moodley VM, Manabe YC, Nicol MP, Holshouser M, et al. (2014) Diagnostic accuracy of a rapid urine lipoarabinomannan test for tuberculosis in HIV-infected adults. J Acquir Immune Defic Syndr 66: 270–279. pmid:24675585
- 38. Kawasaki M, Echiverri C, Raymond L, Cadena E, Reside E, et al. (2019) Lipoarabinomannan in sputum to detect bacterial load and treatment response in patients with pulmonary tuberculosis: Analytic validation and evaluation in two cohorts. PLoS Med 16: e1002780. pmid:30978194
- 39. Balcha TT W N. Sturegard E. Skogmar S. Reepalu A., Jemal ZH, Tibesso G, et al. (2014) Detection of lipoarabinomannan in urine for identification of active tuberculosis among HIV-positive adults in Ethiopian health centres. Trop Med Int Health 19: 734–742. pmid:24684481
- 40.
Stephen D Lawn ADK, Monica Vogt, Robin Wood (2012) Diagnostic accuracy of a low-cost, urine antigen, point-of-care screening assay for HIV-associated pulmonary tuberculosis before antiretroviral therapy: a descriptive study.
- 41. Esmail Aliasgar a, b Pooran Anil,a,b Sabur Natasha F.,a,c Fadul Mohammed,a,b Brar Mantaj S.,d Oelofse Suzette,a,b, Tomasicchio Michelea, b Dhedaa Keertan,b,e,f (2020) An Optimal Diagnostic Strategy for Tuberculosis in Hospitalized HIV-Infected Patients Using GeneXpert MTB/RIF and Alere Determine TB LAM Ag. Journal of Clinical Microbiology. pmid:32727831
- 42.
Jean Claude Semuto Ngabonziza WS, Florence Mutua, Gabriela Torrea, Augustin Dushime,, Michel Gasana EA, Schifra Uwamungu, Alaine Umubyeyi Nyaruhirira, Dufton Mwaengo and, Muvunyi CM (2016) Diagnostic performance of smear microscopy and incremental yield of Xpert in detection of pulmonary tuberculosis in Rwanda. BMC Infect Dis.
- 43.
Mecky Matee LM, Tarja Lounasvaara, Wendy Wieland-Alter,, Richard Waddell JL, Muhammad Bakari,, Reyn KPaCFv (2008) Sputum microscopy for the diagnosis of HIV-associated pulmonary tuberculosis in Tanzania.
- 44. Boehme Catharina C. MD, Nabeta Pamela, M.D., Hillemann Doris, Ph.D., Nicol Mark P.Ph.D.,, Shenai Shubhada PD, Krapp Fiorella, M.D., et al. (2010) Rapid Molecular Detection of Tuberculosis and Rifampin Resistance. The New England Journal of Medicine 363. pmid:20825313
- 45. Balcells ME, García P, Chanqueo L, Bahamondes L, Lasso M, et al. (2012) Rapid molecular detection of pulmonary tuberculosis in HIV-infected patients in Santiago, Chile. Int J Tuberc Lung Dis 16: 1349–1353. pmid:22863872
- 46.
Paul K. Drain M, MPH1,2, Elena Losina, PhD1,3,4, Sharon M. Coleman, MS6, Janet Giddy, MBBCh7, Douglas Ross, MBBCh, MBA8, Jeffrey N. Katz, MD4, and Ingrid V. Bassett, MD, MPH (2015) Value of Urine Lipoarabinomannan Grade and Second Test for Optimizing Clinic-based Screening for HIV-associated Pulmonary Tuberculosis.
- 47. Peter J, Theron G, Chanda D, Clowes P, Rachow A, et al. (2015) Test characteristics and potential impact of the urine LAM lateral flow assay in HIV-infected outpatients under investigation for TB and able to self-expectorate sputum for diagnostic testing. BMC Infect Dis 15: 262. pmid:26156025
- 48. Berhanu RH, David A, da Silva P, Shearer K, Sanne I, et al. (2018) Performance of Xpert MTB/RIF, Xpert Ultra, and Abbott RealTime MTB for Diagnosis of Pulmonary Tuberculosis in a High-HIV-Burden Setting. Journal of Clinical Microbiology 56.
- 49. Saeed M, Rasheed F, Iram S, Hussain S, Ahmad A, et al. (2018) False Negativity of Ziehl-Neelsen Smear Microscopy: Is the Scale-up the Worth It in Developing Countries? J Coll Physicians Surg Pak 28: 201–205. pmid:29544576
- 50. Nel JS, Lippincott CK, Berhanu R, Spencer DC, Sanne IM, et al. (2017) Does Disseminated Nontuberculous Mycobacterial Disease Cause False-Positive Determine TB-LAM Lateral Flow Assay Results? A Retrospective Review. Clin Infect Dis 65: 1226–1228. pmid:28575238
- 51. Oluwole Akanbi Maxwell, Achenbach Chad, Taiwo Babafemi, Idoko John, Ani Agatha, et al. (2017) Evaluation of gene xpert for routine diagnosis of HIV-associated tuberculosis in Nigeria: A prospective cohort study. BMC Pulmonary medicine.