Computer-aided X-ray screening for tuberculosis and HIV testing among adults with cough in Malawi (the PROSPECT study): A randomised trial and cost-effectiveness analysis

Artificial Intelligence-based automated CT brain interpretation to accelerate treatment for acute stroke in rural India: An interrupted time series study

In resource-limited settings, timely treatment of acute stroke is dependent upon accurate diagnosis that draws on non-contrast computed tomography (NCCT) scans of the head. Artificial Intelligence (AI) based devices may be able to assist non-specialist physicians in NCCT interpretation, thereby enabling faster interventions for acute stroke patients in these settings. We evaluated the impact of an AI device by comparing the time to intervention (TTI) from NCCT imaging to significant intervention before (baseline) and after the deployment of AI, in patients diagnosed with stroke (ischemic or hemorrhagic) through a retrospective interrupted time series analysis at a rural hospital managed by non-specialists in India. Significant intervention was defined as thrombolysis or antiplatelet therapy in ischemic strokes, and mannitol for hemorrhagic strokes or mass effect. We also evaluated the diagnostic accuracy of the software using the teleradiologists' reports as ground truth. Impact analysis in a total of 174 stroke patients (72 in baseline and 102 after deployment) demonstrated a significant reduction of median TTI from 80 minutes (IQR: 56·8-139·5) during baseline to 58·50 (IQR: 30·3-118.2) minutes after AI deployment (Wilcoxon rank sum test-location shift: -21·0, 95% CI: -38·0, -7·0). Diagnostic accuracy evaluation in a total of 864 NCCT scans demonstrated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) in detecting intracranial hemorrhage to be 0·89 (95% CI: 0·83-0·93), 0·99 (0·98-1·00), 0·96 (0·91-0·98) and 0·97 (0·96-0·98) respectively, and for infarct these were 0·84 (0·79-0·89), 0·81 (0·77-0·84), 0·58 (0·52-0·63), and 0·94 (0·92-0·96), respectively. AI-based NCCT interpretation supported faster abnormality detection with high accuracy, resulting in persons with acute stroke receiving significantly earlier treatment. Our results suggest that AI-based NCCT interpretation can potentially improve uptake of lifesaving interventions for acute stroke in resource-limited settings.

Head-to-head comparison of diagnostic accuracy of TB screening tests: Chest-X-ray, Xpert TB host response, and C-reactive protein

Background

Accessible, accurate screening tests are necessary to advance tuberculosis (TB) case finding and early detection in high-burden countries. We compared the diagnostic accuracy of available TB triage tests.

Methods

We prospectively screened consecutive adults with ≥2 weeks of cough presenting to primary health centers in the Philippines, Vietnam, South Africa, Uganda, and India. All participants received the index tests: chest-X-ray (CXR), venous or capillary Cepheid Xpert TB Host Response (HR) testing, and point-of-care C-reactive protein (CRP) testing (Boditech iChroma II). CXR images were processed using computer-aided detection (CAD) algorithms. We assessed diagnostic accuracy against a microbiologic reference standard (sputum Xpert Ultra, culture). Optimal cut-points were chosen to achieve sensitivity ≥90% and maximize specificity. Two-test screening algorithms were considered, using two approaches: 1) sequential negative serial screening in which the second screening test is conducted only if the first is negative and positive is defined as positive on either test and 2) sequential positive serial screening, in which the second screening test is conducted only if the first is positive and positive is defined as positive on both tests.

Results

Between July 2021 and August 2022, 1,392 participants with presumptive TB had valid results on index tests and the reference standard, and 303 (22%) had confirmed TB. In head-to-head comparisons, CAD4TB v7 showed the highest specificity when using a cut-point that achieves 90% sensitivity (70.3% vs. 65.1% for Xpert HR, difference 95% CI 1.6 to 8.9; 49.7% for CRP, difference 95% CI 17.0 to 24.3). Among the possible two-test screening algorithms, three met WHO target product profile (TPP) minimum accuracy thresholds and had higher accuracy than any test alone. At 90% sensitivity, the specificity was 79.6% for Xpert HR-CAD4TB [sequential negative], 75.9% for CRP-CAD4TB [sequential negative], and 73.7% for Xpert HR-CAD4TB [sequential positive].

Conclusions

CAD4TB achieves TPP targets and outperforms Xpert HR and CRP. Combining screening tests further increased accuracy. Cost and feasibility of two-test screening algorithms should be explored.

Registration

NCT04923958.

Missed opportunities for TB diagnostic testing among people living with HIV in Zimbabwe: Cross-sectional analysis of the Zimbabwe Population-based HIV Impact Assessment (ZIMPHIA) survey 2015-16

Background

Using data from the Zimbabwe Population-based HIV Impact Assessment survey 2015-2016, we examined the TB care cascade and factors associated with not receiving TB diagnostic testing among adult PLHIV with TB symptoms.

Methods

Statistical Analysis was limited to PLHIV aged 15 years and older in HIV care. Weighted logistic regression with not receiving TB testing as outcome was adjusted for covariates with crude odd ratios (ORs) with p < 0.25. All analyses accounted for multistage survey design.

Results

Among 3507 adult PLHIV in HIV care, 2288 (59.7 %, 95 % CI:58.1-61.3) were female and 2425 (63.6 %, 95 % CI:61.1-66.1) lived in rural areas. 1197(48.7 %, 95 % CI:46.5-51.0) reported being screened for TB symptoms at their last HIV care visit. In the previous 12 months, 639 (26.0 %, 95 % CI:23.9-28.1) reported having symptoms and of those, 239 (37.8 %, 95 % CI:33.3-42.2) received TB testing. Of PLHIV tested for TB, 36 (49.5 %, 95 % CI:35.0-63.1) were diagnosed with TB; 32 (90.3 %, 95 % CI:78.9-100) of those diagnosed with TB received treatment. Never having used IPT was associated with not receiving TB testing.

Conclusion

The results suggest suboptimal utilization of TB screening and diagnostic testing among PLHIV. New approaches are needed to reach opportunities missed in the HIV/TB integrated services.

Outcomes for people with TB by disease severity at presentation

BACKGROUND

There is substantial heterogeneity in disease presentation for individuals with TB disease, which may correlate with disease outcomes. We estimated disease outcomes by disease severity at presentation among individuals with TB during the pre-chemotherapy era.

METHODS

We extracted data on people with TB enrolled between 1917 and 1948 in the USA, stratified by three disease severity categories at presentation using the U.S. National Tuberculosis Association diagnostic criteria. These criteria were based largely on radiographic findings ("minimal", "moderately advanced", and "far advanced"). We used Bayesian parametric survival analysis to model the survival distribution overall, and by disease severity and Bayesian logistic regression to estimate the severity-level specific natural recovery odds within 3 years.

RESULTS

People with minimal TB at presentation had a 2% (95% CrI 0-11%) probability of TB death within 5 years vs. 40% (95% CrI 15-68) for those with far advanced disease. Individuals with minimal disease had 13.62 times the odds (95% CrI 9.87-19.10) of natural recovery within 3 years vs. those with far advanced disease.

CONCLUSION

Mortality and natural recovery vary by disease severity at presentation. This supports continued work to evaluate individualized (e.g., shortened or longer) regimens based on disease severity at presentation, identified using radiography..

Concordance of randomised controlled trials for artificial intelligence interventions with the CONSORT-AI reporting guidelines

The Consolidated Standards of Reporting Trials extension for Artificial Intelligence interventions (CONSORT-AI) was published in September 2020. Since its publication, several randomised controlled trials (RCTs) of AI interventions have been published but their completeness and transparency of reporting is unknown. This systematic review assesses the completeness of reporting of AI RCTs following publication of CONSORT-AI and provides a comprehensive summary of RCTs published in recent years. 65 RCTs were identified, mostly conducted in China (37%) and USA (18%). Median concordance with CONSORT-AI reporting was 90% (IQR 77-94%), although only 10 RCTs explicitly reported its use. Several items were consistently under-reported, including algorithm version, accessibility of the AI intervention or code, and references to a study protocol. Only 3 of 52 included journals explicitly endorsed or mandated CONSORT-AI. Despite a generally high concordance amongst recent AI RCTs, some AI-specific considerations remain systematically poorly reported. Further encouragement of CONSORT-AI adoption by journals and funders may enable more complete adoption of the full CONSORT-AI guidelines.

Impact of active case-finding for tuberculosis on case-notifications in Blantyre, Malawi: A community-based cluster-randomised trial (SCALE)

Active case-finding (ACF) for tuberculosis can help find the "missing millions" with undiagnosed tuberculosis. In a cluster-randomised trial, we investigated impact of ACF on case-notifications in Blantyre, Malawi, where ACF has been intensively implemented following 2014 estimates of ~1,000 per 100,000 adults with undiagnosed TB. Following a pre-intervention prevalence survey (May 2019 to March 2020), constrained randomisation allocated neighbourhoods to either door-to-door ACF (sputum microscopy for reported cough >2 weeks) or standard-of-care (SOC). Implementation was interrupted by COVID-19. Cluster-level bacteriologically-confirmed case-notification rate (CNR) ratio within 91 days of ACF was our redefined primary outcome; comparison between arms used Poisson regression with random effects. Secondary outcomes were 91-day CNR ratios comparing all tuberculosis registrations and all non-ACF registrations. Interrupted time series (ITS) analysis of CNRs in the SOC arm examined prevalence survey impact. (ISRCTN11400592). 72 clusters served by 10 study-supported tuberculosis registration centres were randomised to ACF (261,244 adults, 58,944 person-years follow-up) or SOC (256,713 adults, 52,805 person-years). Of 1,192 ACF participants, 13 (1.09%) were smear-positive. Within 91 days, 113 (42 bacteriologically-confirmed) and 108 (33 bacteriologically-confirmed) tuberculosis patients were identified as ACF or SOC cluster residents, respectively. There was no difference by arm, with adjusted 91-day CNR ratios 1.12 (95% CI: 0.61-2.07) for bacteriologically-confirmed tuberculosis; 0.93 (95% CI: 0.68-1.28) for all tuberculosis registrations; and 0.86 (95%CI: 0.63-1.16) for non-ACF (routinely) diagnosed. Of 7,905 ACF and 7,992 SOC pre-intervention survey participants, 12 (0.15%) and 17 (0.21%), respectively, had culture/Xpert-confirmed tuberculosis. ITS analysis showed no survey impact on SOC CNRs. Despite residual undiagnosed tuberculosis of 150 per 100,000 population, there was no increase in tuberculosis notifications from this previously successful approach targeting symptomatic disease, likely due to previous TB ACF and rapid declines in TB burden. In such settings, future ACF should focus on targeted outreach and demand creation, alongside optimised facility-based screening. Trial Registration: ISRCTN11400592.

Prevalence of bacteriologically-confirmed pulmonary tuberculosis in urban Blantyre, Malawi 2019-20: Substantial decline compared to 2013-14 national survey

Recent evidence shows rapidly changing tuberculosis (TB) epidemiology in Southern and Eastern Africa, with need for subdistrict prevalence estimates to guide targeted interventions. We conducted a pulmonary TB prevalence survey to estimate current TB burden in Blantyre city, Malawi. From May 2019 to March 2020, 115 households in middle/high-density residential Blantyre, were randomly-selected from each of 72 clusters. Consenting eligible participants (household residents ≥ 18 years) were interviewed, including for cough (any duration), and offered HIV testing and chest X-ray; participants with cough and/or abnormal X-ray provided two sputum samples for microscopy, Xpert MTB/Rif and mycobacterial culture. TB disease prevalence and risk factors for prevalent TB were calculated using complete-case analysis, multiple imputation, and inverse probability weighting. Of 20,899 eligible adults, 15,897 (76%) were interviewed, 13,490/15,897 (85%) had X-ray, and 1,120/1,394 (80%) sputum-eligible participants produced at least one specimen, giving 15,318 complete cases (5,895, 38% men). 29/15,318 had bacteriologically-confirmed TB (189 per 100,000 complete-case (cc) / 150 per 100,000 with inverse weighting (iw)). Men had higher burden (cc: 305 [95% CI:144-645] per 100,000) than women (cc: 117 [95% CI:65-211] per 100,000): cc adjusted odds ratio (aOR) 2.70 (1.26-5.78). Other significant risk factors for prevalent TB on complete-case analysis were working age (25-49 years) and previous TB treatment, but not HIV status. Multivariable analysis of imputed data was limited by small numbers, but previous TB and age group 25-49 years remained significantly associated with higher TB prevalence. Pulmonary TB prevalence for Blantyre was considerably lower than the 1,014 per 100,000 for urban Malawi in the 2013-14 national survey, at 150-189 per 100,000 adults, but some groups, notably men, remain disproportionately affected. TB case-finding is still needed for TB elimination in Blantyre, and similar urban centres, but should focus on reaching the highest risk groups, such as older men.

Economic evaluations of artificial intelligence-based healthcare interventions: a systematic literature review of best practices in their conduct and reporting

Objectives: Health economic evaluations (HEEs) help healthcare decision makers understand the value of new technologies. Artificial intelligence (AI) is increasingly being used in healthcare interventions. We sought to review the conduct and reporting of published HEEs for AI-based health interventions. Methods: We conducted a systematic literature review with a 15-month search window (April 2021 to June 2022) on 17th June 2022 to identify HEEs of AI health interventions and update a previous review. Records were identified from 3 databases (Medline, Embase, and Cochrane Central). Two reviewers screened papers against predefined study selection criteria. Data were extracted from included studies using prespecified data extraction tables. Included studies were quality assessed using the National Institute for Health and Care Excellence (NICE) checklist. Results were synthesized narratively. Results: A total of 21 studies were included. The most common type of AI intervention was automated image analysis (9/21, 43%) mainly used for screening or diagnosis in general medicine and oncology. Nearly all were cost-utility (10/21, 48%) or cost-effectiveness analyses (8/21, 38%) that took a healthcare system or payer perspective. Decision-analytic models were used in 16/21 (76%) studies, mostly Markov models and decision trees. Three (3/16, 19%) used a short-term decision tree followed by a longer-term Markov component. Thirteen studies (13/21, 62%) reported the AI intervention to be cost effective or dominant. Limitations tended to result from the input data, authorship conflicts of interest, and a lack of transparent reporting, especially regarding the AI nature of the intervention. Conclusion: Published HEEs of AI-based health interventions are rapidly increasing in number. Despite the potentially innovative nature of AI, most have used traditional methods like Markov models or decision trees. Most attempted to assess the impact on quality of life to present the cost per QALY gained. However, studies have not been comprehensively reported. Specific reporting standards for the economic evaluation of AI interventions would help improve transparency and promote their usefulness for decision making. This is fundamental for reimbursement decisions, which in turn will generate the necessary data to develop flexible models better suited to capturing the potentially dynamic nature of AI interventions.

Chest X-ray predicts cases of pulmonary tuberculosis among women of reproductive age with acute respiratory symptoms: A multi-center cross-sectional study

Background

Tuberculosis (TB) prevalence is increasing among women of reproductive age (WRA) in sub-Saharan Africa, yet undiagnosed and untreated cases remain rather high with serious health and socio-economic consequences. We aimed to assess the prevalence and predictors of TB in WRA seeking health care for acute respiratory symptoms.

Methods

We consecutively enrolled outpatient WRA with acute respiratory symptoms seeking care at four healthcare facilities in Ethiopia between July 2019 and December 2020. Data on sociodemographic characteristics and clinical information were collected using a structured questionnaire administered by trained nurses. Posteroanterior chest X-ray was performed in non-pregnant WRA and interpreted independently by two radiologists. Sputum samples were collected from all patients and tested for pulmonary TB using Xpert MTB/RIF and/or smear microscopy. Predictors of bacteriologically confirmed TB cases were determined using binary logistic regression, with clinically relevant variables included in the final Firth's multivariate-penalized logistic regression model.

Results

We enrolled 577 participants, of whom 95 (16%) were pregnant, 67 (12%) were living with HIV, 512 (89%) had cough of less than 2 weeks, and 56 (12%) had chest-x-ray findings suggestive of TB. The Overall prevalence of TB was 3% (95% CI: 1.8%-4.7%) with no significant difference observed between patient groups categorized by duration of cough or HIV serostatus (P-value = 0.9999). In multivariable analysis, TB-suggestive CXR abnormality (AOR 18.83 [95% CI, 6.20-57.18]) and history of weight loss (AOR 3.91 [95% CI, 1.25-12.29]) were associated with bacteriologically-confirmed TB cases.

Conclusions

We found a high TB prevalence among low-risk women of reproductive age with acute respiratory symptoms. Routine CXR may improve early case detection and thereby TB treatment outcomes.

Economic analysis of different throughput scenarios and implementation strategies of computer-aided detection software as a screening and triage test for pulmonary TB

BACKGROUND

Artificial Intelligence (AI) systems have demonstrated potential in detecting tuberculosis (TB) associated abnormalities from chest X-ray (CXR) images. Thus, they might provide a solution to radiologist shortages in high TB burden countries. However, the cost of implementing computer-aided detection (CAD) software has thus far been understudied. In this study, we performed a costing analysis of CAD software when used as a screening or triage test for pulmonary TB, estimated the incremental cost compared to a radiologist reading of different throughput scenarios, and predicted the cost for the national scale-up plan in Pakistan.

METHODS

For the study, we focused on CAD software reviewed by the World Health Organization (CAD4TB, Lunit INSIGHT CXR, qXR) or listed in the Global Drug Facility diagnostics catalogue (CAD4TB, InferRead). Costing information was obtained from the CAD software developers. CAD4TB and InferRead use a perpetual license pricing model, while Lunit and qXR are priced per license for restricted number of scans. A major implementer in Pakistan provided costing information for human resource and software training. The per-screen cost was estimated for each CAD software and for radiologist for 1) active case finding, and 2) facility based CXR testing scenarios with throughputs ranging from 50,000-100,000 scans. Moreover, we estimated the scale-up cost for CAD or radiologist CXR reading in Pakistan based on the National Strategic Plan, considering that to reach 80% diagnostic coverage, 50% of TB patients would need to be found through facility-based triage and 30% through active case finding (ACF).

RESULTS

The per-screen cost for CAD4TB (0.25 USD- 2.33 USD) and InferRead (0.19 USD- 2.78 USD) was lower than that of a radiologist (0.70 USD- 0.93 USD) for high throughput scenarios studied. In comparison, the per-screen cost for Lunit (0.94 USD- 1.69 USD) and qXR (0.95 USD-1.9 USD) were only comparable with that of the radiologists in the highest throughput scenario in ACF. To achieve 80 percent diagnostic coverage at scale in Pakistan, the projected additional cost of deploying CAD software to complement the current infrastructure over a four-year period were estimated at 2.65-19.23 million USD, whereas Human readers, would cost an additional 23.97 million USD.

CONCLUSIONS

Our findings suggest that using CAD software could enable large-scale screening programs in high TB-burden countries and be less costly than radiologist. To achieve minimum cost, the target number of screens in a specific screening strategy should be carefully considered when selecting CAD software, along with the offered pricing structure and other aspects such as performance and operational features. Integrating CAD software in implementation strategies for case finding could be an economical way to attain the intended programmatic goals.

Tuberculosis in pregnancy

Due to COVID-19 pandemic, the latest progress of the End Tuberculosis (TB) Strategy was far from optimal and services for TB needs to be quickly restored. Pregnancy is a unique opportunity to screen and manage TB, and it is an essential step in TB eradication. Early diagnosis and treatment for active disease can reduce maternal and neonatal morbidities and mortality. The more widespread utilization of newer rapid molecular assays with drug-susceptibility testing has significantly shortened the diagnostic process for active TB disease. First-line anti-TB drugs are proven to be safe in pregnancy. Management of latent TB infection (LTBI) during pregnancy is controversial, but puerperium is a period of increased susceptibility to progress to active disease. Extrapulmonary TB (EPTB), multidrug-resistant TB (MDR-TB) and HIV co-infection remain significant issues surrounding TB management during pregnancy and often require input from a multidisciplinary team including TB experts.

Artificial intelligence for strengthening healthcare systems in low- and middle-income countries: a systematic scoping review

In low- and middle-income countries (LMICs), AI has been promoted as a potential means of strengthening healthcare systems by a growing number of publications. We aimed to evaluate the scope and nature of AI technologies in the specific context of LMICs. In this systematic scoping review, we used a broad variety of AI and healthcare search terms. Our literature search included records published between 1st January 2009 and 30th September 2021 from the Scopus, EMBASE, MEDLINE, Global Health and APA PsycInfo databases, and grey literature from a Google Scholar search. We included studies that reported a quantitative and/or qualitative evaluation of a real-world application of AI in an LMIC health context. A total of 10 references evaluating the application of AI in an LMIC were included. Applications varied widely, including: clinical decision support systems, treatment planning and triage assistants and health chatbots. Only half of the papers reported which algorithms and datasets were used in order to train the AI. A number of challenges of using AI tools were reported, including issues with reliability, mixed impacts on workflows, poor user friendliness and lack of adeptness with local contexts. Many barriers exists that prevent the successful development and adoption of well-performing, context-specific AI tools, such as limited data availability, trust and evidence of cost-effectiveness in LMICs. Additional evaluations of the use of AI in healthcare in LMICs are needed in order to identify their effectiveness and reliability in real-world settings and to generate understanding for best practices for future implementations.

Early risk assessment in paediatric and adult household contacts of confirmed tuberculosis cases by novel diagnostic tests (ERASE-TB): protocol for a prospective, non-interventional, longitudinal, multicountry cohort study

INTRODUCTION

The WHO End-TB Strategy calls for the development of novel diagnostics to detect tuberculosis (TB) earlier and more accurately. Better diagnostics, together with tools to predict disease progression, are critical for achieving WHO End-TB targets. The Early Risk Assessment in TB Contacts by new diagnoStic tEsts (ERASE-TB) study aims to evaluate novel diagnostics and testing algorithms for early TB diagnosis and accurate prediction of disease progression among household contacts (HHCs) exposed to confirmed index cases in Mozambique, Tanzania and Zimbabwe.

METHODS AND ANALYSIS

A total of 2100 HHCs (aged ≥10 years) of adults with microbiologically-confirmed pulmonary TB will be recruited and followed up at 6-month intervals for 18-24 months. At each time point, a WHO symptom screen and digital chest radiograph (dCXR) will be performed, and blood and urine samples will be collected. Individuals screening positive (WHO symptom screen or dCXR) will be requested to provide sputum for Xpert MTB/Rif Ultra. At baseline, HHCs will also be screened for HIV, diabetes (HbA1c), chronic lung disease (spirometry), hypertension and anaemia. Study outcomes will be coprevalent TB (diagnosed at enrolment), incident TB (diagnosed during follow-up) or no TB at completion of follow-up. Novel diagnostics will be validated using fresh and biobanked samples with a nested case-control design. Cases are defined as HHCs diagnosed with TB (for early diagnosis) or with incident TB (for prediction of progression) and will be matched by age, sex and country to HHCs who remain healthy (controls). Statistical analyses will include assessment of diagnostic accuracy by constructing receiver operating curves and calculation of sensitivity and specificity.

ETHICS AND DISSEMINATION

ERASE-TB has been approved by regulatory and ethical committees in each African country and by each partner organisation. Consent, with additional assent for participants <18 years, is voluntary. Attestation by impartial witnesses is sought in case of illiteracy. Confidentiality of participants is being maintained throughout. Study findings will be presented at scientific conferences and published in peer-reviewed international journals.

TRIAL REGISTRATION NUMBER

NCT04781257.Cite Now.

Pooled testing of sputum with Xpert MTB/RIF and Xpert Ultra during tuberculosis active case finding campaigns in Lao People's Democratic Republic

INTRODUCTION

Active case finding (ACF) of individuals with tuberculosis (TB) is a key intervention to find the 30% of people missed every year. However, ACF requires screening large numbers of individuals who have a low probability of positive results, typically <5%, which makes using the recommended molecular tests expensive.

METHODS

We conducted two ACF surveys (in 2020 and 2021) in high TB burden areas of Lao PDR. Participants were screened for TB symptoms and received a chest X-ray. Sputum samples of four consecutive individuals were pooled and tested with Xpert Mycobacterium tuberculosis (MTB)/rifampicin (RIF) (Xpert-MTB/RIF) (2020) or Xpert-Ultra (2021). The agreement of the individual and pooled samples was compared and the reasons for discrepant results and potential cartridge savings were assessed.

RESULTS

Each survey included 436 participants, which were tested in 109 pools. In the Xpert-MTB/RIF survey, 25 (sensitivity 89%, 95% CI 72.8% to 96.3%) of 28 pools containing MTB-positive samples tested positive and 81 pools containing only MTB-negative samples tested negative (specificity 100%, 95% CI 95.5% to 100%). In the Xpert-Ultra survey, all 32 (sensitivity 100%, 95% CI 89.3% to 100%) pools containing MTB-positive samples tested positive and all 77 (specificity 100%, 95% CI 95.3% to 100%) containing only MTB-negative samples tested negative. Pooling with Xpert-MTB/RIF and Xpert-Ultra saved 52% and 46% (227/436 and 199/436, respectively) of cartridge costs alone.

CONCLUSION

Testing single and pooled specimens had a high level of agreement, with complete concordance when using Xpert-Ultra. Pooling samples could generate significant cartridge savings during ACF campaigns.

Accuracy of computer-aided chest X-ray in community-based tuberculosis screening: Lessons from the 2016 Kenya National Tuberculosis Prevalence Survey

Community-based screening for tuberculosis (TB) could improve detection but is resource intensive. We set out to evaluate the accuracy of computer-aided TB screening using digital chest X-ray (CXR) to determine if this approach met target product profiles (TPP) for community-based screening. CXR images from participants in the 2016 Kenya National TB Prevalence Survey were evaluated using CAD4TBv6 (Delft Imaging), giving a probabilistic score for pulmonary TB ranging from 0 (low probability) to 99 (high probability). We constructed a Bayesian latent class model to estimate the accuracy of CAD4TBv6 screening compared to bacteriologically-confirmed TB across CAD4TBv6 threshold cut-offs, incorporating data on Clinical Officer CXR interpretation, participant demographics (age, sex, TB symptoms, previous TB history), and sputum results. We compared model-estimated sensitivity and specificity of CAD4TBv6 to optimum and minimum TPPs. Of 63,050 prevalence survey participants, 61,848 (98%) had analysable CXR images, and 8,966 (14.5%) underwent sputum bacteriological testing; 298 had bacteriologically-confirmed pulmonary TB. Median CAD4TBv6 scores for participants with bacteriologically-confirmed TB were significantly higher (72, IQR: 58-82.75) compared to participants with bacteriologically-negative sputum results (49, IQR: 44-57, p<0.0001). CAD4TBv6 met the optimum TPP; with the threshold set to achieve a mean sensitivity of 95% (optimum TPP), specificity was 83.3%, (95% credible interval [CrI]: 83.0%-83.7%, CAD4TBv6 threshold: 55). There was considerable variation in accuracy by participant characteristics, with older individuals and those with previous TB having lowest specificity. CAD4TBv6 met the optimal TPP for TB community screening. To optimise screening accuracy and efficiency of confirmatory sputum testing, we recommend that an adaptive approach to threshold setting is adopted based on participant characteristics.

Isoniazid or rifampicin preventive therapy with and without screening for subclinical TB: a modeling analysis

BACKGROUND

Short-course, rifamycin-based regimens could facilitate scale-up of tuberculosis preventive therapy (TPT), but it is unclear how stringently tuberculosis (TB) disease should be ruled out before TPT use.

METHODS

We developed a state-transition model of a TPT intervention among two TPT-eligible cohorts: adults newly diagnosed with HIV in South Africa (PWH) and TB household contacts in Pakistan (HHCs). We modeled two TPT regimens-4 months of rifampicin [4R] or 6 months of isoniazid [6H]-comparing each to a reference of no intervention. Before initiating TPT, TB disease was excluded either through symptom-only screening or with additional radiographic screening that could detect subclinical TB but might limit access to the TPT intervention. TPT's potential curative effects on both latent and subclinical TB were modeled, as were both acquisitions of resistance and prevention of drug-resistant disease. Although all eligible individuals received the screening and/or TPT interventions, the modeled TB outcomes comprised only those with latent or subclinical TB that would have progressed to symptomatic disease if untreated.

RESULTS

When prescribed after only symptom-based TB screening (such that individuals with subclinical TB were included among TPT recipients), 4R averted 45 active (i.e., symptomatic) TB cases (95% uncertainty range 24-79 cases or 40-89% of progressions to active TB) per 1000 PWH [17 (9-29, 43-94%) per 1000 HHCs]; 6H averted 37 (19-66, 52-73%) active TB cases among PWH [13 (7-23, 53-75%) among HHCs]. With this symptom-only screening, for each net rifampicin resistance case added by 4R, 12 (3-102) active TB cases were averted among PWH (37 [9-580] among HHCs); isoniazid-resistant TB was also reduced. Similarly, 6H after symptom-only screening increased isoniazid resistance while reducing overall and rifampicin-resistant active TB. Screening for subclinical TB before TPT eliminated this net increase in resistance to the TPT drug; however, if the screening requirement reduced TPT access by more than 10% (the estimated threshold for 4R among HHCs) to 30% (for 6H among PWH), it was likely to reduce the intervention's overall TB prevention impact.

CONCLUSIONS

All modeled TPT strategies prevent TB relative to no intervention, and differences between TPT regimens or between screening approaches are small relative to uncertainty in the outcomes of any given strategy. If most TPT-eligible individuals can be screened for subclinical TB, then pairing such screening with rifamycin-based TPT maximizes active TB prevention and does not increase rifampicin resistance. Where subclinical TB cannot be routinely excluded without substantially reducing TPT access, the choice of TPT regimen requires weighing 4R's efficacy advantages (as well as its greater safety and shorter duration that we did not directly model) against the consequences of rifampicin resistance in a small fraction of recipients.

Computer-aided X-ray screening for tuberculosis and HIV testing among adults with cough in Malawi (the PROSPECT study): A randomised trial and cost-effectiveness analysis

BACKGROUND

Suboptimal tuberculosis (TB) diagnostics and HIV contribute to the high global burden of TB. We investigated costs and yield from systematic HIV-TB screening, including computer-aided digital chest X-ray (DCXR-CAD).

METHODS AND FINDINGS

In this open, three-arm randomised trial, adults (≥18 years) with cough attending acute primary services in Malawi were randomised (1:1:1) to standard of care (SOC); oral HIV testing (HIV screening) and linkage to care; or HIV testing and linkage to care plus DCXR-CAD with sputum Xpert for high CAD4TBv5 scores (HIV-TB screening). Participants and study staff were not blinded to intervention allocation, but investigator blinding was maintained until final analysis. The primary outcome was time to TB treatment. Secondary outcomes included proportion with same-day TB treatment; prevalence of undiagnosed/untreated bacteriologically confirmed TB on day 56; and undiagnosed/untreated HIV. Analysis was done on an intention-to-treat basis. Cost-effectiveness analysis used a health-provider perspective. Between 15 November 2018 and 27 November 2019, 8,236 were screened for eligibility, with 473, 492, and 497 randomly allocated to SOC, HIV, and HIV-TB screening arms; 53 (11%), 52 (9%), and 47 (9%) were lost to follow-up, respectively. At 56 days, TB treatment had been started in 5 (1.1%) SOC, 8 (1.6%) HIV screening, and 15 (3.0%) HIV-TB screening participants. Median (IQR) time to TB treatment was 11 (6.5 to 38), 6 (1 to 22), and 1 (0 to 3) days (hazard ratio for HIV-TB versus SOC: 2.86, 1.04 to 7.87), with same-day treatment of 0/5 (0%) SOC, 1/8 (12.5%) HIV, and 6/15 (40.0%) HIV-TB screening arm TB patients (p = 0.03). At day 56, 2 SOC (0.5%), 4 HIV (1.0%), and 2 HIV-TB (0.5%) participants had undiagnosed microbiologically confirmed TB. HIV screening reduced the proportion with undiagnosed or untreated HIV from 10 (2.7%) in the SOC arm to 2 (0.5%) in the HIV screening arm (risk ratio [RR]: 0.18, 0.04 to 0.83), and 1 (0.2%) in the HIV-TB screening arm (RR: 0.09, 0.01 to 0.71). Incremental costs were US$3.58 and US$19.92 per participant screened for HIV and HIV-TB; the probability of cost-effectiveness at a US$1,200/quality-adjusted life year (QALY) threshold was 83.9% and 0%. Main limitations were the lower than anticipated prevalence of TB and short participant follow-up period; cost and quality of life benefits of this screening approach may accrue over a longer time horizon.

CONCLUSIONS

DCXR-CAD with universal HIV screening significantly increased the timeliness and completeness of HIV and TB diagnosis. If implemented at scale, this has potential to rapidly and efficiently improve TB and HIV diagnosis and treatment.

TRIAL REGISTRATION

clinicaltrials.gov NCT03519425.