Evaluation of a population-wide, systematic screening initiative for tuberculosis on Daru island, Western Province, Papua New Guinea.

BACKGROUND: A population-wide, systematic screening initiative for tuberculosis (TB) was implemented on Daru island in the Western Province of Papua New Guinea, where TB is known to be highly prevalent. The initiative used a mobile van equipped with a digital X-ray device, computer-aided detection (CAD) software to identify TB-related abnormalities on chest radiographs, and GeneXpert machines for follow-on diagnostic testing. We describe the results of the TB screening initiative, evaluate its population-level impact and examine risk factors associated with TB detection. METHODS: Through a retrospective review of screening data, we assessed the effectiveness of the screening by examining the enrolment coverage and the proportion of people with TB among screened subjects. A cascade analysis was performed to illustrate the flow of participants in the screening algorithm. We conducted univariate and multivariate analyses to identify factors associated with TB. Furthermore, we estimated the number of additional cases detected by the project by examining the trend of routine TB case notifications during the intervention period, compared to the historical baseline cases and trend-adjusted expected cases. RESULTS: Of the island's 18,854 residents, 8,085 (42.9%) were enrolled and 7,970 (98.6%) had chest X-ray interpreted by the CAD4TB software. A total of 1,116 (14.0%) participants were considered to have abnormal CXR. A total of 69 Xpert-positive cases were diagnosed, resulting in a detection rate of 853 per 100 000 population screened. 19.4% of people with TB had resistance to rifampicin. People who were in older age groups (aOR 6.6, 95%CI: 1.5-29.1 for the 45-59 age group), were severely underweight (aOR 2.5, 95%CI:1.0-6.1) or underweight (aOR 2.1, 95%CI: 1.1-3.8), lived in households < 5 people (aOR 3.4, 95%CI:1.8-6.6) and had a past history of TB (aOR 2.1, 95%CI: 1.2-3.6) were more likely to have TB. The number of bacteriologically confirmed TB notified during the intervention period was 79.3% and 90.8% higher than baseline notifications and forecasted notifications, respectively. CONCLUSION: The screening project demonstrated its effectiveness with the high Xpert-positive TB prevalence among the participants and by successfully yielding additional cases of bacteriologically confirmed TB including rifampicin-resistant TB. The results and lessons learnt from the project should inform future TB screening initiatives in Papua New Guinea.

Using artificial intelligence to read chest radiographs for tuberculosis detection: A multi-site evaluation of the diagnostic accuracy of three deep learning systems.

Deep learning (DL) neural networks have only recently been employed to interpret chest radiography (CXR) to screen and triage people for pulmonary tuberculosis (TB). No published studies have compared multiple DL systems and populations. We conducted a retrospective evaluation of three DL systems (CAD4TB, Lunit INSIGHT, and qXR) for detecting TB-associated abnormalities in chest radiographs from outpatients in Nepal and Cameroon. All 1196 individuals received a Xpert MTB/RIF assay and a CXR read by two groups of radiologists and the DL systems. Xpert was used as the reference standard. The area under the curve of the three systems was similar: Lunit (0.94, 95% CI: 0.93-0.96), qXR (0.94, 95% CI: 0.92-0.97) and CAD4TB (0.92, 95% CI: 0.90-0.95). When matching the sensitivity of the radiologists, the specificities of the DL systems were significantly higher except for one. Using DL systems to read CXRs could reduce the number of Xpert MTB/RIF tests needed by 66% while maintaining sensitivity at 95% or better. Using a universal cutoff score resulted different performance in each site, highlighting the need to select scores based on the population screened. These DL systems should be considered by TB programs where human resources are constrained, and automated technology is available.

An evaluation of automated chest radiography reading software for tuberculosis screening among public- and private-sector patients.

Computer-aided reading (CAR) of medical images is becoming increasingly common, but few studies exist for CAR in tuberculosis (TB). We designed a prospective study evaluating CAR for chest radiography (CXR) as a triage tool before Xpert MTB/RIF (Xpert).Consecutively enrolled adults in Dhaka, Bangladesh, with TB symptoms received CXR and Xpert. Each image was scored by CAR and graded by a radiologist. We compared CAR with the radiologist for sensitivity and specificity, area under the receiver operating characteristic curve (AUC), and calculated the potential Xpert tests saved.A total of 18 036 individuals were enrolled. TB prevalence by Xpert was 15%. The radiologist graded 49% of CXRs as abnormal, resulting in 91% sensitivity and 58% specificity. At a similar sensitivity, CAR had a lower specificity (41%), saving fewer (36%) Xpert tests. The AUC for CAR was 0.74 (95% CI 0.73-0.75). CAR performance declined with increasing age. The radiologist grading was superior across all sub-analyses.Using CAR can save Xpert tests, but the radiologist's specificity was superior. Differentiated CAR thresholds may be required for different populations. Access to, and costs of, human readers must be considered when deciding to use CAR software. More studies are needed to evaluate CAR using different screening approaches.

The burden of non-communicable disease in transition communities in an Asian megacity: baseline findings from a cohort study in Karachi, Pakistan.

BACKGROUND: The demographic transition in South Asia coupled with unplanned urbanization and lifestyle changes are increasing the burden of non-communicable disease (NCD) where infectious diseases are still highly prevalent. The true magnitude and impact of this double burden of disease, although predicted to be immense, is largely unknown due to the absence of recent, population-based longitudinal data. The present study was designed as a unique 'Framingham-like' Pakistan cohort with the objective of measuring the prevalence and risk factors for hypertension, obesity, diabetes, coronary artery disease and hepatitis B and C infection in a multi-ethnic, middle to low income population of Karachi, Pakistan. METHODS: We selected two administrative areas from a private charitable hospital's catchment population for enrolment of a random selection of cohort households in Karachi, Pakistan. A baseline survey measured the prevalence and risk factors for hypertension, obesity, diabetes, coronary artery disease and hepatitis B and C infection. RESULTS: Six hundred and sixty-seven households were enrolled between March 2010 and August 2011. A majority of households lived in permanent structures (85%) with access to basic utilities (77%) and sanitation facilities (98%) but limited access to clean drinking water (68%). Households had high ownership of communication technologies in the form of cable television (69%) and mobile phones (83%). Risk factors for NCD, such as tobacco use (45%), overweight (20%), abdominal obesity (53%), hypertension (18%), diabetes (8%) and pre-diabetes (40%) were high. At the same time, infectious diseases such as hepatitis B (24%) and hepatitis C (8%) were prevalent in this population. CONCLUSION: Our findings highlight the need to monitor risk factors and disease trends through longitudinal research in high-burden transition communities in the context of rapid urbanization and changing lifestyles. They also demonstrate the urgency of public health intervention programs tailored for these transition communities.