.
BACKGROUND:Tuberculosis (TB) remains a leading cause of infectious disease mortality worldwide, and treatment failure contributes to ongoing transmission, drug resistance, and poor clinical outcomes. Artificial intelligence (AI) and machine learning (ML) approaches have attracted growing interest in predicting TB treatment outcomes, but the literature is heterogeneous and lacks a comprehensive synthesis. METHODS:We systematically searched PubMed/MEDLINE and Embase (January 2000-October 2025) for studies developing or validating AI/ML models to predict TB treatment failure. Two reviewers independently screened records and extracted data on study characteristics, predictor modalities, algorithms, validation strategies, and performance metrics, i.e., area under the curve (AUC), sensitivity, specificity, and confidence intervals. Studies reporting AUC with confidence intervals or sufficient data for calculation were included in random-effects meta-analysis. Missing standard errors were estimated from sample sizes and event rates using established methods. Risk of bias was assessed using PROBAST. Subgroup analyses and meta-regression explored heterogeneity, and publication bias was assessed using funnel plots, Egger's test, and trim-and-fill analysis. The study is registered with PROSPERO (CRD420251101443). RESULTS:Thirty-four studies met the inclusion criteria. Publications increased markedly from 2019 onwards (91% of studies). Tree-based methods predominated (52.9%), and multimodal models (≥3 data types) were used in 41.2% of the studies. Nineteen studies (100,790 participants) contributed to the meta-analysis. The pooled AUC was 0.836 (95% CI 0.799-0.868), with substantial heterogeneity (I² = 97.9%). In subgroup analyses, studies including HIV-positive participants showed lower discrimination (AUC 0.748) than those excluding them (0.924). Only eight studies (23.5%) performed external validation, and only one study (2.9%) was rated low risk of bias overall (PROBAST), primarily due to analytical domain deficiencies. Egger's test suggested publication bias (p = 0.024). Major evidence gaps included underrepresentation of high-burden countries, HIV-affected populations, social determinants, pediatric TB, and extrapulmonary disease. CONCLUSIONS:AI/ML models for predicting TB treatment failure show promising discrimination but are not yet ready for routine clinical implementation. Performance varies substantially across populations and settings, and methodological limitations, including inadequate validation, poor calibration assessment, and high risk of bias, limit confidence in current estimates. Future research should prioritize rigorous external validation, calibration assessment, and development in underrepresented populations, particularly HIV-affected and high TB burden settings.
BACKGROUND:Prevention, screening, and diagnostic services for hepatitis B virus (HBV) and hepatitis C virus (HCV) can prevent morbidity and mortality in people receiving HIV care. However, there is limited information about the availability of HBV and HCV services at HIV clinics globally. METHODS:The International epidemiology Databases to Evaluate AIDS (IeDEA) conducted surveys of service delivery and practices at participating HIV treatment centers from 7 regions. We used 2023 survey data to measure availability of HBV vaccination, HBV and HCV screening, HBV surface antigen (HBsAg), HBV DNA, HCV antibody, and HCV RNA testing. Multivariable logistic regression models were used to test associations of site characteristics with HBV and HCV services. RESULTS:HBV vaccination was available on-site at 67.7% of 204 HIV treatment sites. Screening for HBV and HCV at HIV care enrollment was reported by 72.1% and 50% of sites, respectively. HBsAg, HBV DNA, HCV antibody, and HCV RNA testing were available on-site at 77%, 47.6%, 61.8%, and 44.6% of sites, respectively. Sites serving predominately rural (vs. urban) populations were less likely to report on-site availability of HBV DNA [odds ratio (OR):0.07; 95% confidence interval: 0.01 to 0.68; P = 0.02], HCV antibody (OR = 0.18; 95% CI: 0.04 to 0.92; P = 0.04), and HCV RNA (OR = 0.10; 95% CI: 0.01 to 0.90; P = 0.04) testing. CONCLUSION:Life-saving services such as HBV vaccination, HBsAg, and HCV antibody testing were available on-site at most HIV treatment sites participating in the IeDEA network. Lower availability at rural sites suggests that expansion of services is important to eliminate HBV and HCV as public health problems in people receiving HIV care.
Abstract A 28 year old woman presented with local wound infection two weeks after sustaining an anterior thigh laceration in a high energy motorbike crash that killed the co-riding motorcyclist. X-ray revealed a non-identifiable radio-opaque fragment and computed tomography identified a bony foreign body deep in the thigh adjacent to the femoral vessels, with the host femur intact. A 6 x 3 cm fragment of cortical bone was removed at surgery. Histopathology and genetic studies to confirm its origin were not feasible in this resource-limited setting, but an intact host femur suggested it originated from the deceased rider. The wound healed by secondary intention, HIV and hepatitis serology were negative at 6 weeks. This case highlights the unusual occurrence of a bone missile and the importance of early advanced imaging and prompt post-exposure prophylaxis following high-energy, multi-victim trauma.
Background: Non-operative management is a common approach for treating closed ankle fractures at Mulago National Referral Hospital (MNRH) due to its potential benefits, including shorter hospital stays and the absence of surgical risks. Evaluating functional and radiological outcomes is essential for determining the success of this treatment. This study aimed to evaluate the functional and radiological outcomes of ankle fractures managed non-operatively and to identify the factors associated with these outcomes. Methods: This cross-sectional study assessed 93 adults with ankle fractures managed non-operatively at six months post-injury at Mulago National Referral Hospital. Functional outcomes were evaluated using the AOFAS ankle-hindfoot score, while radiological outcomes included fracture union and ankle alignment parameters. Data were analyzed using STATA, and modified Poisson regression was applied to identify factors associated with functional and radiological outcomes. Results: The mean age and standard deviation were 41+/-12 years. More females, n=49(52.7%), had ankle fractures. The commonest fracture type was Weber B n=65(69.9%)> Weber C n=16(17.2%)> Weber A n=12 (12.9%). The mean AOFAS-AH score was 82.9 ± 14.9. Patients had an AOFAS-AH score categorized as good (54.8%)>excellent (23.7%)>fair (16.1%)>poor (5.4%). Eight-six (92.5%) had radiological union, and 7(7.5%) had nonunion. 97.7% of the patients had a normal MCS, 80.2% normal TFO, 50% normal TCA, and 13.9% normal TFCS. Significant predictors of poor functional outcomes included being HIV-positive, Weber B fractures, and Weber C fractures. Predictors of good functional outcomes included receiving physiotherapy and initiating weight bearing at 4-6 weeks or after 6 weeks Conclusion: The majority of participants achieved radiological union, though all united fractures resulted in malunion. Worse functional outcomes were associated with Weber classifications B and C, as well as HIV-positive status.
Kidney disease affects 850 million people worldwide, with Sub-Saharan Africa bearing a significant burden. People living with HIV (PWH) are at increased risk due to nephrotoxicity of antiretroviral therapy (ART), in part due to widespread use of tenofovir disoproxil fumarate. In response, Uganda recommends routine kidney disease screening by doing a serum creatinine test at ART initiation. However, the extent of adherence to these guidelines remains poorly understood. We extracted clinical data for adults initiating ART between 2017 and 2024 at three large-volume HIV clinics in Uganda. To determine if kidney disease screening rates had increased appropriately over time, we divided the observation period into three eras as per national guidelines: (1) Test and Treat (2017-2019), that recommended screening only PWH and diabetes or hypertension; (2) DTG rollout/COVID-19 (2020-2022); and (3) creatinine-for-all (2023-2024), recommending screening everyone initiating ART. Logistic regression models were fit to identify correlates of renal screening. Of the 17,485 participants, only 22.4% (3,909/17,485) were screened for kidney disease at ART initiation. Screening was more common at the urban site (54.2%) compared to rural sites (10.0%). At rural sites, screening declined over time and individuals were 83% less likely to be screened in the creatinine-for-all era compared to the baseline era (aOR 0.17, 95% CI: 0.13-0.22) while it increased at urban site (aOR 9.27, 95% CI: 7.37-11.66). Male sex (aOR 1.37, 95% CI: 1.20-1.57), older age (≥45 years), hypertension, and non-TDF-based ART regimens were associated with higher screening odds at rural sites. Diabetes, opportunistic infections, and TDF use were not significantly associated with screening likelihood at any site. Kidney disease screening for PWH at ART initiation remains poor in Uganda, even when using a single creatinine test, particularly in rural clinics, highlighting critical challenges in translating national guidelines into practice. Future research should focus on understanding multilevel barriers to screening and evaluating strategies to improve guideline uptake.
