People living with HIV (PLHIV) have an increased risk of tuberculosis (TB) and severe COVID-19. TB and COVID-19 present with overlapping symptoms and co-infection can lead to poor outcomes. We assessed the frequency of SARS-CoV-2 positive serology and SARS-CoV-2 infection and the risk of mortality at 6 months in PLHIV with TB disease and SARS-CoV-2 infection. This multi-country, prospective, observational study, conducted between 7th September 2020 and 7th April 2022, included ambulatory adult PLHIV investigated for TB (with symptoms of TB or advanced HIV disease) in Kenya, Uganda, and South Africa. Testing included CD4 cell count, Xpert MTB/RIF Ultra assay (sputum), Determine TB LAM Ag assay (urine), chest X-ray, blood SARS-CoV-2 serology test and SARS-CoV-2 PCR (only if TB or COVID-19 symptoms). Individuals were followed for 6 months. Among 1254 participants, 1204 participants had SARS-CoV-2 serology (54% women, median CD4 344 cells/µL [IQR 132–673]), and 487 had SARS-CoV-2 PCR. SARS-CoV-2 serology positivity was 27.0% (325/1204), lower in PLHIV with CD4 counts <200 cells/µL (19.9%, 99/497) than in those with CD4 counts ≥200 cells/µL (31.6%, 222/703), p<0.001. SARS-CoV-2 PCR positivity was 8.6% (42/487) and 27.7% (135/487) had probable or confirmed SARS-CoV-2 infection. Among PLHIV with symptoms of TB or of COVID-19, 6.6% (32/487) had SARS-CoV-2 infection and TB disease. In multivariable analyses, the risk of death was higher in PLHIV with both SARS-CoV-2 infection and TB compared to those with only SARS-CoV-2 infection (adjusted hazard ratio [aHR] 8.90, 95%CI 1.47-53.96, p=0.017), with only TB (aHR 3.70, 95%CI 1.00-13.72, p=0.050) or with none of them (aHR 6.83, 95%CI 1.75-26.72, p=0.006). These findings support SARS-CoV-2 testing in PLHIV with symptoms of TB, and SARS-CoV-2 vaccination, especially for those with severe immunosuppression. PLHIV with COVID-19 and TB have an increased risk of mortality and would benefit from comprehensive management and close monitoring.
BACKGROUND
Viral load (VL) suppression is key to ending the global HIV epidemic, and predicting it is critical for healthcare providers and people living with HIV (PLHIV). Traditional research has focused on statistical analysis, but machine learning (ML) is gradually influencing HIV clinical care. While ML has been used in various settings, there’s a lack of research supporting antiretroviral therapy (ART) programs, especially in resource-limited settings like Guinea. This study aims to identify the most predictive variables of VL suppression and develop ML models for PLHIV in Conakry (Guinea).
METHODS
Anonymized data from HIV patients in eight Conakry health facilities were pre-processed, including variable recoding, record removal, missing value imputation, grouping small categories, creating dummy variables, and oversampling the smallest target class. Support vector machine (SVM), logistic regression (LR), naïve Bayes (NB), random forest (RF), and four stacked models were developed. Optimal parameters were determined through two cross-validation loops using a grid search approach. Sensitivity, specificity, predictive positive value (PPV), predictive negative value (PNV), F-score, and area under the curve (AUC) were computed on unseen data to assess model performance. RF was used to determine the most predictive variables.
RESULTS
RF (94% F-score, 82% AUC) and NB (89% F-score, 82% AUC) were the most optimal models to detect VL suppression and non-suppression when applied to unseen data. The optimal parameters for RF were 1,000 estimators and no maximum depth (Random state = 40), and it identified Regimen schedule_6-Month, Duration on ART (months), Last ART CD4, Regimen schedule_Regular, and Last Pre-ART CD4 as top predictors for VL suppression.
CONCLUSION
This study demonstrated the capability to predict VL suppression but has some limitations. The results are dependent on the quality of the data and are specific to the Guinea context and thus, there may be limitations with generalizability. Future studies may be to conduct a similar study in a different context and develop the most optimal model into an application that can be tested in a clinical context.
We investigated people living with HIV (PLWH)’s exposure to COVID-19 pandemic stressors and their association with distress, psychological growth, and substance use. PLWH in the ANRS CO3 AQUIVIH-NA cohort’s QuAliV study (Nouvelle Aquitaine, France) completed an adapted CAIR Lab Pandemic Impact Questionnaire (C-PIQ) and reported substance use between 9/2021 to 3/2022. We described cumulative stressor exposure (score 0-16) and explored variation by PLWH characteristics (demographic, HIV-related, risk factors, psychosocial). Associations with distress (score 0-23), psychological growth (score 0-20), and substance use were assessed using regression models. Participants reported exposure to a median of 2 (IQR: 1-4) stressors. Stressor exposure was higher in working-age (<60) and psychosocially vulnerable PLWH. Exposure to an additional stressor correlated with a 0.7-point increase in distress scores (95% C.I. 0.5-1.0, p<0.001), a 0.04-point increase (95% C.I. 0.01-0.07, p=0.002) in psychological growth scores in working-age PLWH. In older PLWH, additional stressor correlated with a 0.8-point (95% C.I. 0.4-1.2, p<0.001) increase in distress and a 0.1-point increase (95% C.I. 0.06-0.2, p=0.001) in growth scores. Each additional stressor was associated with 1.2 (95% C.I. 1.0-1.4, p=0.02) higher adjusted odds of cannabis use in working-age PLWH, and 1.2 (95% C.I. 1.0-1.4, p=0.004) higher adjusted odds of drug use. Exposure to stressors was linked to increased distress, cannabis and drug use but also growth. Providers should not only be aware of risk (of severe COVID-19) but also be mindful of the social and psychological challenges PLWH face as these may affect their retention in care, especially during challenging times.
INTRODUCTION
The retention in care of patients undergoing antiretroviral therapy (ART) is a cornerstone for preventing AIDS‐associated morbidity and mortality, as well as further transmission of HIV. Adherence to ART poses particular challenges in conflict‐affected settings like the Central African Republic (CAR). The study objective was to estimate the rate of lost‐to‐follow‐up (LTFU) and determine factors associated with LTFU among patients living with HIV under ART in CAR.
METHODS
A retrospective cohort analysis was conducted using data from patients being managed at 42 representative ART dispensing sites (i.e. management of ≥200 patients) in the seven health regions of CAR which started ART between January 2019 to September 2021 and followed up to December 2021. The outcome of LTFU was defined as a failure of a patient to attend a scheduled ART refill appointment for at least 90 days from the last appointment. Patients were censored at the first LTFU event.
RESULTS
A total of 6844 patients enrolled in ART care were included in the analysis, of whom 67.5% were females. The mean age (standard deviation) was 35.3 years (10.5). Forty‐two per cent (n = 2874/6844) had an LTFU event during the follow‐up period. However, 23.2% (n = 666/2874) returned to care after LTFU. Overall retention in antiretroviral care at 12 months was 64.2% (CI 63.0−65.5), which ranged from 76.1% in the capital to 48.2% in the inner country region. Risk factors related to LTFU were being male (adjusted hazard ratio [aHR] 1.33; CI 1.1−1.5), age < 25 (aHR 1.46; CI 1.1−1.9), living in regions outside the capital (aHR 1.83; CI 1.6−2.3) and undernutrition (aHR 1.13; CI 1.0−1.3).
CONCLUSIONS
Retention to care in CAR is suboptimal, especially in the inner country. Our results underline the difficulties involved in retaining patients in ART in complex settings, the interplay between poor retention, social unrest, stigma, food insecurity and HIV epidemic control, and the need for tailored programming and interventions like differentiated treatment strategies and complementary food provision.
BACKGROUND
Targeted preventive strategies in persons living with HIV (PLWH) require markers to predict visceral leishmaniasis (VL). We conducted a longitudinal study in a HIV-cohort in VL-endemic North-West Ethiopia to 1) describe the pattern of Leishmania markers preceding VL; 2) identify Leishmania markers predictive of VL; 3) develop a clinical management algorithm according to predicted VL risk levels.
METHODS
The PreLeisH study followed 490 adult PLWH free of VL at enrolment for up to two years (2017-2021). Blood RT-PCR targeting Leishmania kDNA, Leishmania serology and Leishmania urine antigen test (KAtex) were performed every 3-6 months. We calculated the sensitivity/specificity of the Leishmania markers for predicting VL and developed an algorithm for distinct clinical management strategies, with VL risk categories defined based on VL history, CD4 count and Leishmania markers (rK39 RDT & RT-PCR).
FINDINGS
At enrolment, 485 (99%) study participants were on antiretroviral treatment; 360/490 (73.5%) were male; the median baseline CD4 count was 392 (IQR 259-586) cells/μL; 135 (27.5%) had previous VL. Incident VL was diagnosed in 34 (6.9%), with 32 (94%) displaying positive Leishmania markers before VL. In those without VL history, baseline rK39 RDT had 60% sensitivity and 84% specificity to predict VL; in patients with previous VL, RT-PCR had 71% sensitivity and 95% specificity. The algorithm defined 442 (92.3%) individuals at low VL risk (routine follow-up), 31 (6.5%) as moderate risk (secondary prophylaxis) and six (1.2%) as high risk (early treatment).
INTERPRETATION
Leishmania infection markers can predict VL risk in PLWH. Interventional studies targeting those at high risk are needed.
FUNDING
The PreLeisH study was supported by grants from the Department of Economy, Science and Innovation of the Flemish Government, Belgium (757013) and the Directorate-General for Development Cooperation and Humanitarian Aid (DGD), Belgium (BE-BCE_KBO-0410057701-prg2022-5-ET).