Journal of Acquired Immune Deficiency Syndromes · Nov 18, 2026

Development and Internal Validation of a Prediction Model for Reversible Cognitive Frailty in People Living with HIV

By Michael HU, Brouillette MJ, Tamblyn R et al.

Background: Reversible cognitive frailty may be an important target for early identification among people with HIV (PWH). We developed a prediction model for reversible cognitive frailty and assessed its internal validity using clinical, laboratory, and medication-related data. Setting: Positive Brain Health Now cohort, including adults recruited from five Canadian HIV clinics between 2013 and 2018. Methods: Data from 824 adults with HIV aged 35 years or older were analysed. Reversible cognitive frailty was defined as co-occurring physical frailty and self-reported cognitive difficulty without dementia. Twenty candidate predictors were entered into least absolute shrinkage and selection operator logistic regression (LASSO) with 10-fold cross-validation to select the final predictor set. Internal validity was assessed through discrimination (apparent and optimism-corrected C-statistic via bootstrap resampling), calibration, and clinical utility (decision curve analysis). Results: Reversible cognitive frailty was present in 7.5% of participants. The final model retained 10 predictors; the strongest were female sex, greater comorbidity burden, higher anxiety and depression scores, and higher anticholinergic medication burden, with additional contributions from hepatic, nutritional, HIV-related, and medication variables. On internal assessment, the model showed strong discrimination (optimism-corrected C-statistic 0.882), acceptable calibration, and net benefit over assess-all and assess-none strategies. At the optimal threshold, sensitivity was 91.9%, specificity 76.1%, and negative predictive value 99.1%. Conclusion: This is the first prediction model developed for reversible cognitive frailty in PWH, showing strong discrimination, acceptable calibration, and clinical utility within the development cohort. External validation in independent samples is needed before clinical use.

doi.org/10.1097/qai.0000000000004005PubMed 43050450Biomarkers

AgingResearch.News