Sedative co-medication patterns across frailty states in people with HIV: a network-based study.

Abstract

Introduction: People with HIV are living longer, but frailty may increase vulnerability to adverse drug interactions, particularly with sedative medications.

Aim: To describe sedative co-medication patterns across frailty states using network-based analysis, and to identify key medications driving interaction complexity.

Method: This cross-sectional study analyzed 321 participants using sedatives from the Positive Brain Health Now Cohort (mean age: 53 years), categorized as robust (30.2%), prefrail (47.0%), or frail (22.7%). Sedative use was classified using the Sedative Load Model, and frailty was assessed with a modified Fried Frailty Phenotype. Co-medication networks were constructed for robust, prefrail, and frail groups, with metrics such as Neighborhood Shift Scores (NESH) and ΔBetweenness used to evaluate network dynamics. Edge-level Observed-to-Expected ratios highlighted significant drug pairings and interaction risks.

Results: Frail individuals exhibited the most interconnected sedative network (graph density: 0.24; average degree: 7.31) followed by prefrail (graph density: 0.18; average degree: 6.67) and robust groups (graph density: 0.13; average degree: 4.61). Medications with high network influence included mirtazapine (robust-to-prefrail: NESH = 2.05; ΔBetweenness = 0.23), gabapentin (robust-to-frail: NESH = 2.46, ΔBetweenness = 0.10), and pregabalin (prefrail-to-frail: NESH = 2.55; ΔBetweenness = 0.71). High-risk sedative pairs in frail individuals included hydromorphone-clonazepam (Observed-to-Expected ratio: 3.07; interaction severity: D), amitriptyline-oxycodone (Observed-to-Expected ratio: 1.92; severity: D), and quetiapine-citalopram (Observed-to-Expected ratio: 2.12; severity: D). Prefrail individuals had intermediate-risk combinations, including clobazam-levetiracetam (Observed-to-Expected ratio: 81.5; severity: C), and gabapentin-hydroxyzine (Observed-to-Expected ratio: 3.58; severity: D). Robust individuals showed fewer and lower-risk patterns, such as amitriptyline-pregabalin (Observed-to-Expected ratio: 4.00; severity: C). Network differences reflect increasing polypharmacy and adverse interaction risks with advancing frailty.

Conclusion: Sedative co-medication patterns vary meaningfully across frailty states in people with HIV. Frailty is associated with more complex networks and a greater likelihood of high-risk drug interactions. Medications influencing these patterns can help identify opportunities for safer prescribing. The prefrail stage may offer a timely window for interventions aimed at minimizing polypharmacy and improving safety in middle-aged and older adults with HIV.