# Network analysis of comorbid mental health disorders among people with HIV in the All of Us Research Program

## TL;DR

This study explores how mental health disorders cluster among people with HIV, identifying key conditions that may be central to broader comorbidity patterns.

## Contribution

The paper introduces a network analysis approach to uncover comorbidity patterns of mental health disorders in people with HIV.

## Key findings

- Two main clusters of mental health disorders were identified: affective/anxiety and substance-related conditions.
- Recurrent depression and substance use disorders were found to be central in the comorbidity network.
- Sensitivity analyses confirmed the stability of the network structure and centrality rankings.

## Abstract

People living with HIV experience high rates of mental health disorders, but the comorbidity patterns of these conditions remain poorly understood. Identifying how disorders cluster and which diagnoses are most central may guide more effective screening and integrated treatment strategies.

This cross-sectional study used electronic health records and survey data from the National Institutes of Health All of Us Research Program (Release 8; May 2018–October 2023). Of 6,664 participants with confirmed HIV, 5,868 had electronic health record data; 3,078 with at least two encounters bearing mental health disorder diagnoses were included in the analysis. Diagnoses were identified using ICD-10 codes. Comorbidity networks were estimated with a psychometric network analysis based on conditional log-odds associations. Network centrality was evaluated using strength and expected influence measures, and clusters were identified with the Louvain algorithm. Sensitivity analyses included relaxing the diagnostic threshold to one recorded encounter and assessing robustness through nonparametric bootstrapping.

Participants were predominantly aged 50 years or older (69.2%), Black or African American (47.7%), non-Hispanic (77.5%), and male (64.4%); 91.2% had health insurance. The most common disorders were tobacco-related conditions (49%), major depressive disorder single episode (45%), other anxiety disorders (45%), recurrent depression (28%), adjustment and stressor-related disorders (25%), and substance-related conditions including alcohol (21%), cocaine (20%), and other psychoactive substances (19%). Two main clusters were identified: affective and anxiety-related disorders, and substance use and psychotic disorders. The most central diagnoses in the network were multiple psychoactive substance use, cannabis-related disorders, personality disorders, cocaine-related disorders, and recurrent depression. Sensitivity analyses supported the stability of the network structure and centrality rankings.

Mental health disorders among people with HIV organize into two primary clusters centered on affective/anxiety and substance-related conditions. A small subset of highly central disorders, particularly recurrent depression and substance use conditions, appear to drive broader comorbidity patterns. Interventions targeting these central conditions may offer substantial benefit by disrupting interconnected clusters of psychiatric morbidity in this medically and socially vulnerable population.