Could the telescoping effect and comorbidity with substance use disorders account for sex differences in the processes and therapeutic outcomes? A latent-class moderation analysis.

Dacosta-Sánchez, Daniel; Mancheño-Velasco, Cinta; Narváez-Camargo, Marta; Lozano, Óscar M · Drug Alcohol Depend · 2026

retrospective_cohort · Level III

Where this comes from

Abstract

Sex differences in substance use disorder (SUD) treatment outcomes remain unclear; protocols focused on male prevalence may overlook female-specific needs. This study tested interactions among sex, the telescoping effect, and comorbid profiles in outcomes. Retrospective multisite electronic health record study (N = 4818) of patients with co-occurring disorders. Interaction regressions and latent class analysis tested sex moderation of links between years of substance use (telescoping) and outcomes within SUD and co-occurring diagnostic groups. Slower progression from substance use onset to diagnosis was associated with lower readmission likelihood at 3 years in males than females; similar patterns appeared at 18 and 24 months for alcohol dependence and at 18 months for co-occurring mood disorders. Six comorbid classes were identified (feature-based labels): behavioral dysregulation (Class 1), anxious distress (Class 2), adolescence-cannabis (Class 3), emotional-interpersonal dysregulation (Class 4), affective-alcohol (Class 5), and reality distortion (Class 6). Only Classes 2 and 4 showed the telescoping effect. Latent-class moderation revealed males in Classes 1 and 3 had higher readmission over 3 years, whereas females in Class 4 had lower adherence to scheduled appointments and higher predicted readmissions across all follow-ups. Females in Class 6 showed better appointment adherence than males. Telescoping and sex differences may contribute to higher readmission in females, particularly with alcohol dependence and mood or personality disorders. Males with impulse-control or childhood disorders may require more post-treatment care. Sex-specific interventions and moderation/latent class approaches may improve targeting.