# Long-term follow-up of insomnia patients: symptom trajectories and predictors of treatment outcomes in a mobile-based CBT-I program

## TL;DR

This study tracks insomnia patients over four years using a mobile CBT-I program, finding that depression is a key predictor of long-term treatment outcomes.

## Contribution

The study identifies baseline depression as the strongest predictor of long-term treatment outcomes in insomnia patients using a mobile-based CBT-I program.

## Key findings

- Depressive symptoms at baseline strongly predicted poorer outcomes across sleep, mood, and somatic domains.
- Anxiety showed limited independent prognostic value after adjusting for depression.
- Older age was associated with better treatment response in key symptom domains.

## Abstract

To investigate the longitudinal trajectories and prognostic factors of insomnia-related symptoms over a four-year follow-up period, using pharmacotherapy combined with a mobile-based cognitive behavioral therapy (CBT-I) for insomnia intervention.

The clinical data including 1,022 patients of insomnia within January 2017 to January 2024 were obtained from the Sleep Center of the First People’s Hospital of Yunnan Province. Participants completed standardized assessments via the “Good Sleep 365” mobile app, including Pittsburgh Sleep Quality Index (PSQI), Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7), PHQ-15, Epworth Sleepiness Scale (ESS), and Insomnia Severity Index (ISI) at baseline and multiple follow-up points. Symptom trends were analyzed descriptively, and linear mixed-effects models were used to examine trajectories and identify baseline predictors of treatment outcomes.

PSQI, GAD-7, PHQ-9, and PHQ-15 scores improved markedly during the first 12 months after treatment, but showed mild relapse thereafter, whereas ESS scores remained largely stable. Higher baseline PHQ-9 scores predicted poorer outcomes across all symptom domains, while baseline GAD-7 showed limited independent prognostic value after adjusting for depression. Older age was associated with better treatment response (β ≈ −0.05 for GAD7 and PHQ9).

Baseline depressive symptoms were the strongest predictor of long-term outcomes across sleep, mood, and somatic domains, whereas anxiety added only modest prognostic value. These findings highlight the importance of routine depression screening to guide risk stratification and follow-up in chronic insomnia.