Exploring the associations between data-driven insomnia disorder combined with mild anxiety or/and depressive symptoms and the efficacy of Cognitive-Behavioral Therapy for insomnia

Objectives

The endophenotype of insomnia disorder is complex and the treatment is not targeted. The data-driven typing method might provide some bases for precise treatment. The present study was based on a post hoc analysis, aiming to explore the association between subtypes of insomnia disorder and the efficacy of cognitive-behavioral therapy for insomnia (CBT-I).

Methods

The present study was conducted on data of 118 patients with chronic insomnia disorder combined mild anxiety or/and depressive symptoms, who had completed an 8-week randomized controlled trial of CBT-I vs CBT-I plus (CBT-I combined with modules targeting anxiety and depressive symptoms). The silhouette coefficient determined the optimal number of clusters, and a K-means clustering analysis was performed. T-tests were conducted to assess baseline differences at eight weeks, and the changes in self-reported total sleep time (sTST), Pittsburgh Sleep Quality Index (PSQI), Hamilton Depression Rating Scale (HRSD-17), and Hamilton Anxiety Rating Scale (HAMA) scores in order to explore the impact of subtypes and treatment approaches (CBT-I and CBT-I plus) on insomnia and emotional symptoms.

Results

The analysis revealed no significant demographic differences between the two clusters. Subtype 2 was characterized by significantly poorer baseline sleep quality (PSQI: 16.59 vs 12.74, t = -9.90, p < 0.01), higher depressive (HRSD: 18.47 vs 13.21, t = -8.37, p < 0.01), and anxiety levels (HAMA: 17.47 vs 13.46, t = -6.23, p < 0.01), and shorter sTST (4.67 vs 6.09 h, t = 8.31, p < 0.01) compared to Subtype 1. Post-treatment analyses showed significant improvements in both subtypes, with Subtype 2 experiencing a larger increase in sleep duration (csTST: 0.58 vs 1.77 h, t = -7.18, p < 0.01) and more pronounced improvements in sleep quality (cPSQI: 6.92 vs 8.88, t = -3.57, p < 0.001), depression (cHRSD: 8.07 vs 10.59, t = -2.71, p = 0.008), and anxiety (cHAMA: 9.28 vs 11.22, t = -2.56, p = 0.012). Despite these improvements, Subtype 1 maintained significantly better outcomes in sleep quality (PSQI: 5.81 vs 7.71, p < 0.01), depression (HRSD: 5.14 vs 7.89, p < 0.01), and anxiety (HAMA: 4.18 vs 6.25, p < 0.01) at 8 weeks. No significant differences in baseline characteristics were found between treatment groups within subtypes, indicating homogeneity. Within Cluster 1, CBT-I plus was more effective in reducing depressive symptoms (cHRSD: t = -2.48, p = 0.016), whereas CBT-I was superior in enhancing sTST in Cluster 2 (t = 2.01, p = 0.049), with no significant differences in other measures.

Conclusions

The study underscores the heterogeneity within ID subtypes and the differential response of sleep quality and depressive symptoms to CBT-I and CBT-I plus, highlighting the importance of personalized treatment strategies based on insomnia subtypes.