Real World Effectiveness of Digital Cognitive Behavioral Therapy for Insomnia: A Retrospective Study in Chinese Population
Menglin Lu,Yaoyun Zhang,Junhang Zhang,Songfang Huang,Fei Huang,Luo Si,Tingna Wang,Fei Wu,Hongjing Mao,Zhengxing Huang
DOI: https://doi.org/10.2139/ssrn.4172078
2022-01-01
Abstract:Background: Although digital cognitive behavioral therapy for insomnia (dCBT-I) has been investigated in many random clinical trial (RCT) studies and recommended as a first-line treatment option, few studies have systematically examined its efficacy, engagement, durability, and adaptability in the real-world.Methods: We conducted a real-world study using longitudinal data collected by a mobile App named Good Sleep 365 from November 14, 2018 to February 28, 2022. Three therapeutic modes, namely dCBT-I, medication, and their combination, were measured and compared for treatment efficacy. Strict cohort eligibility criteria were established and the effect of confounders was eliminated using propensity score-based inverse probability of treatment weighted (IPTW). The Pittsburgh Sleep Quality Index (PSQI) score and its essential subitems were used as the primary outcomes. Effects on comorbid somnolence, anxiety, depression and somatic symptoms were used as the secondary outcomes. In addition, the efficacy of dCBT-I in different subgroups, and the effect of its therapeutic sessions were examined.Findings: 1,418 patients (mean [standard deviation] age, 46·22 [11·68] years; 1,102 [77·7%] females) were selected for dCBT-I (n=62), medication (n=279), and their combination (n=1,077). In comparison with medication, both dCBT-I (Cohen d, -0·7; 95% CI, -0·98 to -0·41; p=0·013; SMD=0·534) and combination therapy (Cohen d, 0·51; 95% CI, 0·38 to 0·64; p<0·001, SMD=0·54) achieved remarkable reductions of PSQI score at 6-month follow-up, and dCBT-I monotherapy had a comparable promoting effect (p=0·544; SMD=0·101) with combination therapy. The response rate (≥50% reduction in baseline score) and remission rate (PSQI score < 8) of combination therapy were higher than those of dCBT-I and medication (52·20% vs 37·84% and 32·65%, 53·21% vs 40·54% and 34·69%). Compared to medication therapy, changes in all secondary outcomes indicated statistically significant benefits from dCBT-I (GAD-7, Cohen d, 0·78, 95% CI, 0·49 to 1·06, p=0·006; PHQ-9, Cohen d, 0·87 , 95% CI, 0·58 to 1·15, p=0·002; PHQ-15, Cohen d, 0·79, 95% CI, 0·5 to 1·07, p=0·007) and combination therapy (0·58, 0·45 to 0·71, p<0·001; 0·56, 0·43 to 0·69, p<0·001; 0·45, 0·32 to 0·58, p=0·001). dCBT-I conferred benefits in young patients (OR, 1·53; 95% CI, 0·37 to 6·31) or in those with a medication history (1·54; 0·48 to 5·01) or those taking hypnotics (2·06; 0·63 to 6·79), whereas dCBT-I appeared to be unreliable for patients with low education (0·34; 0·08 to 1·36) or a family history (0·54; 0·15 to 1·94).Interpretation: Real-world evidence suggests that combination therapy is optimal, and dCBT-I is more effective than medication therapy, with long-term benefits for insomnia. However, the therapeutic effect of dCBT-I varies by sub-population, requiring future studies to analyze its real-world efficacy and reliability on distinct sub-populations to enable “targeted therapy”.Funding: This work was partially supported by the National Nature Science Foundation of China under Grant No. 61672450. Thanks to Wenyue Ma, Yunlong Gao and others in Hangzhou slan-health Co., Ltd for their support in data access.Declaration of Interest: None to declare. Ethical Approval: The use of clinical information, and data collection protocols were approved by the ethical committee of College of Biomedical Engineering and Instrument Science, Zhejiang University (number: Zheda Shengyi 2022-3).