Smartphone-Based Intervention for Sleep Disturbance in Individuals Recovering From Alcohol Use Disorder

Overview

In order to assess the efficacy of the rhythm stabilization component of Health Rhythms' product in addressing sleep problems, investigators will recruit 128 outpatients over the age of 18, who are in early recovery (at least 2 weeks but no more than 6 months) from Alcohol Use Disorder (AUDIT-C score \>4) and are experiencing sleep problems (ISI score \>/= 10). 64 of these individuals will be randomly allocated to receive the full HealthRhythms package (sensing, trending, and stabilization) on their smartphones, while 64 will be asked to consent to sensing only. The duration of the study will be 14 weeks. The primary outcome measures of the RCT will be the ISI (Insomnia Severity Index). The investigators hypothesize that those receiving the full package will demonstrate lower levels of sleep disturbance.

We will conduct a 14-week randomized controlled trial (RCT) of CueToSleep (behavioral monitoring via smartphone sensors and phone-based Cognitive Behavior Therapy for Insomnia) versus Measure (smartphone-based monitoring only). The sample will be 128 individuals with alcohol use disorder who have been abstinent for at least two weeks and are experiencing sleep difficulties. Using a block randomization scheme, these individuals will be randomly allocated on a 1:1 basis to the experimental condition (CueToSleep) or to the control condition (Measure). The components of the CueToSleep intervention include: 1) education about sleep and its regulation and sleep hygiene; 2) alcohol withdrawal-specific psychoeducation 3) personalized and context-specific suggestions for behavior change to address sleep problems using data obtained from the monitoring portion of CueToSleep; 4) feedback to the user on his/her progress; 5) feedback from the user about the usefulness of the suggestions; 6) adjustment of recommendations based on progress, adherence and feedback from the user. The primary outcome is change in sleep difficulties as measured by the Insomnia Severity Index (ISI). The study is powered to detect at least medium Cohen's d effect size for this primary outcome. The secondary outcome of interest is proportion relapsing to heavy drinking (≥4 drinks in a day for women; ≥5 drinks in a day for men). Number of drinks consumed on the previous day will be queried daily via smartphone and relapse to heavy drinking will be determined from these participant self-reports. For our primary outcome, we will fit a 2-level linear mixed effects model with a random intercept and slope to test for group (experimental vs. control) differences in the slope of ISI scores over 14 weeks. This model will regress repeatedly measured ISI score on week, group, the week by group interaction, and covariates (age, sex, living status, employment status). Secondarily, to allow further flexibility in the trajectory of ISI scores over the course of the study, we will fit a mixed effects model with categorical time, and specific contrasts to test changes from baseline to 14 weeks. For our secondary outcome, we will fit a cox proportional hazards model to regress time to relapse on group and covariates (age, sex, living status, employment status). This model accounts for participants who are censored due to early dropout from the study. Secondarily, we will use simple non-parametric statistics to compare the proportion relapsing to heavy drinking in each group. All analyses will use an intent-to-treat (ITT) strategy where all participants are included regardless of adherence to the randomized condition.