Statistical analysis All analyses were conducted using SAS 9.2 (SAS, Inc., Cary, NC). Because this is a pilot trial, no a priori power calculations were conducted prior to initiating enrollment; sample sizes were selected based on a sufficient number to estimate the treatment effect size. The primary and secondary analyses were conducted using multilevel random effects

models. For the primary outcome, ISI score was modeled specifying random intercepts for participants (i.e., accounting for variance in the initial levels of insomnia across participants at baseline) with group (HUC vs. UC) and time (baseline vs. post-treatment) as fixed effects. Inhibitors,research,lifescience,medical The group × time interaction was interpreted as the differential change of the HUC group compared with the UC group. Secondary outcomes were similarly modeled, with the follow-up Inhibitors,research,lifescience,medical period added to examine the duration of change. To estimate the size of effect, Cohen’s d was calculated for all outcome measures to index the size of the group differences in terms of within-group standard deviations (e.g., 1.2 standard deviation difference between the groups). Although arbitrary ranges, standard deviation differences ≤0.2 are often

Inhibitors,research,lifescience,medical considered “small”, d = 0.5 are considered “medium,” and d > 0.80 are “large.” Descriptive statistics are presented as means (SD) or frequency counts (%) as appropriate. All point estimates of differential change are presented Inhibitors,research,lifescience,medical with 95% confidence intervals. Where appropriate, all hypothesis testing is two-tailed with P < 0.05 interpreted as statistical significance. Results Baseline data and subject flow A total of 28 subjects were

enrolled in the study at Wake Forest Baptist Health (Fig. 3). Recruitment took place from March 1, 2011, through May 1, 2011. Twenty participants Inhibitors,research,lifescience,medical were assigned to either the wait-list UC or HUC group. Demographics and baseline characteristics (Table 1) were not statistically different between the two selleck chemical groups. There were slightly more comorbidities noted in the HUC group (Table 2). Antidepressants were used by three subjects in the HUC group, and one in the UC group. All patients continued their usual care throughout the course of the study; HIRREM was added to usual care during the primary intervention epoch. All subjects completed the primary intervention period, and primary not data collection visits. All 10 participants in the HUC group received HIRREM (mean of 10.3 sessions) and nine of 10 UC subjects subsequently received HIRREM after crossover. One in the UC group had a job change and the schedule prevented further participation. One subject from each group receiving HIRREM was not available for the late telephone follow-up. Figure 3 Subject recruitment and flow through the study. Table 1 Baseline demographics Table 2 Self-reported comorbidities Primary outcome Mean baseline ISI for each group was identical, at the enrollment visit (mean = 18.6, P = 1.0).