This was a prospective clinical pilot study conducted as a singlearm trial. Participants were recruited from the psychiatric day care hospital and outpatient clinic of Seoul Metropolitan Eunpyeong Hospital. Patients with schizophrenia who had difficulty in sleeping and wanted treatment options other than drugs were recruited. This study was approved by the Institutional Review Board of Seoul Metropolitan Eunpyeong Hospital, Korea (IRB No. Eunpyeong 2022-06). The researchers explained the aims and methods of the study to patients, and the patients who voluntarily consented to participate were assessed for eligibility.

The inclusion criteria were as follows: 1) diagnosis of schizophrenia according to the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5); 2) Pittsburgh Sleep Quality Index (PSQI-K) of ≥5, which is indicative of clinically significant sleep problems; and 3) age 18–65 years. The exclusion criteria were as follows: 1) Psychotic Symptom Rating Scale (PSYRATS) score >2 for either delusions or hallucinations; 2) hospitalization in closed wards for at least 3 months before study enrollment; 3) changes in psychiatric medication for at least the past month; and 4) diagnosis of sleep apnea, alcohol or substance use disorders, organic brain damage, or intellectual disability.

All patients included in the study completed the baseline (before sessions) and follow-up assessments (after sessions 2, 3, and 4) and were analyzed. Psychiatric evaluations were conducted by two psychiatrists who were also evaluators of the scales used in this study.

All participants received four sessions of individual or group CBT-I (Table 1). Each session was completed in groups; missed sessions were supplemented with individual sessions. Our CBT-I program was modified for patients with schizophrenia with relatively stable psychotic symptoms based on the clinical guidelines for insomnia in psychosis [21]. Each session lasted approximately 60 min, with some flexibility in the session duration. The patients were provided with a resource pack at each session that outlined the skills addressed in that session and a sleep diary.

The PSQI-K [22] and Insomnia Severity Index (ISI) [23] were used as the main outcomes to determine the effectiveness of CBT-I. The PSQI-K components include subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbance, use of sleep medications, and daytime dysfunction. These scores for these seven components were summed to yield the total PSQI-K. The ISI is a self-report questionnaire that measures the patients’ perceptions of insomnia; the index is scored as follows: 0 to 7 indicating “no clinically significant insomnia”; 8–14 indicating “subthreshold insomnia;” 15–21 indicating “moderate clinical insomnia;” and 22–28 indicating “severe clinical insomnia.” The Dysfunctional Beliefs and Attitudes about Sleep Scale-16 (DBAS-16) [24] was used to measure maladaptive cognition associated with sleep. This scale comprises the following four subscales: “consequences,” “worry/helplessness,” “expectations,” and “medication.” The Korean version of the Epworth Sleepiness Scale (K-ESS) [25] was used to evaluate the average sleep propensity in daily life. Total sleep time [26] and sleep onset latency [27] were also assessed from each sleep diary.

All statistical analyses were performed using the IBM SPSS 27 software (IBM Corp., Armonk, NY, USA). Continuous variables are reported as mean and standard deviation (SD) and categorical variables as frequency and percentage. As the primary results of this pilot study, the mean and standard deviation of all outcomes were calculated at the following timepoint: baseline before session 1 and after sessions 2, 3, and 4. The primary significance values for each outcome variable were calculated using one-way repeated measures analysis of variance (ANOVA), with the timepoint as a within-subjects factor. When ANOVA findings were significant, post-hoc comparison tests were performed to compare the data at baseline before session 1 and after sessions 2, 3, and 4. A p-value of <0.013, using the Bonferroni correction for multiple comparisons, was considered statistically significant. Partial eta-squared (η2) values were used when indicated to reflect the effect sizes; partial eta-squared represented a small effect size, while 0.06 denoted a medium effect size, and 0.14 indicated a large effect size, as established by previous research [28].

As shown in the study flowchart (Figure 1), 38 patients with schizophrenia who had difficulty in sleeping participated and were assessed for eligibility. Five patients were excluded for meeting the exclusion criteria and two were unable to participate in the CBT-I schedule. Finally, 31 patients received four weekly sessions of CBT-I; there were no dropouts during the treatment period. All 31 patients completed the baseline (before sessions) and follow-up assessments (after sessions 2, 3, and 4) and were analyzed.

The demographic and baseline clinical characteristics of the participants are presented in Table 2. The mean age of the participants (n=31; female: 54.8% [n=17]) was 46.97 years (SD=11.46). The mean body mass index was 24.55 kg/m2. The PSYRATS score was used to exclude insomnia symptoms caused by worsening psychotic symptoms, and the baseline total PSYRATS score was 3.57 (SD=3.84). The mean score on the auditory hallucination subscale was 3.32 (SD=3.69), and the mean score on the delusion subscale was 0.26 (SD=1.00). The mean baseline Patient Health Questionnaire-9 (PHQ-9) score was 4.06 (SD=3.79).

The complete CBT-I program was completed by 31 (93.9%) patients who signed the consent form (n=33); the complete program included four treatment sessions and both pre- and posttreatment assessments.

Compared with the baseline data, the mean total DBAS-16 score, ESS score, and ISI significantly decreased after sessions 2, 3, and 4. Table 3 presents the changes in the clinical outcomes of CBT-I over the study period. Repeated-measures ANOVA revealed a significant effect of time on DBAS-16 score [F(50.39); p<0.001], ESS score [F(33.85); p<0.001], and ISI [F(29.23); p<0.001]. The sleep onset latency from the sleep diary at each study timepoint showed a significant improvement over the period [F(29.70); p< 0.001]. Similarly, the mean total PSQI-K decreased significantly after sessions 2, 3, and 4. However, some scores of the PSQI-K subscales (sleep duration, use of sleep medication, and daytime dysfunction) showed no reliable improvement over this period. The total sleep time from the sleep diary also showed no significant improvement over the study period.