The study used a pre-post testing design to assess differences in loneliness and other mental health outcomes in participants before and after interventions. Participants were assigned to the cognitive behavioral therapy (CBT) group or the social prescribing (SP) group. The University of Washington Institutional Review Board approved the study (STUDY00018180).

Participants were between ages 18-25, spoke English, met the cutoff score for loneliness (scored ≥ 6 on the UCLA 3-item Loneliness scale) (26), had at least one primary care visit between July 2022 - October 2023 in a health system in Washington state, and were willing to participate in virtual group-based interventions. The health system is an academically affiliated urban health system with 16 primary care practices. Individuals who were receiving CBT treatments at the time of recruitment were excluded. As this was a pilot feasibility trial, the study was not powered to detect significant differences between two conditions.

We used electronic health records (EHR) data to identify emerging adult patients in our primary care health system. We contacted the patients via text messaging and invited them to complete a REDCap survey to assess their eligibility. The survey included the UCLA-3 item loneliness scale (27) and participants’ demographics (i.e., age, gender, sex at birth, race and ethnicity, sexual orientation, relationship status, and living arrangements). Patients received $15 incentives for survey completion. We texted those who were eligible a link for electronic consent to join the study. Patients who provided consent and completed the baseline assessment were then assigned to either CBT or SP interventions using a sequential assignment method, where patients were assigned to groups based on the availability of open slots in each group following a predetermined sequence, to ensure they were balanced and even groups. Two CBT groups and two SP groups took place between fall 2023 – winter 2024.

Participants completed baseline and post-intervention surveys on REDCap. The baseline surveys were administered prior to assignment. The post-intervention surveys were administered within two weeks following the completion of the intervention. Participants received $25 for completing the baseline survey and $50 for completing the post-intervention survey. Both baseline and post-intervention surveys included the UCLA Loneliness-20, PHQ-9, and GAD-7.

The primary outcome was loneliness measured by the 20-item UCLA loneliness scale (version 3) (32). Using a 4-point rating scale (1= never; 4 = always), participants were asked to answer 20 questions on “how often they feel” about a positive or negative description of social interactions and perceptions. Scores range from 20-80, with higher scores indicating greater degrees of loneliness: 20-34 (low), 35-49 (moderate), 50-64 (moderately high), 65-80 (high) (33).

Secondary outcomes were depression and anxiety. Depressive symptoms were measured using the Patient Health Questionnaire-9 (PHQ-9) (34). Using a 4-point rating scale (0= not at all; 3 = nearly every day), participants were asked to measure symptoms of major depressive disorder over a 2-week period. Ratings are summed up and scores range are: 0-4 (minimal), 5-9 (mild), 10-14 (moderate), 15-19 (moderately severe), and 20-27 (severe). Anxiety symptoms were measured using the Generalized Anxiety Disorder 7-item Questionnaire (GAD-7) (35). Participants reported the presence and severity of anxiety symptoms over a 2-week period on a 4-point scale (0= not at all; 3 = nearly every day). Ratings were summed up and scores range are: 0-4 (minimal), 5-9 (mild), 10-14 (moderate), and 15-21 (severe).

To assess treatment adherence and participant engagement, both CBT and SP group facilitators completed a REDCap questionnaire for each participant after each session that included three questions: i) Did the participant attend the full treatment session? [Yes/no], ii) Overall, how much effort did the participant exert during group activities (e.g., in-group exercise, sharing, etc.)? and iii) How much was the participant engaged in the discussion and review of take-home practice? Group facilitators provided their subjective ratings on the latter two questions using a 5-point scale (1= none; 5= very often/very high effort). Ratings were summed up across all sessions for each participant and the total score ranged from 5 – 25, the higher the score, the greater the perceived level of participant’s engagement.

All participants were offered post-intervention interviews and 10 agreed to participate. These participants had varying levels of participation in the interventions. The individual interviews were conducted by our research staff using a semi-structured interview guide that asks about the experience of loneliness, barriers and facilitators to study activities, recommended adaptations for future intervention design, and behavioral changes after study participation. Each interview lasted about 30–45 minutes and participants received $50 after completing the interviews. The interviews were audio recorded, de-identified, and transcribed.

Our final sample (N = 15) included participants who completed baseline survey, postintervention survey, and attended at least one session of the CBT or SP intervention. Descriptive statistics were used to analyze and aggregate data on participant demographics and their treatment adherence. Paired T-tests were conducted separately for CBT and SP groups to examine whether there were significant pre- and post-intervention differences in outcomes. Group comparisons (CBT vs. SP) were not conducted as it was beyond the scope of this pilot study.

For qualitative interview data, we used Dedoose-9.0.46 (36) for data management and content analysis (37). The analysists (AJ and KC) first familiarized themselves with the data by reading each of the transcripts. A codebook was developed based on the semi-structured interview guide and the transcripts. Initially, two transcripts were coded by AJ and KC to verify accurate application of the codes. Then, the remaining transcripts were coded independently, and the team met regularly to discuss any inconsistencies within the code application. Once consensus was reached, emerging themes supported with quotes were used to summarize the findings.

Figure 1 depicts the flow diagram of participants through the study. Of 2,199 patients identified from the EHR, 243 (11%) of them completed the screening survey to assess eligibility. Of those who completed the screening, 112 (46%) were found eligible and expressed interest in participating in the study. The majority were excluded due to not meeting the cutoff score on UCLA-3 loneliness scale. Twenty-six (23%) eligible patients consented to enroll in the study; 20 of them completed the baseline assessment and were assigned to CBT (n = 10) and SP (n = 10) groups.

Overall attrition rate was at 25% (n = 5); four participants who were assigned did not start their assigned interventions (3 CBT, 1 SP) and one CBT participant did not complete the post-intervention survey. No significant demographic differences were found between those who dropped out vs. those who received the interventions. The five participants were excluded from the final analyses.

Table 2 summarized the demographic characteristics of total participants that were included in the final pre-post analysis (N = 15). The average age was 22 (SD = 2.4). Most were female (87%) and identified as women (73%). Participants were diverse in their race, ethnicities, and sexual orientations (see Table 2 ).

Of the 15 participants, 14 (93%) completed at least three out of five sessions in either intervention. Six (40%) participants completed all five treatment sessions: three each from the CBT and SP groups. The group facilitator-rated levels of participant effort and engagement were high in both groups. CBT participants were rated with mean scores of 19.7 and 19.3 (out of 25) for engaging in group activities and in take-home practice, while the SP participants were rated with mean scores of 16.6 and 17 respectively.

Results from paired t-tests showed a decrease in participants’ scores on loneliness, depression, and anxiety symptoms at post-intervention as compared to baseline for both interventions (see Table 3 ). The pre-post reductions in all outcomes were statistically non-significant, but the large reductions in UCLA-20 loneliness (d = -1.09) and in GAD-7 (d = -1.08) in CBT group suggested a potentially substantial magnitude of change and might be considered as clinically meaningful differences (28, 38, 39) (see Table 3 ).

Ten interviews were completed (4 CBT and 6 SP). Four themes emerged including the experience of loneliness, participant reported changes in self and behavior, barriers/facilitators to participation, and adaptations to the intervention design.

Given the exploratory nature of this pilot feasibility study, the findings should be interpreted in the context of three study limitations. First, the study used a small convenience sample of emerging adults identified from the EHR data in a large urban health system; hence there could be sampling bias, and the results may not be generalizable. Future studies will include larger samples that are powered to detect within- and between-group differences in intervention outcomes. Second, a control group was not included and thus we cannot conclude participants’ reported changes in loneliness, depression and anxiety were due to the intervention and not to other causes. There were also potential allocation and temporal biases with the sequential assignment method, which could result in unequal group characteristics unrelated to the intervention. Our future study will use a randomized controlled trial design to examine the treatment effects of these interventions against the control group and will include longitudinal follow-ups as well as analyses controlling for covariates. Third, challenges with regular attendance and varying levels of engagement may have affected outcomes; this was not analyzed due to the small sample size. Despite the limitations, results from this pilot work are informative to refine the interventions and to prepare for a future RCT.