Loneliness, the subjectively perceived lack of social relationships, is affecting a significant share of people across countries and age groups [1]. As this is associated with numerous adverse health consequences [2], effective interventions to tackle loneliness are highly needed. Although a substantial body of evidence has examined such interventions and found that psychological approaches in particular can effectively reduce loneliness [3], considerably more research is still required. Loneliness is linked to hypervigilance and negative social cognition, reinforcing withdrawal in a cycle [4], which cognitive-behavioral techniques such as social skills training and cognitive restructuring aim to break. Particularly, low-threshold digital interventions are an interesting yet underinvestigated approach with few available studies [5]. Consequently, the objective of this preregistered [6] pilot study was the evaluation of a digital app-based intervention against loneliness.

Out of the 72 included participants, 67 (93%) provided answers to the UCLA-9 after 4 weeks from the intervention start. UCLA-9 scores decreased from baseline (mean 25.6, SD 3.7) to week 4 (mean 23.5, SD 4.2). A paired t test showed a significant medium-to-large effect for loneliness (t₆₆=5.8; P<.001; d=0.71), with an average reported decrease on the UCLA-9 of 2.2 (SD 3.1). Approximately 75% (50/67) of participants reported a decrease, and 45% (30/67) of all participants reported a decrease of at least 3 points. This has been considered clinically meaningful in past work, following a reliable change index derived for the UCLA-9 in a comparable sample [4]. Higher baseline UCLA-9 scores were associated with greater reduction (r=0.25; P=.04). The effect remained stable for the 44 participants who responded to the questionnaires after 8 weeks, with scores decreasing from baseline (mean 26.2, SD 4.2) to week 8 (mean 24.0, SD 5.2; t₄₃=3.8; P<.001; d=0.57), reporting an average reduction of 2.21 (SD 3.9). Additionally, effects were found for interpersonal competencies (t₄₃=−3.1; P=.004; d=−0.46) as well as perceived social support (t₄₃=−2.4; P=.02; d=−0.37). However, while there was an average decrease of 0.77 (SD 6.8) in perceived stress, this effect was not statistically significant (t₄₃=0.75; P=0.45; d=0.13). No significant baseline differences were found between completers and noncompleters (ie, no 8-week questionnaire) at 8 weeks across demographic or clinical variables, suggesting limited attrition bias. Noncompleters showed markedly lower overall engagement (mean 14.8, SD 13.3 vs 30.9, SD 20.1 events; t_69.87=4.10; P<.001).

On average, the included participants used the app for approximately 128 (SD 119) minutes, corresponding to a weekly usage of around 21 (SD 19.8) minutes. On average, 25 (SD 19.3) exercises were completed, as well as 6 (SD 4.6) behavioral experiments, and the tracking mechanism was used 18 (SD 19.3) times. The small (r=0.14) correlation between usage time and reduction was nonsignificant in the given sample (P=.24).

In this uncontrolled pilot study, an app-based digital health intervention targeting loneliness in individuals with at least low-to-moderate levels of loneliness showed preliminary signals of reduced loneliness. The observed within-group effect size (d=0.71) falls within the range reported across existing internet-based interventions (d=0.33 to d=1.47) [5], corroborating the general promise of this intervention format. However, given the single-arm design, this interpretation is limited, and randomized controlled trials are needed to confirm the intervention’s effectiveness.

The sample was predominantly female and highly educated, with considerable dropout, limiting generalizability. Additionally, user satisfaction and barriers to use were not systematically assessed. Finally, the association between baseline severity and reduction may partly reflect regression to the mean. Interestingly, contrary to our initial hypothesis, the effect on perceived stress in this study was not significant. This may be due to limited statistical power to detect small effects.

Taken together, these preliminary results suggest app-based cognitive behavior therapy interventions may offer a promising approach to addressing loneliness as a pressing public health challenge.