Background: The highly prevalent late-life loneliness, together with its deleterious health impacts, calls for increasing attention to the need for effective interventions targeting on this growing public health problem. With the increasing evidence on interventions for combating loneliness, it is timely to identify their comparative effectiveness.

Objective: This systematic review, meta-analysis and network meta-analysis was to identify and compare the effects of various non-pharmacological interventions on loneliness in community-dwelling older adults.

Methods: Systematic search was conducted in nine electronic databases from inception to 30th March 2023 for studies investigating the effects of non-pharmacological interventions on loneliness among community-dwelling older adults. The interventions were categorized according to the nature and purpose of use. Pairwise meta-analysis and network meta-analyses were sequentially performed to identify the effects of each category of interventions and their comparative intervention effectiveness, respectively. Meta-regression was performed to examine any influence of study design and participants' characteristics on the intervention effectiveness. The study protocol was registered at PROSPERO (CRD42022307621).

Results: A total of 60 studies with 13,295 participants were included. The interventions were categorized as psychological interventions, social support interventions (by digital and non-digital means), behavioral activation, exercise intervention with and without social engagement, multi-component intervention and health promotion. Pairwise meta-analysis identified the positive effect of psychological interventions (Hedges' g = -2.33; 95%CI [-4.40, -0.25]; Z = -2.20, p = 0.003), non-digital social support interventions (Hedges' g = -0.63; 95%CI [-1.16, -0.10]; Z = 2.33, p = 0.02) and multi-component interventions (Hedges' g = -0.28 95%CI [-0.54, -0.03]; Z = -2.15, p = 0.03) on reducing loneliness. Subgroup analysis provided additional insights: i) social support and exercise interventions which integrated active strategies to optimize the social engagement demonstrated more promising intervention effects; ii) behavioral activation and multicomponent interventions worked better for older adults who were male or reported loneliness, respectively, and iii) counseling-based psychological interventions was more effective than mind-body practice. Network meta-analysis consistently pointed to the greatest therapeutic benefits of psychological interventions, and this was followed by exercise-based interventions, non-digital social support interventions and behavioral activation. Meta-regression further suggested that the therapeutic effects of the tested interventions were independent of the various factors relating to study design and participants' characteristics.

Conclusions: This review highlights the more superior effects of psychological interventions in improving loneliness among older adults. Interventions which have an attribute to optimize social dynamic and connectivity may also be effective.

Tweetable abstract: Psychological intervention is the best to beat late-life loneliness, but increasing social dynamic and connectivity may add an impact.