Digital interventions to promote psychological resilience: a systematic review and meta-analysis.

"supplementary data 3 presented the characteristic of 101 included studies (comprising 20010 participants) published between 2007 and 2023 with the vast majority being published from 2015 onwards (83 2%).; most studies were performed in the united states (31 studies30 7%) germany (11 studies10 9%) and the united kingdom (7 studies6 9%; see supplementary data 3 for all countries).; the remaining 28 studies (27 7%) used other theoretical approaches (see supplementary data 3 ) with only three studies (3 0%) employing interventions grounded in specific resilience theories.; there was a moderate to high risk of bias (see fig 2 and supplementary data 4 ).; neither contour-enhanced funnel plots (see supplementary data 5 ) nor meta-regression models provided evidence for a publication bias at post-intervention assessments for mental distress qm (1) = 0 77 p = 0 384 and positive mental health qm (1) = 1 08 p = 0 302 while the meta-regression model on resilience factors suggested funnel plot asymmetry qm (1) = 4 52 p = 0 039.; at follow-up assessments we found no evidence for a publication bias neither using a regression-based approach [mental distress: qm (1) = 0 05 p = 0 821; positive mental health: qm (1) = 0 49 p = 0 488; resilience factors: qm (1) = 3 61 p = 0 076] nor based on the inspection of contour-enhanced funnel plots (see supplementary data 6 ).; eighty-five studies (reporting 150 effect estimates) contributed to the meta-analysis on mental distress (see supplementary data 7 for forest plot).; seventy-seven studies (reporting 123 effect estimates) contributed to the meta-analysis on positive mental health outcomes (see supplementary data 8 for forest plot).; forty-five studies reported 64 effect estimates for resilience factors (see supplementary data 9 for forest plot).; according to grade ratings certainty of evidence was very low for all outcome categories (see supplementary data 10 ).; analyses on mental distress at follow-up assessments were based on 38 studies (comprising 64 effect estimates; see supplementary data 11 for forest plot) and yielded again evidence for a small favorable effect of digital resilience interventions over comparators smd = -0 24 95% ci [-0 35 -0 13] with substantial heterogeneity q (63) = 205 32 p < 0 001 mainly resulting from between-study differences while between-outcome differences were of minor relevance qm (3) = 0 63 p = 0 602.; at follow-up assessments based on 57 effect estimates from 36 studies (see supplementary data 12 for forest plot) there was evidence for small favorable effects of digital resilience interventions on positive mental health smd = 0 19 95% ci [0 11 0 26] with substantial heterogeneity q (56) = 139 66 p < 0 001 mainly resulting from between-study differences while between-outcome differences were of minor relevance qm (7) = 1 68 p = 0 352.; eighteen studies (reporting 25 effect estimates) assessed resilience factors at follow-up assessments (see supplementary data 13 for forest plot) finding an overall small favorable effect smd = 0 19 95% ci [0 08 0 30] with substantial heterogeneity q (24) = 83 62 p < 0 001 which derived mostly from between-study differences while between-outcome differences were non-significant qm (5)= = 1 11 p = 0 405.; also for follow-up assessments grade ratings indicated a very low certainty of evidence for all outcome categories (see supplementary data 14 ).; for follow-up assessments the results of the moderator analysis are presented in supplementary data 15 .; = 0 80) impacted on our results (see supplementary data 16 ).; however in some cases previously significant effects were non-significant as those analyses were based on a smaller number of effect estimates (see supplementary data 17 ).; we found no evidence for differences between studies examining digital resilience interventions during the covid-19 pandemic and pre-pandemic studies p >= 0 281 (see table 2 and supplementary data 15 ).; we examined whether effect estimates were different for studies with small digital component and found no evidence for a difference neither at post-intervention nor follow-up assessment p >= 0 350 (see table 2 and supplementary data 15 ).; we searched five databases from 2019 to 2022 and identified studies published before 2019 by means of systematic reviews on resilience and health-promoting interventions (see supplementary data 2 for our search rationale).; moreover we made minor changes from the preregistration of the review which are described in supplementary data 1 .; differences between the preregistration of the review (prospero preregistration-id: crd42021286780) and the final review are presented as supplementary data 1 .; second we searched for primary studies to cover the period from january 1 2019 to august 15 2022 (see supplementary data 2 for the rationale of this two-step search strategy).; search strategies are presented in supplementary data 2 ."

"Competing interests Authors SKS, LvB, FK, TW, FZ, MW, KL, and OT have been involved in the development of a non-commercial digital resilience intervention (i.e., ResiLIR). This intervention is not part of this systematic review and had no impact on the review design, our analyses, reporting or interpretation of our results. Authors MW and AK were involved in one study included in this review (Behrendt et al. 2023), which had no impact on the design of our review and the interpretation of our findings. All authors declare no financial competing interests."

"Funding Open Access funding enabled and organized by Projekt DEAL."

"This systematic review adheres to the standards of the Cochrane Collaboration and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). Differences between the preregistration of the review (PROSPERO preregistration-ID: CRD42021286780) and the final review are presented as Supplementary Data . In our protocol, we planned several moderator analyses (PROSPERO preregistration-ID: CRD42021286780). Due to the substantial between-study heterogeneity in our primary analyses on all outcome categories, these analyses were performed for the main analyses on mental distress, positive mental health, and resilience factors. For categorical variables (e.g., type of control group) we used subgroup analyses, while meta-regressions were used for omnibus moderation tests and continuous moderators (e.g., publication year), with a significant QM statistic indicating the presence of a moderator effect. Moderator analyses were performed for sociodemographic sample characteristics (i.e., age, gender balance per sample, population type), delivery format (eHealth vs. mHealth vs. mixed), theoretical foundation (CBT vs. coping literature vs. mindfulness vs. positive psychology vs. mixed), availability of guidance (unguided vs. guided [i.e., availability of human guides or coaches who support intervention delivery]), intervention type (standalone vs. blended), individualization (standardized vs. individualized), intervention intensity in weeks, publication year (as proxy of improvements over time), and type of control group (passive controls [i.e., no intervention/waitlist] vs. low-intensity active controls vs. high-intensity active controls)."