Figure 1
Revised PRISMA Flow diagram.
Note: The diagram shows the iterative steps of literature screening and selection. Numbers refer to the reports identified. For some interventional studies, more than one report was located. The bottommost box shows total numbers of included reports and underlying SP projects.
Figure 2
a: Risk of bias within studies: EPHPP for non-randomized studies, summary plot. b: Risk of bias within studies: RoB2 for randomized studies, traffic light plot.
Note: Figure 2a: Stacked bars in the summary plot represent percentages of studies with a corresponding rating for individual EPHPP domains and the overall quality judgment. Figure 2b: RoB 2 assessment for randomized studies only. Rows show individual domain and overall ratings for each study, visualized as traffic light symbols.
Figure 3
Effect sizes for mental health and mental wellbeing.
Note: HADS-A Hospital Anxiety and Depression Scale – Anxiety; HADS-D Hospital Anxiety and Depression Scale – Depression; PHQ-9 Patient Health Questionnaire-9; GAD-7 Generalized Anxiety Disorder Scale-7; WEMWBS Warwick Edinburgh Mental Wellbeing Scale; SWEMWBS Short Warwick Edinburgh Mental Wellbeing Scale; Time frame: short (0–3 months), intermediate (>3–6 months) and long term (>6 months); Follow-up time in months: in case of original study reporting follow-up in weeks, these were converted to 30-day months; Kimberlee 2016 [76]: follow-up time of 3 months was intended, but length unclear in half of cases; Woodall 2018 [90]: mean follow-up time; If different numbers of months are shown, follow-up length was different for sub-populations; NA not addressed in original report; B–A before-after study; cRCT cluster randomized controlled trial; n number of patients analyzed for outcome, numbers refer to intervention/control in case of controlled studies; SMD and SE rounded by software package to two digits, confidence intervals were calculated by R package “meta” [37] from SMD/SE values calculated by R package “effect size” [35]. For Mercer 2019 [92, 93], effect sizes were taken from the original publication (adjusted SMD).
Figure 4
Effect sizes for loneliness, quality of life, and general health.
Note: UCLA University of California Los Angeles Loneliness Scale; CEL Campaign to End Loneliness ; EQ-VAS European Quality of Life Visual Analogue Scale; WHO-Quol-Bref, WHO Quality of Life short form; EQ-5D-3L/5L European Quality of Life 5 Dimensions 3 Level/5 Level; Time frame: short (0–3 months), intermediate (>3–6 months) and long term (>6 months); Follow-up time in months: in case of original study reporting follow-up in weeks, these were converted to 30-day months; B-A before-after study; cRCT cluster randomized controlled trial; n number of patients analyzed for outcome, numbers refer to intervention/control in case of controlled studies; SMD and SE rounded by software package to two digits, confidence intervals were calculated by R package “meta” [37] from SMD/SE values calculated by R package “effect size” [35]. In cases of 0.00 as a confidence interval bound, the graph shows whether the original study reported a significant (s) or non-significant (ns) result, as the plotting function only shows two digits. For Mercer 2019 [92, 93], effect sizes were taken from the original publication (adjusted SMD).
Figure 5
Effect sizes for health care utilization.
Note: GP general practitioner; ED emergency department; Time frame: short (0–3 months), intermediate (>3–6 months) and long term (>6 months); Follow-up time in months: in case of original study reporting follow-up in weeks, these were converted to 30-day months; B-A before-after study; n number of patients analyzed for outcome; SMD and SE rounded by software package to two digits, confidence intervals were calculated by R package “meta” [37] from SMD/SE values calculated by R package “effect size” [35]. In cases of 0.00 as a confidence interval bound, the graph shows whether the original study reported a significant (s) or non-significant (ns) result, as the plotting function only shows two digits.