Attrition and adherence in smartphone-delivered interventions for mental health problems: A systematic and meta-analytic review.

Abstract

Objectives: Although the efficacy of smartphone-delivered interventions for mental health problems is emerging, randomized controlled trials (RCTs) of smartphone interventions are characterized by high rates of attrition and low adherence. High attrition and low adherence may threaten the validity of RCT findings, so a better understanding of these phenomena is needed. We examined attrition and adherence in 70 RCTs of smartphone interventions. Method: Four online databases were searched for RCTs of mental health interventions delivered via smartphones. Results: The mean meta-analytic study attrition rate was 24.1% (95% CI [19.3, 29.6]) at short-term follow up and 35.5% (95% CI [26.7, 45.3]) at longer-term follow up. These rates varied according to target mental health condition. Attrition rates were significantly lower in trials that delivered an acceptance-based intervention, offered participants monetary compensation, and reminded participants to engage in the intervention, and were significantly higher in trials that used an online enrollment method (relative to telephone or in-person enrollment). No participant-level baseline characteristic reliably predicted attrition. Evidence of attrition bias came from many RCTs not conducting intention-to-treat analyses. However, the mean difference in the between-groups effect size on primary outcomes in trials that reported both per protocol an intention-to-treat analyses was only ⌬d ϭ 0.18. Adherence rates were also suboptimal based on our qualitative synthesis; several participants failed to download the intervention, and intervention usage consistently declined over the course of the trial. Conclusion: Study attrition and low adherence are common, problematic, and may undermine the validity of findings in RCTs of smartphonedelivered interventions for mental health problems. What is the public health significance of this article?This study suggests that a significant proportion of people enrolled in randomized trials of mental health smartphone apps drop out from the study and fail to use the intervention as intended. Developing retention strategies may be important for improving mental health problems.

Insomnia Interventions in the Workplace: A Systematic Review and Meta-Analysis

“…As for the use of smartphones in mental health interventions, another meta-analysis pointed out that psychological interventions delivered via these devices reduced anxiety [ 50 ]. Nevertheless, a recent meta-analysis has showed that such interventions with smartphones were characterized by high rates of attrition and low adherence [ 51 ].…”

A survey study of attitudes toward, and preferences for, e‐therapy interventions for eating disorder psychopathology

“…Emerging empirical support also exists for the efficacy of e‐therapy interventions for EDs (Beintner, Jacobi, & Taylor, 2012; Beintner, Jacobi, & Taylor, 2014; de Zwaan et al, 2017; Loucas et al, 2014; Melioli et al, 2015), with cognitive, behavioral, and dissonance‐based prevention (e.g., Student Bodies , the Body Project ) and treatment (e.g., enhanced cognitive behavior therapy, Salut ) programs producing moderate reductions in core risk factors (Cohen's d 's range from 0.20 to 0.60) and symptoms (Cohen's d 's range from 0.40 to 1.00). However, this enthusiasm for e‐therapy has, at times, been balanced by concerns regarding high rates of attrition (Eysenbach, 2005; Linardon & Fuller‐Tyszkiewicz, 2020), and concerns that efficacy is lower for entirely automated or self‐guided e‐therapy use (Linardon et al, 2019). The notion that self‐guided e‐therapy may be best suited as a low intensity treatment within a stepped care approach has some empirical support (Karyotaki et al, 2017), but is unlikely to be satisfactory for all.…”

Manaaki – a cognitive behavioral therapy mobile health app to support people experiencing gambling problems: a randomized control trial protocol

“…Smartphone apps have been shown to support selfmanagement and behaviour change for smoking cessation [23], cardiac rehabilitation [24], healthy lifestyle [25], diabetes [26], HIV [27], nutrition [28], mental illness [29] and youth driving [30]. In a recent metaanalysis of smartphone interventions for mental health problems, smartphone apps outperformed waitlist controls with small to medium effect sizes [31]. The effect size was greater for apps that had a theoretical basis in CBT.…”

“…Stratified block randomisation (created by the study statistician), using block sizes of 2 and 4, will be used to randomly allocate participants. To ensure balance on potential confounders, randomisation will be stratified by ethnicity (Māori, Pacific, Other), sex and gambling symptom severity using the Gambling Symptom Assessment Scale (G-SAS): dichotomised to mild to moderate gambling symptom severity (G-SAS score 0-30) and severe to extreme gambling symptom severity (G-SAS score [31][32][33][34][35][36][37][38][39][40] to ensure a balance in these key characteristics.…”

“…Total scores on the G-SAS range from 0 to 48, with higher scores indicating greater gambling symptom severity. Scores on the G-SAS are categorised as extreme (41)(42)(43)(44)(45)(46)(47)(48), severe (31)(32)(33)(34)(35)(36)(37)(38)(39)(40), moderate (21)(22)(23)(24)(25)(26)(27)(28)(29)(30), or mild (8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). The G-SAS has demonstrated high internal consistency and good convergent validity with other measures of gambling symptom severity [50].…”