Dropout in self-guided web-based interventions for depression can be predicted by several variables
•,.
...
ABSTRACT FROM: Karyotaki E, Kleiboer A, Smit F, et al. Predictors of treatment dropout in self-guided web-based interventions for depression: an ‘individual patient data’ meta-analysis. Psychol Med 2015;45:2717–26.
What is already known on this topic
Web-based interventions may constitute an effective form of treatment for depression compared with face-to-face treatments.1 However, self-guided interventions show less-promising results and higher dropout rates than guided web-based interventions,2 since human support increases treatment adherence through accountability to a therapist. It is therefore essential to identify the characteristics of individuals and interventions related to treatment dropout in unguided interventions to increase their efficacy and foster adherence.
Methods of the study
The study brings together data from separate studies to undertake an individual patient data (IPD) meta-analysis to identify sociodemographic, clinical and intervention characteristics that predict dropout in self-guided web-based interventions for people with depressive symptoms. The primary studies were selected from an existing database of randomised controlled trials (RCTs) (http:evidencebasedpsychotherapies.org). The quality of the studies included was assessed by two independent reviewers according to four criteria from the Cochrane risk of bias assessment tool. Data extraction and preparation were also carried out by two independent authors who identified the variables common to most data sets (ie, age, education, anxiety symptoms, number of modules completed).
The statistical analysis only included data from the experimental groups, since the focus was on the identification of the variables associated with dropout. This was accomplished using a three-step Poisson regression model to obtain the relative risk of treatment dropout. IPD meta-analysis included data from 2705 participants in 10 trials. Most of the interventions were based on cognitive–behavioural therapy (CBT), one used problem-solving therapy and one interpersonal therapy. The majority of participants was women (72%), highly educated (university), aged between 25 and 34 and with a high degree of severity (mean score at baseline on Center for Epidemiologic Studies Depression Scale: 35.5 and 28.4 on Beck Depression Inventory). Forty per cent dropped out before completing 25% of the treatment modules, 59% before completing half the treatment modules and 70% before completing 75% of the treatment modules. Overall, only 17% completed all the treatments modules (452 of 2705).
What does this paper add
Results revealed that being male (RR=1.08, 95% CI 1.03 to 1.13), having lower educational attainment (RR=1.26, 95% CI 1.14 to 1.39), being older (RR=0.94, 95% CI 1.01 to 1.38) and having comorbid anxiety symptoms (RR=1.18, 95% CI 1.01 to 1.38) significantly increased the risk of dropping out before the completion of treatment.
The effect of the intervention type (CBT vs non-CBT) was less clear due to differences in the number of modules completed in CBT as compared to non-CBT.
Severity of depression (RR=1.05; 95% CI 0.99 to 1.11), employment status (RR=0.97; 95% CI 0.79 to 1.20) and relationship status (RR=1.09; 0.95 to 1.25) were non-significant variables for dropout.
Limitations
Only 10 of 13 eligible RCTs had IPD sets that could be accessed and contributed to the analysis. This might have resulted in a lower capacity to predict effects in some of the variables of interest.
Some predictor variables were not reported across all 10 studies, which also affected the power of the analysis.
Participants in the studies were regarded as different from patients in clinical samples since they were recruited in the community and were proactively seeking help, meaning that the findings cannot be generalised to the population with depression as a whole.
The fact that patients were aware of the group to which they had been assigned (experimental vs control) might not only have affected the response to the intervention,3 but also compliance and the use of adjunct interventions, with those in the control group being less willing to comply and having lower expectations of the benefits.
What next in research
Future RCTs should address the particular needs of individuals with comorbid anxiety symptoms, who are male, with low educational attainment, and young in order to tailor web-based interventions to meet their requirements and assess the improvement in adherence.4
Do these results change your practices and why?
I am not a clinician; however, I think that these findings change our practice as researchers in that they will be taken into account when designing self-guided interventions that are tailored more to reduce dropouts. More effective web-based interventions can have an impact on clinical practice as clinicians are now more inclined to recommend the use of these tools to a wider range of patients with mild depression as a compliment or substitute of face-to-face interventions when there are barriers to attend a clinic.
Competing interests: None declared.
Provenance and peer review: Commissioned; internally peer reviewed.
Cuijpers P, Donker T, van Straten A, et al. Is guided self-help as effective as face-to-face psychotherapy for depression and anxiety disorders? A systematic review and meta-analysis of comparative outcome studies. Psychol Med2010; 40:1943–57.
Richards D, Richardson T. Computer-based psychological treatments for depression: a systematic review and meta-analysis. Clin Psychol Rev2012; 32:329–42.
Cuijpers P, Karyotaki E, Andersson G, et al. The effects of blinding on the outcomes of psychotherapy and pharmacotherapy for adult depression: a meta-analysis. Eur Psychiatry2015; 30:685–93.
Kelders SM, Kok RN, Ossebaard HC, et al. Persuasive system design does matter: a systematic review of adherence to web-based interventions. J Med Internet Res2012; 14:e152.