Identifying change-dropout patterns during an Internet-based intervention for depression by applying the Muthen-Roy model. Issue 1 (2nd January 2020)

• Record Type: Journal Article
• Authors: Arndt, Alice; Lutz, Wolfgang; Rubel, Julian; Berger, Thomas; Meyer, Björn; Schröder, Johanna; Späth, Christina; Hautzinger, Martin; Fuhr, Kristina; Rose, Matthias; Hohagen, Fritz; Klein, Jan Philipp; Moritz, Steffen
• Is Part Of: Cognitive behaviour therapy. Volume 49:Issue 1(2020)
• Page Start: 22
• Page End: 40
• Publication Date: 2020-01-02
• Subjects: Change -- dropout -- Internet-based interventions -- NMAR -- prediction; Cognitive therapy -- Periodicals; Behavior therapy -- Periodicals; 616.89142
• DOI: 10.1080/16506073.2018.1556331
• Languages: English
• ISSNs: 1650-6073

Abstract

ABSTRACT: To date, only few studies have attempted to investigate non-ignorable dropout during Internet-based interventions by applying an NMAR model, which includes missing data indicators in its equations. Here, the Muthen-Roy model was used to investigate change and dropout patterns in a sample of patients with mild-to-moderate depression symptoms ( N = 483) who were randomized to a 12-week Internet-based intervention (deprexis, identifier: NCT01636752). Participants completed the PHQ-9 biweekly during the treatment. We identified four change-dropout patterns: Participants showing high impairment, improvement and low dropout probability (C3, N = 134) had the highest rate of reliable change at 6- and 12-month follow-up. A further pattern was characterized by high impairment, deterioration and high dropout probability (C2, N = 32), another by low impairment, improvement and high dropout probability (C1, N = 198). The last pattern was characterized by high impairment, no change and low dropout probability (C4, N = 119). In addition to deterioration, also rapid improvement may lead to dropout as a result of a perceived "good enough" dosage of treatment. This knowledge may strengthen sensitivity for the mechanisms of dropout and help to consider its meaning in efforts to optimize treatment selection.