Issue link: https://read.uberflip.com/i/1503810
2.6 | Measures and data sources Two validated patient reported outcome measures are routinely used in IAPT services to monitor depression and anxiety symptoms: The PHQ-9 is a self-report measure of depression that has been widely used in research and is a regular screening measure in primary care settings (Spitzer et al., 1999). Each item is rated on 0 to 3 scale, yield- ing a summary score ranging from 0 to 27, where larger scores reflect a greater severity of depressive symptoms. A cut-off ≥10 is used to detect clinically significant depression symptoms (Kroenke et al., 2001) and a reliable change index of ≥6 points is recommended to monitor improvement or deterioration over time (McMillan et al., 2010). The GAD-7 comprises of 7 items measuring symptoms and severity of anxiety based on the DSM-IV diagnostic criteria for GAD. Each item is also rated on a 0–3 scale, yielding a total anxiety severity score between 0 and 21. A cut-off score ≥8 is recommended to identify the likely presence of a diagnosable anxiety disorder (Kroenke et al., 2007) with a reliable change index of ≥4 points (National Health Service (NHS), 2018a, 2018b; Richards & Borglin, 2011). Participants were sent reminders via email to complete these questionnaires on the SilverCloud platform at each follow-up time point. If the questionnaires were not completed following two reminder emails, a member of the research team would call the partic- ipants and offer them the opportunity to complete the questionnaires over the phone. A small number of participants availed of the option to complete hardcopy versions of the questionnaires and post them to the research team. 2.7 | Statistical analysis Within IAPT, the definition for reliable recovery is a score that moves from caseness to non-caseness (i.e., moves from above to below clini- cal cut-offs), and exceeds the measurement error of that question- naire (IAPT, 2014). A participant was considered to have met the criteria for reliable recovery if they (1) completed treatment with a score of ≤9 on the PHQ-9 and ≤7 on the GAD-7 and (2) demonstrated a reliable improvement of ≥6 points on the PHQ-9 or ≥4 points on the GAD-7, from baseline to end-of-treatment (i.e., 3 months). A par- ticipant was considered to have relapsed if (1) their post-treatment score at 3-, 6- or 9-month follow-up was ≥10 on the PHQ-9 or ≥8 on the GAD-7 and (2) they showed a reliable deterioration of ≥6 points on the PHQ-9 questionnaire or ≥4 points on the GAD-7, from the time of end-of-treatment to 3-, 6-, or 9-month follow-up. Participants were therefore categorized as in remission or relapsed at each time point. Therefore, participants who did not experience a relapse event by 9-month follow-up were considered to still be in remission. Baseline demographic and clinical characteristics were described for the total sample making up our secondary analysis, and compari- son between relapse and remission groups at baseline was done using chi-square and t tests as appropriate. Usage data, including number of logins and time spent on the platform, were collected for all participants and compared between the relapse and remission groups using independent sample t tests. A survival analysis was used to assess the durability of iCBT treat- ment effects over 9 months following treatment completion, and to determine predictors to relapse. Time-to-relapse, measured in months, was assessed using non-parametric Kaplan–Meier (KM) curves (Kaplan & Meier, 1958). Kaplan–Meier curves plot the probability of survival (i.e., remission) over time. Hazard ratios predicting time-to- relapse were also estimated with a semi-parametric Cox proportional hazards model (Cox, 1972). This model makes no assumptions about the shape of the baseline hazard function however, there are assump- tions such as (i) independence of survival times, (ii) a change in predic- tors produce proportional changes in the hazard regardless of time and (iii) a linear association between the natural algorithm of the rela- tive hazard and the predictors. For these analyses, missing data were accounted for through censoring. Data points for each participant are censored when their survival time (i.e., time to relapse, in this study) cannot be accurately determined at any given time point. This happens for a number of reasons including (1) when data are not available for a participant at a given follow-up time point, (2) when a participant drops out of the study and (3) when a participant does not experience the event, that is, does not relapse (or the study ends before they do). There- fore, each estimate at each time point is based solely on the observed data. 3 | R E S U L T S 3.1 | Sample characteristics Of the 241 eligible participants in the main RCT, 89 met the eligi- bility criteria for inclusion in this secondary analysis. The mean age in the sample was 34.83 (SD = 13.39; range = 18 to 74); 77.5% (n = 69) were females; 81.8% (n = 72) were from a White British background; and 18.2% (n = 16) were identified as having a long- term health condition. At the time of treatment completion, 76.5% (n = 65) were employed either full-time or part-time; 27.4% (n = 23) were taking medication; and 25% (n = 22) had a formal diagnosis of depression and/or anxiety disorder based on the Mini International Neuropsychiatric Interview 7.0.2 (MINI). Table 1 pre- sents an overview of the PHQ-9 and GAD-7 scores at each time point for all 241 participants in the intervention arm, and the sub- set of 89 participants who met the criteria for this secondary analysis. In terms of residual symptoms at the end of treatment, 41.6% (n = 37) showed sub-threshold symptoms of depression and 33.7% (n = 30) showed sub-threshold symptoms of anxiety. Descriptive sta- tistics and results from t tests and chi-square tests comparing partici- pants who did (n = 26, 29.2%), and did not (n = 63, 70.8%) experience a relapse event at any point during follow-up are pres- ented in Table 2. PALACIOS ET AL. 1771