The primary outcome measure will be the time to relapse or recurrence meeting DSM-IV criteria for a major depressive episode (American Psychiatric Association, 1994) on the Structured Clinical Interview for DSM-IV (SCID, Spitzer et al., 1992). Occurrence of relapse or recurrence will be assessed after treatment (T1), and at three (T2), six (T3), nine (T4) and twelve (T5) months thereafter by trained psychologists blind to participants' treatment condition. If the interview establishes that symptoms meeting diagnostic criteria for major depression have been present since the last assessment, we shall ask participants when this episode of depression started (and ended, if they are no longer symptomatic). 'Return to treatment' will also be treated as a relapse or recurrence if, in the judgment of a blind rater, the participant experienced exacerbation of their symptoms that would have met the criteria for Major Depression in the absence of immediate treatment.
The analysis will be by 'intention to treat' (ITT). The time (in weeks) of relapse or recurrence to Major Depression, as defined above, will be the dependent variable in survival analysis. The treatment group and stratification variables will be used as predictors.
For participants who are lost from the trial we shall use their available measures and then censor them at the time of their last observation. Since only a participant's first relapse or recurrence to Major Depression will contribute to the survival analysis, the subsequent loss of that participant will not affect the analysis.
Participants who miss one or more follow-up assessments, but are then assessed at a later time point will be asked whether they have experienced a relapse or recurrence according to SCID diagnostic criteria since the last assessment, including time periods which would have been covered in missed assessments. This will enable us to assess the time to relapse and thus to censoring.
We shall use the clinician-rated Hamilton Rating Scale for Depression (HRSD, to assess severity of depression at all time points. We shall follow De Rubeis et al. (2005) and Hollon et al (2005) in using a score of 14 or more on the HRSD to indicate relapse, thus complementing the SCID diagnosis. HRSD scores also provide a quantitative measure of outcome that strengthens the dichotomised outcome of diagnosis, because, as a quasi-continuous measure, it has more power to detect differences between groups.
The other quantitative measures used at baseline, before treatment, and at times T1 to T5 are the Beck Depression Inventory (BDI-II), Beck Hopelessness Scale (BHS), Beck Scale for Suicide Ideation (BSS current) and the EQ5D. We shall calculate the 'area under the curve' (AUC) of each measure to give a single score. For the EQ5D this is known as a QALY.
Missing items within individual outcome measures will be treated according to the instructions for that measure. If two or more observations for quantitative measures are available between T1 and T5, then we shall use linear regression to estimate the missing values and the AUC. If there is only one observation available between T1 and T5 then we shall estimate the remaining values from the general slope estimated from all participants with two or more observations. If no observations for T1 to T5 are available then the AUC is also missing. If the trend among participants with full data departs from linear, then we shall use the observed trend to estimate the missing values.
For the quantitative measures we shall use a mixed-model analysis of covariance (AnCova). We shall use baseline measures as covariates, and multi-level modeling to take account of the clustering of participants within classes within centres. As covariates we shall also use the stratification variables and treatment group, together with the number of sessions attended and, in the MBCT group, number of hours of home practice. These will allow us to estimate how the response to the two treatments depends on their 'dose'.
Potential moderators to be examined include gender, residual symptoms of depression (e.g. HRSD score and BDI-II score at baseline) and stability of remission, course of previous history (chronic vs episodic), and age of onset.
Recurrence of suicidal ideation, both within episodes of major depression and over the follow-up period, is an important secondary outcome. Initially we shall examine recurrence of suicidality specifically for those participants who relapse to Major Depression and had a history of suicidal ideation or behaviour at entry to the trial. Then we shall compare severity of suicidal symptoms, as measured by the Beck Scale for Suicide Ideation (BSS current) and the MINI suicide-tracking measure, across the follow-up period for all participants, whether or not they relapsed or became suicidal. Finally we shall compare suicidal cognitions (and ability to let go of cognitions that occur), at baseline, T1 and T5 only between groups using AnCova, with T0 as covariate and T1 and T5 as separate outcomes.
By assessing cognitive measures relevant to risk of relapse to depression, namely mindfulness, suppression, self-compassion, rumination, self guides, autobiographical memory and executive capacity, before and immediately after treatment and at the end of follow-up, we can use them in an explanatory analysis to study factors that mediate efficacy. We shall use regression analysis (binary logistic regression for the dichotomous outcome of relapse and linear regression for the worst HRSD score during follow-up) to explore whether the change from T0 to T1 in each of these measures can account for the effects of treatment on risk of relapse.
Where necessary we shall transform variables closer to Normal distributions for AnCova. Where data are missing, we shall use data from alternative sources, notably therapists on the trial and referring general practitioners. We shall also use sensitivity analysis to assess the effect of including participants whose data on relapse is collected by these means.