MDD diagnosis: All 465 participants had completed the K-SADS interview at baseline. The vast majority (n=454, 97.6%) fulfilled criteria for depression, while 11 (2.4%) did not meet criteria. At follow-up, K-SADS data was available for 282 participants (60.7%) of whom 212 no longer met criteria for depression (45.6%) and 70 still met criteria (15.1%). For an additional 33 patients diagnostic status could reliably be derived from the permuted imputations (i.e. participant was consistently classified as either depressed or recovered in >75% of permutations; at a threshold of 75%, the ratio of diagnosed vs non-diagnosed in the real+imputed dataset was comparable as the ratio in the real dataset only [0.33]). For the remainder of patients no imputation could reliably be estimated, hence for these mdd diagnosis at follow-up was left missing. The final imputed data for MDD status at follow-up consisted of 315 participants of whom 76 met diagnostic criteria for depression and 239 had recovered.