The Influence of Unmeasured Confounding Effects in a Study of Antipsychotic Discontinuation in First-Episode Schizophrenia

Abstract

BPD) at 12 weeks in subsamples of people who were adherent with medication during this period, those without coexisting substance misuse, and those who had a recent history of deliberate self-harm. While conducting this analysis, we identified a typographical error that appeared in our article (1) but not in the monograph that has subsequently been published that provides additional information about the trial (2). The number of participants in the placebo arm of the trial who reported deliberate self-harm in the 6 months prior to randomization was 87 (63%), not 51 (37%) as erroneously stated in our article. Among 98 participants who did not report substance misuse at baseline, the difference in the mean ZAN-BPD score at 12 weeks between those prescribed lamotrigine and those prescribed placebo was 0.90 (95% CI521.56 to 3.37). Among 181 participants who adhered to study medication during the first 12 weeks, the difference in the mean ZAN-BPD score between those prescribed lamotrigine and those prescribed placebo was 0.52 (95% CI5 22.41 to 1.36), and among the 143 participants who reported deliberate self-harm at baseline, the difference in the mean ZAN-BPD score at 12 weeks between those prescribed lamotrigine and those prescribed placebo was 20.60 (95% CI522.73 to 1.54).Since the publication of the results of the LABILE trial, many people have told us of their surprise that the study generated negative findings and have asked us whether there was something about the group we recruited or the design of the trial that led to this result. In keeping with findings from previous research examining the views of psychiatrists (3), some clinicians have told us that they have seen firsthand how people with borderline personality disorder make a good response when prescribed medication. However, we would argue that this experience is entirely in keeping with the results of the LABILE trial that demonstrate that, when people are offered clear information by staff who provide structured follow-up, there are likely to be clinically important reductions in symptoms of emotional distress.