Disagreement between therapist raters and independent evaluators in a controlled clinical trial of interpersonal psychotherapy for depressed pregnant women.

Abstract

The randomized controlled trial in which both the patient and the treating clinician are kept blinded to the treatment is the "gold standard" for treatment research assignment. However, in psychotherapy research, evaluations can only be single blind; thus, such studies are inherently more limited. A 12-week, bilingual, parallel-design, controlled clinical treatment trial compared interpersonal psychotherapy for antepartum depression (IPT-P) with a parenting education program (PEP) provided to a control group. An outpatient sample of 142 women who met DSM-IV criteria for major depressive disorder was randomly assigned to IPT-P or PEP between September 2005 and May 2011. Only 110 cases were assessed at baseline and had at least 1 other treatment week of paired ratings by a therapist and a blinded independent evaluator (IE). The 17-item Hamilton Depression Rating Scale and the Clinical Global Impressions Scale were administered weekly by a therapist and every 4 weeks by a blinded IE. We examined cross-informant agreement on ratings of mood and global improvement and severity. Nonblinded therapists consistently rated the IPT-P treatment group as more improved than the PEP control group throughout treatment, whereas the ratings by the blinded IE were significantly higher than the therapist ratings, indicating less improvement in the IPT-P group compared with the control group. The ratings suggest that rater bias may have caused the therapist raters to perceive subjects as more improved because of the expectation that IPT-P would be more effective than the PEP control condition. Ratings in psychotherapy research must be made by anonymous participation in treatment and an independent clinical evaluator who is blind to all therapy.