Proposal of Measurement Occasions for Unbiased Evaluation of Psychotherapy

Abstract

Studies complementing the assessment of symptoms right before (t1), right after therapy (t2), and at follow-up (t3) with an assessment of symptoms preceding the waiting period without intervention (t0) have revealed substantial t0–t1 changes. We discuss this phenomenon based on our own data and address the following questions: does it make sense to compare symptoms at the beginning of therapy (t1) with symptoms at the end of therapy (t2) or at follow-up (t3)? Or does it make more sense to use t0 instead of t1? We argue for the latter alternative based on the following reasons. (1) Symptom descriptions at t0 are realistic. (2) Expecting therapy success mitigates symptom descriptions at t1. (3) Security signals emitted from the therapy context also mitigate symptoms, especially anxiety, at t1. (4) Regression toward the mean reduces the validity of single occasion assessments. Controlling for regression requires two occasions of measurement with a short time interval at t0 (t01 and t02). It follows from this reasoning that therapy success should be evaluated using the t02–t2 and t02–t3 intervals. Single case evaluations require reliable critical differences. This will be illustrated using a concrete example. The validity of treatment evaluation can be increased via the elimination of non-pathological symptom scores. A simplified calculation of cut-off scores can facilitate applied treatment evaluation. Unspecific t0–t1 changes do not challenge therapy effects according to t1–t2 changes. Rather, they are part of the whole therapy process.