Therapist effects over time: A multilevel modelling analysis

Therapists are differentially effective, a concept that has been termed ‘therapist effects’. Research has shown that therapist effects account for around 5% of the variability in outcomes of psychological therapy. However, there has been little research investigating whether such therapist effects are stable over time.
A systematic review was conducted to provide a contemporary synopsis of therapist effects research. The review comprised 21 studies that focussed on therapist effects for outcomes, extending the most recent review of Baldwin and Imel (2013). Results found an average therapist effect of 5% which was in common with previous findings. New research areas included low intensity treatment settings and comparisons of different outcome measures.
In order to investigate the stability of therapist effects over time, the research report analysed data from steps 2 (low intensity) and 3 (high intensity) of an Improving Access to Psychological Therapies service, comprising 12,949 patients and 141 therapists. Multilevel modelling was used to determine the therapist effect of the whole service over 40 months. Then, for five equal time periods, Markov chain Monte Carlo procedures compared therapist effects over time. Results found an overall therapist effect of 4.9% with no statistical difference between time periods. Therapist effects for step 2 of 2.9% and for step 3 of 4.9% were found. However, such effects were not statistically stable over time. Further studies with higher patient and therapist sample sizes are recommended.