In the past post, I mentioned that that concepts and techniques specific to various types of psychotherapy may account for as little as 1% of outcomes. Here is a fuller accounting of the numbers for “percentage of variability of outcomes” in psychotherapy - care of Laska and Gurman (2014):

  • Common factors to all effective therapies (therapeutic alliance, empathy, goal consensus/collaboration, positive regard/affirmation, congruence/genuineness and therapist characteristics): 43%
  • Differences between treatments - <1.0%
  • Specific ingredients (dismantling) - 0.0%
  • Adherence to protocol - <0.1%
  • Rated competence in delivering particular treatment – 0.5%

And the rest care of Lambert, M. J., & Bergin, A. E. (1994):

  • Extra-therapeutic events: 40%
  • Placebo effect (expectations): 15%

The above adds up to about 100%.

The Laska and Gurman paper posit the following common factors as “necessary and sufficient for change: (a) an emotionally charged bond between the therapist and patient, (b) a confiding healing setting in which therapy takes place, (c) a therapist who provides a psychologically derived and culturally embedded explanation for emotional distress, (d) an explanation that is adaptive (i.e., provides viable and believable options for overcoming specific difficulties) and is accepted by the patient, and (e) a set of procedures or rituals engaged by the patient and therapist that leads the patient to enact something that is positive, helpful, or adaptive.” (p. 469)

Laska and Gurman argue that therapies and treatments (“interventions”) that contain all these factors are likely to be efficacious. Note that some of these factors enhance the placebo effect - hope and higher expectations would be inspired by a caring therapist with a believable explanation for one’s troubles and believable options for change. This makes any clear division between “common factors” and “placebo effect” somewhat problematic.

The two papers also didn’t provide estimates for the contribution of client characteristics to outcomes, although I’d imagine client characteristics are an important source of variance. And where does “regression towards the mean” enter the picture? Bottom line: the above percentages are suggestive but the actual numbers and variables could use some tweaking.

Laska and Gurman point out that although random controlled trials (RCTs) are the gold standard in research, they are often flawed and so the conclusions one can draw from them are limited. For example, some RCTs have active comparison interventions that are supposed to match the main intervention in important respects (e.g., length of time, number of sessions). If the main intervention has better outcomes than the comparison intervention, the researchers may conclude there is something specific to the main intervention that made the difference. But most comparison interventions don’t include all the “necessary and sufficient” factors common to all effective interventions, so successful outcomes in the main intervention may not be attributable to anything special it does.

What we need are studies that compare interventions that share all the common factors and aim to inspire the same degree of hope, expectation, and buy-in, not just for the subjects but for the therapists as well. Then maybe we’ll be closer to designing therapies that offer more than what is commonly available.

References

Kevin M. Laska and Alan S. Gurman Expanding the Lens of Evidence-Based Practice in Psychotherapy: A Common Factors Perspective 2014, Vol. 51, No. 4, 467–481.

Lambert, M. J., & Bergin, A. E. (1994). The effectiveness of psychotherapy. In A. E. Bergin & S. L. Garfield (Eds.), Handbook of psychotherapy and behavior change (4th ed., pp. 143–189). New York, NY: Wiley.