Academic journal article
By Luborsky, Lester; German, Ramaris E.; Diguer, Louis; Berman, Jeffrey S.; et al.
American Journal of Psychotherapy , Vol. 58, No. 4
This is a dedicated review of the evidence for the relation of having a period of psychotherapy and then comparing it with a measure of improved physical health. We aimed to make it the first intended-to-be-complete review of this type. Three inter-related types of studies were examined: Type 1: reduction in physical illnesses through psychotherapy, especially for the patient's survival time during the interval between diagnosis and an end point, Type 2: reduction in pain in relation to receiving psychotherapy, and Type 3: reduction in costs of treatment in relation to receiving psychotherapy. To find the relevant studies on these topics, we performed a literature search using both Psychinfo and Medline databases. An average of the effect sizes under each type was taken to calculate the mean effect size along with its confidence interval. Our results (1) on survival time for the combined severe patients, did not reach even the lowest significant level of effect size, although the low severity patients seemed to fit the hypothesis better; but the other two reduction topics, (2) and (3), clearly did achieve it.
In this article we bring together evidence for the association of (1) the benefits of a period of psychotherapy, with (2) a measure of improvement in physical health. It is an unlikely association on the face of it because the patient's treatment goals typically are to work out psychological problems not physical problems. In essence, the types of evidence to be covered in our review are about three hypotheses: type 1), the association between receiving psychotherapy with reduction in physical symptoms, type 2), the association between receiving psychotherapy with a reduction in pain and type 3), the association between receiving psychotherapy with reductions in the cost of care.
Of these 3 types of evidence, the one on "survival time" subsequent to receiving psychotherapy was the biggest issue, for it did not reach the level of significance that we expected. In addition, so far as we know, this was the first time all of such studies have been looked at together for the definition of health.
We performed a computerized search using both PsychInfo and Medline databases for all studies through 2003. For type 1, we used survival time and mortality as search terms and crossed these with having been in psychotherapy. For type 2, we used pain, somatic symptoms, physical symptoms, and health and crossed these with psychotherapy. For type 3, we used cost, cost reduction, cost effectiveness, and cost-offset as search words and crossed these with psychotherapy.
We calculated effect sizes and the 95% confidence interval for those studies reporting a statistical analysis. The statistical analyses that we converted to effect sizes were for studies reporting Student's t and F-tests with 1 as the numerator's degrees of freedom and chi-square. The effect size we used was r, which is the Pearson product moment correlation.
When calculating the overall effect size of the number of studies (k) under each type, we state the effect sizes for the studies for that type. For the three studies that had multiple years to follow-up, we used the 10-year follow-up as the limit, when it was present.
The corresponding 95 % confidence intervals for the three mean effect sizes were calculated by transforming all the effect sizes under the type into Zs. Then, we determined the critical value of / at the .05 probability level, given the degrees of freedom under each type. To this critical value we multiplied the standard deviations of all the Zs and divided it by the square root of the k number of studies. We added or subtracted this final computation from the mean of the Zs. Then, the lower and upper limits of the confidence interval were transformed from units of Z back to units of r.
Type 1. Severity of Physical Illness in Relation to Receiving Psychotherapy
There are several types of severity of illness included. …