Twenty-three studies (13 RCTs and 10 before-and-after studies) were included in the review (n=1,870). The included studies were of moderate quality (average score 26.5, range 16 to 36). Follow-up ranged from 1.5 to 60 months. Some 91.3% of studies showed symptom benefits that related to the main physical condition. Dropout rates in the control groups were significantly higher than those in the intervention groups (OR 1.54, 95% CI 1.06 to 2.25; 10 studies).
Short-term follow-up: Compared with controls, short-term psychodynamic psychotherapy significantly improved general psychiatric symptoms (SMD -0.69, 95% CI -0.86 to -0.52; eight studies), depression (SMD -0.97, 95% CI -1.19 to -0.74; six studies), anxiety (SMD -0.74, 95% CI -0.96 to -0.52; eight studies) and somatic symptoms (SMD -0.59, 95% CI -0.78 to -0.40; eight studies).
Long-term follow-up: Compared with controls, long-term psychodynamic psychotherapy significantly improved general psychiatric symptoms (SMD -0.70, 95% CI -0.91 to -0.48; four studies), depression (SMD -2.26, 95% CI -2.75 to -1.77; two studies), anxiety (SMD -2.28, 95% CI -2.76 to -1.80; two studies) and somatic symptoms (SMD - 0.49, 95% CI -0.77 to -0.21; three studies).
Statistically significant heterogeneity was observed for all outcomes (I2>90%). Sensitivity analyses did not materially affect most outcomes. The results of random-effects models showed that some outcomes were no longer statistically significant. The fail-safe N analyses showed that only the results for medium-term outcomes were more subject to publication bias.
Results of assessment of medium-term outcomes were reported.