A wide variety of physical health outcomes were measured but many not reported. Objectively measured physical health in people with pre-existing morbidity either found no difference or an improvement for the intervention groups. Objectively measured health centre visits showed no significant differences between intervention and control (WMD 0.06, 95%CI 0.26 to +0.13, random effects). Self-report health centre visits showed a similar result. Self-assessed health behaviours, where reported, were found not to be different between the two groups. Other questionnaire measures showed conflicting results. In people with pre-existing morbidity, there were improvements in the intervention groups compared to controls for pain, sleep quality, physical dysfunction, physical symptoms, fibromyalgia impact, health interference with daily functioning and perceived somatic symptoms. None of the RCTs demonstrated worse physical health.
For psychological outcomes there was more positive and negative mood for intervention compared to control but no differences in anxiety, depression or impact of events. Other psychological outcomes, where reported, either showed no difference or a mixture of results with no clear trend or conflicting results.
There is no clear evidence to demonstrate the efficacy of the intervention reviewed. This finding is contrary to a previous meta-analysis and numerous recent editorials on the subject. It may be that the rather biased reporting of outcomes has resulted in a more positive impression of the intervention than is actually the case, but neither is there any evidence that it does any harm. Without solid evidence of effectiveness there is a danger of its proliferation in inappropriate circumstances in clinical practice. There is a need for a large, good quality RCT, adequately powered to detect a small effect size, to establish whether this type of emotional disclosure has any effect at all.