Twenty-six randomised controlled trials were included in the review (n=2,398 patients).
Hypnosis was found to have a statistically significant, large, beneficial effect on emotional distress from medical procedures. Using a random-effects model, the authors found a mean effect size of 0.88 (95% CI 0.57 to 1.19). Using a funnel plot they calculated the number of trials with an effect size of 0 that were required to decrease the overall mean effect size to small (123 trials) or medium (28 trials). Statistical analysis showed that the trials were heterogeneous and that 87.14% of the variance was explained by factors other than chance.
The moderating factors were: hypnosis was statistically more effective for children than adults; the sample size was inversely correlated with the effect size, but this may have been driven by one particularly large, negative trial; the type of control condition did not significantly affect the effect size; hypnosis interventions were significantly more effective than suggestion ones; live hypnosis was significantly more effective than recorded hypnosis; hypnosis was significantly more effective when delivered prior to the procedure, or prior and during the procedure, rather than just during the procedure; and there was no difference in the effect size where it had to be imputed.