ABSTRACT
BACKGROUND: One aim of personalized medicine is to determine which treatment is to be preferred for an individual patient, given all patient information available. Particularly in mental health, however, there is a lack of a single objective, reliable measure of outcome that is sensitive to crucial individual differences among patients. METHOD: We examined the feasibility of quantifying the total clinical value provided by a treatment (measured by both harms and benefits) in a single metric. An expert panel was asked to compare 100 pairs of patients, one from each treatment group, who had participated in a randomized clinical trial (RCT) involving interpersonal psychotherapy (IPT) and escitalopram, selecting the patient with the preferred outcome considering both benefits and harms. RESULTS: From these results, an integrated preference score (IPS) was derived, such that the differences between any two patients' IPSs would predict the clinicians' preferences. This IPS was then computed for all patients in the RCT. A second set of 100 pairs was rated by the panel. Their preferences were highly correlated with the IPS differences (r=0.84). Finally, the IPS was used as the outcome measure comparing IPT and escitalopram. The 95% confidence interval (CI) for the effect size comparing treatments indicated clinical equivalence of the treatments. CONCLUSIONS: A metric that combines benefits and harms of treatments could increase the value of RCTs by making clearer which treatments are preferable and, ultimately, for whom. Such methods result in more precise estimation of effect sizes, without increasing the required sample size.
