Half of those who start an antidepressant medication may not show good enough response.Predicting antidepressant response is crucial to make sure that right patients get the right treatment.Previous research has suggested lower income, lower education, earlier age of depression onset, older age, male gender, presence of somatic symptoms, alcohol and drug use,atypical or melancholic features, anxiety symptoms,personality dysfunction,longer depression duration and concurrent psychiatric and medical comorbidities as poor prognostic factors. So far we do not have any one factor with sufficient power to predict outcome on an individual basis. Would combining various factors in a hierarchical manner in to predictive profiles help predict outcomes?
Felipe A. Jain, Aimee M. Hunter, John O. Brooks and Andrew F. Leuchter used signal detection analysis to generate quality receiver operating characteristics (QROC) using data from the Level 1 of the Sequenced Treatment Alternatives to Relieve Depression trial (STAR*D: participants with MDD were treated with citalopram).
Significant predictors of antidepressant treatment response were number of years of education ,employment status, distress from trauma reminders,level of anxiety ,episode duration, and gender. However the effect sizes were small for these individual variables. The best predictor of treatment response was years of education.Subjects with fewer than 14 years of education had a response rate of 41%, whereas those who were more highly educated had a response rate of 54%. Those with a profile of higher levels of education (14 or more years), absence of distress from trauma reminders and female gender had 63% response rate. Those with lower education, no job and anxiety symptoms had a response rate of 31% only.Hierarchical profiles, were more useful than individual factors for predicting response. 20% of the patients in STAR*D were classified into a lowest responder group, with a 31% response rate , and 20% of the patients separated into a highest responder group, which had a 63% response rate.
Socioeconomic indicators (i.e., education in the response profile, and income in the remission profile) were the most important and had greater overall predictive power for antidepressant treatment outcome than depressive symptom burden and comorbid conditions. This suggest that certain services (job training programmes, intensive social support for those from low SEC ) may be important to help minimise the detrimental effects of depression.
Summary of the article:
Jain FA, Hunter AM, Brooks JO 3rd, Leuchter AF. Depress Anxiety. 2013 Jul;30(7):624-30