Predictors of Electroconvulsive Therapy Use in a Large Inpatient Psychiatry Population

Objective There is limited research on reliable and clinically useful predictors of electroconvulsive therapy (ECT) use. We aimed to examine factors that predict ECT use in an inpatient psychiatric population. Design Retrospective analysis of provincial database for inpatient psychiatry. Methods This study is a retrospective analysis of a provincial database for inpatient psychiatry. The study includes all psychiatric inpatients 18 years or older in Ontario, Canada, assessed with the Resident Assessment Instrument for Mental Health (RAI-MH) within the first 3 days of admission between 2009 and 2014 (n = 153,023). The RAI-MH is a validated assessment tool which includes a breadth of information on symptoms, self-harm, functioning, social support, comorbid medical diagnoses, and risk appraisal. Multivariable analyses were performed using SAS. Results One hundred forty-five thousand seven hundred (95.2%) of patients admitted had no history of ECT treatment and were not scheduled to receive ECT. A total of 7323 (or 4.8% of the patient population) had either a history of ECT use or were scheduled to receive ECT. Overall rate of ECT use was highest in patients with a provisional diagnosis of mood disorder (7.2%) compared with schizophrenia/other psychotic disorder (3.1%) or substance-related disorder (1.7%). Women were more likely to receive ECT compared with men (overall rates of ECT use 6.2% and 3.4%, respectively). Overall rate of ECT use increased significantly with increasing age. Number of prior hospitalizations was also a strong predictor of ECT use. Conversely, patients with elevated Risk of Harm to Others, schizophrenia, or a substance use disorder were all significantly less likely to receive ECT. All variables examined were statistically significant (P < 0.0001). Higher Severity of Self Harm Scores predicted past use, but not scheduled use of ECT. Conclusions This is the largest study to date on predictors of ECT use. Utilization of RAI-MH is a novel and clinically useful method for evaluating predictors of ECT use. Predictors of ECT use within an inpatient population include: presence of a mood disorder, female sex, older age, low risk of harm to others, number of lifetime hospitalizations, lack of substance use disorder, and inability to care for self.