Prediction of response to ECT with routinely collected data in major depression

J Affect Disord. 2005 Jun;86(2-3):323-7. doi: 10.1016/j.jad.2005.03.008.

Abstract

Background: Little is known about the possibility to predict response to electroconvulsive therapy (ECT) in patients with major depression. The aim of this study is to create an index for the prediction of response to ECT in an individual patient.

Methods: Fifty-three depressive patients referred for ECT were included. Poor response was defined as a decrease in Hamilton Rating Depression Scale less than 50%. With multivariable analyses a simple index of independent predictors was constructed.

Results: Thirty-one patients (58%) showed poor response. The index comprised age < 65 years, psychotic depression, refractory to antidepressant medication, and personality disorder. It discriminated poor response patients reasonably well with an area under the receiver operating characteristic curve of 0.76 (95% confidence interval 0.63-0.89).

Limitations: Retrospective chart review, a relatively small study size and confining to the short-term response.

Conclusions: Response to ECT may be predicted using an index with four patient characteristics. Before implementation, however, validation of the index in future patients is mandatory.

Publication types

  • Comparative Study

MeSH terms

  • Age Factors
  • Aged
  • Antidepressive Agents / therapeutic use
  • Comorbidity
  • Data Collection
  • Depressive Disorder, Major / diagnosis
  • Depressive Disorder, Major / epidemiology
  • Depressive Disorder, Major / therapy*
  • Electroconvulsive Therapy*
  • Female
  • Humans
  • Male
  • Medical Records / statistics & numerical data
  • Middle Aged
  • Netherlands / epidemiology
  • Outcome Assessment, Health Care / statistics & numerical data
  • Personality Disorders / epidemiology
  • Probability
  • Psychiatric Status Rating Scales
  • Psychotic Disorders / epidemiology
  • ROC Curve
  • Retrospective Studies
  • Treatment Outcome

Substances

  • Antidepressive Agents