A statistical model predicting the seizure threshold for right unilateral ECT in 106 patients

Area of Science:

• Neuroscience
• Psychiatry
• Biostatistics

Background:

• Directly assessing seizure threshold via electroconvulsive therapy (ECT) stimulus titration is the standard practice.
• Only 39% of ECT providers currently utilize this direct assessment method.
• Understanding factors influencing seizure threshold is vital for optimizing ECT efficacy and safety.

Purpose of the Study:

• To develop and validate a multivariate statistical model to predict seizure threshold level (STL) in patients undergoing right unilateral (RUL) ECT.
• To identify demographic and anthropometric variables that significantly influence seizure threshold during ECT.

Main Methods:

• A multivariate ordinal logistic regression model was developed using data from 66 patients undergoing RUL ECT.
• The model's predictive accuracy was cross-validated on an independent group of 40 patients.
• Key variables analyzed included age, race, gender, and inion-nasion distance.

Main Results:

• The final model was statistically significant (p < 0.0001) with an R-squared of 0.313.
• Increasing age, African-American race, and longer inion-nasion distances predicted higher seizure threshold levels.
• Female gender was associated with lower seizure threshold levels, and model prediction accuracy was fair (correlation=0.576).

Conclusions:

• Multivariate modeling can elucidate variables influencing seizure threshold in ECT.
• The developed model demonstrated fair predictive capabilities but is currently impractical for routine clinical estimation of seizure threshold.
• Empirical stimulus titration remains the recommended method for assessing individual seizure thresholds in clinical ECT practice.