Anaesthesia-related factors and seizure quality and duration during modified electroconvulsive therapy: a retrospective patient-clustered analysis of 1379 sessions in 108 patients

Abstract Background Modified electroconvulsive therapy (m-ECT) depends on eliciting an adequate therapeutic seizure. The contributions of anaesthetic and ventilatory factors remain incompletely quantified, and because patients contribute multiple repeated sessions, treating these sessions as independent may overstate precision. We examined factors associated with seizure quality and electroencephalographic (EEG) seizure duration, using patient-clustered analyses for primary inference. Methods We retrospectively analysed 1379 m-ECT sessions in 108 patients (median, 10 per patient) at one centre. Hyperventilation was delivered predominantly by two-handed facemask pressure-controlled ventilation. Good-quality seizures were defined by an ictal EEG regularity score of 5 or 6 and a postictal suppression score of 2 or 3. Primary analyses modelled good-quality seizures by multivariable logistic regression and EEG seizure duration by multiple linear regression, with cluster-robust standard errors clustered on the patient. Explanatory variables were age, sex, body mass index (BMI), anaesthetic agent (thiamylal or propofol), stimulus output, machine-measured weight-indexed minute ventilation (MV/kg), and mask-measured end-tidal carbon dioxide (EtCO 2 ). Session-level models treating sessions as independent were exploratory. A worst-case sensitivity analysis re-included the 12 sessions without an elicited seizure (classified as poor quality, 0 s). Results Good-quality seizures occurred in 1002 of 1379 sessions (72.7%). In patient-clustered models, good-quality seizures were less likely with older age (odds ratio [OR], 0.95 per year), male sex (OR, 0.42), and higher stimulus output (OR, 0.988 per %); the stimulus-output association most plausibly reflects underlying differences in seizure threshold. The anaesthetic agent (propofol vs. thiamylal: OR, 0.50; 95% confidence interval, 0.15–1.65) and mask-measured EtCO 2 were not significantly associated with seizure quality. No variable was significantly associated with EEG seizure duration. Re-including the 12 sessions did not alter these results. Conclusions Under patient-clustered inference, older age, male sex, and higher stimulus output remained associated with poorer seizure quality, whereas no factor was significantly associated with EEG seizure duration. Session-level associations of the anaesthetic agent with seizure quality, and of MV/kg, BMI, and male sex with EEG seizure duration, did not persist under patient clustering; these findings are hypothesis-generating and should not inform anaesthetic management. Mask-measured EtCO 2 showed substantial measurement error, most plausibly from mask leak.