Response to Rosenman ‘electroconvulsive therapy stimulus titration: Not all it seems’

Dr Rosenman raises the point that innovations in clinical practice should be carefully considered and based on reliable evidence – we agree (Rosenman, 2018). However, we disagree with the conclusions he reached:

1. ‘The present form of threshold estimation is unstable and too uncertain for valid optimisation or individualisation of dose’, and also creates ‘potential harms’. The literature clearly shows that seizure threshold (ST) varies markedly across individuals. The most significant predictors, age, gender and electrode placement, each contribute only a small part of the variance in ST. In our clinical experience, now extending over 30 years, the titration method is safe, feasible, reproducible and reliably finds the same ST in the same patient on different occasions, while confirming that there is a wide variation in ST across patients. ST predicts seizure duration and seizure expression, validating its physiological significance and the reliability of its measurement.

2. ‘Threshold stimulation has proven therapeutically ineffective’. It is important to clarify that research has demonstrated right unilateral (RUL) ECT but not bilateral forms of ECT are ineffective when administered near the threshold dose. This underscores the importance of empirically determining ST. This research also established that electrical dose in excess of ST contributes to cognitive side effects regardless of electrode placement.

3. ‘Multiples applied to threshold to obtain efficacy have never been properly investigated or standardised’. In fact, high-quality, controlled trials randomising patients to different dose levels have systematically evaluated the comparative effects of RUL electroconvulsive therapy (ECT) at 1.0, 1.5, 2.5 and 6 times ST, likewise bitemporal (BT) ECT at 1.0, 1.5 and 2.5 times ST (e.g. Sackeim et al., 1993, 2000). Furthermore, other carefully conducted prospective studies and randomised controlled trials (RCTs) with detailed evaluations of efficacy and cognition have reported on the outcomes of unilateral ECT given at 5 times and 8 times ST (e.g. Loo et al., 2014). Four major and recent studies have recently compared 6 times threshold RUL to BT ECT; all found equivalent efficacy and a superior side effect profile for RUL ECT. Dosing relative to threshold was fundamental in resolving the long-standing controversies regarding electrode placement.

4. ‘The therapeutic outcomes of threshold estimation have not been separated from simple dose effects’. Several studies have examined the power of absolute electrical dose and dose relative to threshold to predict efficacy and side effects. Significant effects were reported for dose relative to threshold and not absolute dose. Since these two variables are themselves correlated, Rosenman speculated that problems with collinearity in multiple regression analyses were responsible for these results. However, in the studies by Sackeim et al. (1993, 2000) and McCall et al. (2000) zero-order correlations showed the same pattern.

Dr Rosenman cites several studies which purport to show that there is no advantage from dosing according to empirical ST titration (Aten et al., 2015; Bennett et al., 2012; Shapira et al., 1996). This literature does not support this interpretation. Two of the studies examined only BT ECT and thus could not address the impact of dosage on the efficacy of RUL ECT. The third study compared outcomes at different hospitals that used different dosing strategies, a weak design vulnerable to site differences. In contrast, the studies that have led to the widespread adoption of empirical titration and related dosing practices involved prospective, randomised trials, often in medication-free patients, who underwent double-blind and intensive assessment of therapeutic and cognitive outcomes at multiple time points.

It is also necessary to respond to Dr Rosenman’s comments about ECT-related cognitive side-effects. He takes the position that virtually all of the cognitive side effects of ECT resolve within a few days. He claims that evidence for persistent retrograde amnesia for autobiographical information is inconclusive since this domain is ‘hard to grasp and measure’. Likewise, despite evidence that many patients have persistent subjective complaints about adverse memory effects, Dr Rosenman dismisses their importance since ‘altered memory . . . is not measurable by objective tests’. In contrast, objective and subjective persistent retrograde amnesia has been documented since the 1940s. The possibility of such memory loss is described in the consent documents recommended in a variety of national and international guidelines (e.g. APA, 2001). Recent research has shown that even 6 months later patients treated with BT ECT have greater retrograde amnesia than patients treated with RUL ECT, and that these deficits are far greater than the forgetting that occurs in healthy controls (Sackeim et al., 2007; Sackeim, 2014). Indeed, even the leading critics of the measures used to assess autobiographical amnesia cited by Dr Rosenman have reported the same pattern in a recent large non-inferiority RCT comparing brief pulse RUL and BT ECT (Semkovska et al., 2016). Another replicated phenomenon is that the post ECT magnitude of autobiographical amnesia is predicted by the duration of postictal orientation, providing a clinically useful index of potential long-term adverse effects. Critically, unlike Dr Rosenman’s claims, three recent studies have shown significant associations between the magnitude of retrograde amnesia, objectively assessed, and subjective complaints about ECT’s cognitive impact. Furthermore, scales have been developed that are feasible for clinical use in monitoring and assessing memory and cognitive effects of ECT (Martin et al., 2017). In conclusion, the literature amply demonstrates that technical factors in the administration of ECT (electrode placement, stimulus dosing, waveform) impact on efficacy as well as short and long-term cognitive outcomes. Best practice ECT should be built on this evidence base.

See Viewpoint by Rosenman 52: 410–414.