Time to replicate

This year marks electroconvulsive therapy’s (ECT) 80th anniversary, and although the scientific knowledge about its use, efficacy and side-effect burden is impressive, a lot of questions and dilemmas, clinicians are faced with on a daily basis, remain unanswered.

As eloquently described by Rosenman (2018), seizure threshold (ST) titration as a method of dose optimization is one of the practices that is unproven to be effective. It has not been shown to work, Rosenman concludes. But neither have other ways of choosing a stimulus dose. ST is determined by a myriad of variables, not in the least medication use. In randomized controlled trials, most patients are tapered off medication, limiting the variability in ST that is seen in daily practice, where combining medication with ECT is rule rather than exception. Rosenman seems to consider it proven that a stimulus should be above threshold in order to have a therapeutic efficacy. Determining the ST thus seems to be a reasonable option and would make formula dosing inadequate. That is not to say that using multiples of the ST is the most adequate (and efficacious) way of treating our patients. Using (half)-age-based dosing, however, is based on just as little evidence.

Rosenman rightly points out that titration can be harmful, but only refers to the fact that it might add a session to the treatment course. And to possible memory side-effects, as a consequence of doses that are (too) high. Empirical dose titration, multiplying the threshold dose by 6, does not imply ‘high dose’ ECT. On the contrary, total charge delivered might as well be lower than a dose based on the age of the patient (e.g. when a briefer pulse width is used, when patients use threshold-lowering drugs such as clozapine).

Another concern, that is not mentioned, is that titration implies, in a substantial number of cases, multiple sub-convulsive stimuli, increasing the risk of bradycardia and asystole. Moreover, seizures at threshold tend to be long, often prolonged, adding to postictal disorientation and agitation, and the need for sedating medication. An alternative adequate and evidence-based dosing-strategy would be welcomed by the field.

The paper of Rosenman highlights an important issue in clinical practice: so-called evidence-based guidelines are only marginally based on randomized evidence, and so is most of ECT-practice. We lack a (replicated) evidence base for the use of a stimulus with a 0.5 ms pulse width (Sienaert et al., 2018), a bifrontal electrode position, a fixed maintenance ECT scheme, a flexible maintenance ECT scheme. In the absence of a significant improvement after the sixth treatment, guidelines argue to switch electrode position. There is no evidence to support such a guideline. Nor do we know how to adjust a stimulus dose the best way.

The only reasonable answer to the very valid questions raised by Rosenman (2018) is research and replication. Only a small minority of practices we consider to be ‘evidence based’ have been replicated. In psychological sciences, only 1.07% of studies are replication studies (Makel et al., 2012). In psychiatry research, among 83 papers recommending effective interventions, only 16 were replicated. Replication studies showed that effects in the original studies were overestimated (Tajika et al., 2015). Only 36% of replications in psychological research proved to be statistically significant compared to 97% of the original studies (White et al., 2017). White and co-workers (2017) conclude that reproducibility should be one of the core principles of science. Unfortunately, replication studies have a low priority, for funding agencies and for journal editors alike: research is to be new and innovative. Perhaps, academic institutions should take it as their responsibility to perform and fund replication studies, to guide clinical practice in the scientific direction.

See Viewpoint by Rosenman 52: 410–414.