We appreciate Dr. Østergaard and Dr. Kellner’s comments on our article. The mortality rate in Tørring et al.1 was 2.1 deaths (95% CI, 1.2 to 3.4) per 100,000 electroconvulsive therapy treatments and 6.0 deaths (95% CI, 2.0 to 23.0) per 100,000 electroconvulsive therapy treatments in our meta-analysis.2 The difference may appear large; however, the 95% CIs overlap and thus, the two studies are less discrepant than they appear at first sight. Three primary reasons may explain the difference between the two studies. First, as pointed out by Østergaard and Kellner, there are statistical differences. Second, our study aimed at all-cause mortality to obtain an unbiased estimate of risk after electroconvulsive therapy. Trying to identify a causal relationship between electroconvulsive therapy and death from a retrospective review of the literature is difficult at best, and exposes the analysis to bias, which we wanted to avoid.3 Third, we included all studies regardless the sample size (Tørring et al. included only studies with a minimum of 3,000 electroconvulsive therapy treatments). We choose to include all studies to limit selection bias but accept the concern of a small study bias.
The mortality rate per patient undergoing electroconvulsive therapy is not reported in the study by Tørring et al. The reason that the reported risk2 per patient is proportionally higher than per electroconvulsive therapy treatment, is that most patients undergo a series of electroconvulsive therapy treatments. In Tørring et al.’s study (table 1 from Tørring et al.1 ), each patient underwent, on average, 8 to 12 treatments, which may translate into a higher mortality rate per patient compared to per electroconvulsive therapy treatment.2
We completely agree with Østergaard and Kellner’s statement to “avoid unfounded fear of electroconvulsive therapy—especially among patients with life-threatening mental disorders where this treatment can be life-saving—it is of utmost importance that the mortality related to electroconvulsive therapy is calculated and presented in a meaningful manner.” This was our goal and we stand behind the findings of our study.
The authors declare no competing interests.