Patients are often concerned about possible negative effects of electroconvulsive therapy (ECT) on their cognitive performance, and as a result may sometimes be referred for neuropsychological testing. It can frequently prove difficult to advise patients, not only because of the emotion surrounding ECT, but also because of the complexities involved in interpreting neuropsychological test results in this clinical group, where many variables affect test scores, including of course depression itself. Kirov et al ^{Reference Kirov, Owen, Ballard, Leighton, Hannigan and Llewellyn1} report potentially very important data from their excellent long-term study to inform how we advise our patients about ECT, cognition and neuropsychological testing. These data are largely very reassuring, particularly in view of the number of important known confounds they controlled for in their study, and were mostly in accordance with previous findings. However there remain a few issues pertaining to specifically neuropsychological testing that may need further exploration in future studies.

The first issue relates to the accepted current practice in cognitive assessment, that in order to more meaningfully interpret an individual patient's neuropsychological test performance, and specifically detect likely significant change, patients need to be compared against their own pre-morbid level of general cognitive ability (or ‘norm’). To achieve this, clinicians use tools such as language-based tests, demographic formulas or the ‘best performance method’ described by Lezak et al ^{Reference Lezak, Howieson, Bigler and Tranel2} to determine pre-morbid intellectual function. Although there were baseline assessments as part of the protocol, the present study does not provide any such data regarding the more individual comparison standard of a patient's pre-morbid general cognitive ability. It is acknowledged, however, that these data might be difficult to obtain and interpret for many patients in this clinical population.

Clinically meaningful change on test performance, as opposed to statically significant change, can sometimes be difficult to detect. Furthermore, change on a given test with repeated testing can be more difficult to identify if the baseline scores are already well below average – the well-known ‘floor effect’ seen when testing patients with widely distributed low performance over different cognitive domains. In Kirov and colleagues' study, their Trail Making Test data, when compared with normative data, ^{Reference Tombaugh3} seem to be potentially already below average at baseline. Again, this might have been a function of participants' depression or other factors that are known to affect performance on a test with high sensitivity such as the Trail Making Test, rather than an already present cognitive impairment related to brain injury. Nevertheless, this possible difficulty with interpreting test performances that may be below the norm at baseline probably needs consideration in the design of future studies.

Potentially related to the above point, it would be helpful to know a bit more about the cognitive performance of a specific subgroup of participants. How many of the 199 patients in the reported study who had never had ECT prior to baseline cognitive testing were there, and when analysing the data from this subgroup what were the findings? While the paper does report a large number of assessments, of which the highest number (122) were with patients who had never had previous ECT, the actual number of patients with no previous ECT was not entirely clear. Perhaps in future studies it would be possible to assess ECT-naive patients for pre-morbid general intellectual ability as a comparison standard to inform the interpretation of subsequent serial neuropsychological test performances. Finally, ECT-naive patients may also, at baseline testing, potentially be further away from floor-level normative data, which could, if showing no significant change over time, provide further evidence to build on the findings from the current landmark study. Such findings may further reassure patients, their families and clinicians when considering cognition and ECT.