Non-memory cognition

In the subacute period following ECT (0–3 days), tests of executive function and processing speed were impaired compared with baseline. However, tests of attention were not impaired at this time point.^{Reference Semkovska and McLoughlin2} When measured between 4 and 14 days most tests had improved significantly compared with baseline and none were significantly below baseline. The improvement is likely to have been a result of successful treatment of the depressive illness as studies suggest that in very severe depression, a high percentage of patients have very significant cognitive impairment.^{Reference Reppermund, Zihl, Lucae, Horstmann, Kloiber and Holsboer4–Reference Douglas, Gallagher, Robinson, Carter, McIntosh and Frampton6}

Longer term, with a wide range of follow-up intervals (2 weeks to 2 years), there was no evidence of impairment in tests of non-memory cognition.^{Reference Semkovska and McLoughlin2} The only caveat to this may involve the use of sine wave ECT machines (discussed below).

Anterograde memory

Individual studies show that there was impairment on tests of anterograde memory from early in a course of ECT^{Reference Porter, Heenan and Reeves7} and meta-analyses suggest that this was still impaired subacutely (0–3 days) but returned to baseline (once again a baseline that may have been adversely affected by depression) between 4 and 14 days after the end of the course of ECT.^{Reference Semkovska and McLoughlin2} The degree of initial impairment depends on treatment and patient factors considered below. Once again, the data refers to mean values and compares performance with a baseline, which is likely to be reduced secondary to the cognitive effects of severe depression itself.^{Reference Douglas, Porter, Knight and Maruff5}

Beyond 14 days, there was no evidence of objectively assessed impairment.^{Reference Semkovska and McLoughlin2}

A caveat, which applies to all domains of cognitive function, is that although studies reporting the results of objective cognitive tests present mean results, the finding of no mean change from baseline does not necessarily mean that no patients experienced a clinically significant reduction in function. If, for instance, some patients improve greatly from baseline, a lack of mean change implies that some patients deteriorate. This is important to bear in mind when explaining risk on a case by case basis in clinical practice. Studies have rarely provided analysis on an individual level. We recommend that future studies provide this type of analysis in order for clinicians to be able to give information to patients regarding the percentage risk of significant decline.

Retrograde memory

When patients report cognitive side-effects it is often loss of autobiographical memory (i.e. loss of memory for events that the person has experienced) that are reported.^{Reference Rose, Wykes, Leese, Bindman and Fleischmann8} However, both episodic memory (memory for experiences) and semantic memory (memory for facts) are affected. Retrograde amnesia for non-personal information has also been reported.^{Reference Lisanby, Maddox, Prudic, Devanand and Sackeim9} Loss of memory in both areas may be distressing and functionally impairing. Objective testing of the loss of autobiographical memory (personal memories – including both episodic and semantic), usually using the Colombia University Autobiographical Memory Interview (CUAMI), has shown that the degree of loss differs depending on treatment parameters and that it exceeds the normal decay of memories over time in healthy matched participants. Furthermore, in studies where it is objectively measured for at least 1 year^{Reference Sackeim10} this loss has been found to persist. Whether it continues for longer than 1 year is not clear from objective studies, but some patients report that it does and it seems unlikely that lost memories could return spontaneously after this point. Although studies report loss of autobiographical memory as a mean, compared with baseline, identification of individuals who experience particularly severe loss has rarely been done. Where this has been done, there is evidence that females and those receiving bilateral ECT are more at risk.^{Reference Sackeim, Prudic, Fuller, Keilp, Lavori and Olfson11}

Subjective impairment

Studies of subjective reports of cognitive side-effects show varying percentages of patients who complain of ‘memory loss’, depending on a range of factors, including who interviewed them (Rose et al^{Reference Rose, Wykes, Leese, Bindman and Fleischmann8}). A recent review suggests that 60% of patients report memory problems, with 40% reporting that these lasted from several weeks to several years.^{Reference Chakrabarti, Grover and Rajagopal12} However, most of the patients involved in these studies received bilateral ECT, limiting generalisation to other forms of ECT. Subjective assessment of cognitive problems following ECT, usually assessed using detailed questionnaires, such as the Cognitive Failures Questionnaire,^{Reference Broadbent, Cooper, FitzGerald and Parkes13} correlates poorly with objective deficits. Subjective impairment often reduces throughout the course of treatment – tracking closely with clinical response. However, studies that used a simple question or seven-point Likert scale asking patients whether ECT ‘helped’ or ‘hurt’ their memory showed that scores on this scale declined (i.e. ECT ‘hurt’ memory on average) and this, in contrast with more complex assessment, correlated with loss of autobiographical memory.^{Reference Berman, Prudic, Brakemeier, Olfson and Sackeim14,Reference Brakemeier, Berman, Prudic, Zwillenberg and Sackeim15,Reference Semkovska, Landau, Dunne, Kolshus, Kavanagh and Jelovac16}

Post-ictal disorientation

Patients may be disoriented for a variable period of time following an individual treatment. In the extreme case, this disorientation may persist in a fluctuating way for hours or days following a treatment – a post-ictal delirium. Sobin et al^{Reference Sobin, Sackeim, Prudic, Devanand, Moody and McElhiney24} showed a correlation between time to orientation post-treatment and loss of autobiographical memory at 2 months after the end of the course. The correlation was driven by greater retrograde amnesia in a small group of patients who had recovery of orientation more than 30 min after the treatment. With most modern treatments it is unusual for this to happen, but recording time to re-orientation and reporting to clinicians when this is greater than, for example, 30 min is very simple and we recommend that this is done. More recently a small study confirmed the predictive effect of re-orientation across both RUL and bilateral ECT.^{Reference Martin, Gálvez and Loo32} A further and more rigorous development of this is the use of a specific questionnaire which, administered at 30 min post-ECT, in a small study correlated with loss of autobiographical memory at the end of the course.^{Reference Martin, Wong, Kumar and Loo33} Staff and families monitoring patients between treatments should be aware of the symptoms of delirium.

Global cognitive scales

Many units use scales measuring global cognitive performance as a means of monitoring. The 3MS has been shown to be sensitive to differences short term (4–14 days) between 2 × weekly and 3 × weekly and between RUL and bilateral electrode placement, and is able to detect the detrimental effects of ECT in the subacute period (0–4 days).^{Reference Semkovska and McLoughlin2} On this basis, it is potentially useful for monitoring. Other similar scales may also be used on the same basis if they are more familiar to clinicians or more available – for example the Montreal Cognitive Assessment^{Reference Nasreddine, Phillips, Bedirian, Charbonneau, Whitehead and Collin34} or the Addenbrooke's Cognitive Examination.

Brief general cognitive batteries

Several previous papers have recommended shorter, more clinically practical batteries of cognitive tests that could be delivered at intervals throughout a course. Porter et al^{Reference Porter, Douglas and Knight35} recommended a battery based on tests that had, in research studies, shown significant differences between different types of ECT (RUL versus bilateral) on the basis that these were therefore sensitive to aspects of ECT that could be modified if necessary. Data on the utility of this battery in clinical practice are awaited. An attempt has also been made to construct extremely short (few minutes) batteries of tasks, changes in which may be predictive of later cognitive difficulties.^{Reference Martin, Katalinic, Ingram, Schweitzer, Smith and Hadzi-Pavlovic36} However, although correlations between changes in this brief cognitive battery and tests of anterograde memory post-ECT, there was no evidence that this was predictive of long-term loss of autobiographical memory. Standardised, computerised batteries of cognitive tests may be a method with future utility.

Anterograde amnesia

One of the most consistent findings, at least in the short term, is impairment of new learning, shown by reduced scores in verbal learning. This has often been demonstrated using the traditional Rey Verbal Learning Task^{Reference Rey37} or the shorter but less sensitive Hopkins Verbal Learning Test.^{Reference Brandt38}

Retrograde amnesia

The planned nature of ECT allows an assessment of stored memories, prior to the start of a course, in order to be able to monitor what has been lost. Usually, amnesia for autobiographic memory is measured. The most systematic approach to this is to elicit a large number of memories, record these and then prompt patients to retrieve the memories later in the course of treatment. This has been done most commonly and systematically using the CUAMI, a scale that was carefully constructed to illicit positive, negative and neutral memories and recent and remote events. The same questions are then asked at intervals throughout a course of ECT to generate a score reflecting the amount of information forgotten. This has been compared with the natural loss of memories over time in healthy participants.^{Reference Sackeim10} It is sensitive to differences in different methods of ECT, with the amount of information being forgotten at 6 months displaying a gradient from healthy participants at one end of the gradient to those having bilateral ECT at the other end, with unilateral ECT in between. The scale is, however, lengthy (281 questions over 1–3.5 h) making it impractical for use in clinical practice.

A shorter form was therefore developed (CUAMI- SF) that utilises only 30 questions, focusing on the previous 1 year. Areas covered are shown in Appendix 2 and illustrate the areas in which questioning has been able to elicit loss of memory. These items were chosen from the larger CUAMI because they produced high and equivalent rates of production of memories at baseline in patients with depression and controls, and because they differentiated RUL from bilateral ECT. Both the CUAMI and its short form have been criticised because they have not undergone the same rigorous standardisation as traditional neuropsychological tests.^{Reference Semkovska and McLoughlin39} However, they represent a specific attempt to perform a particular function, which is to measure loss of autobiographical memory, in research into ECT and, in the case of the CUAMI-SF, to attempt to produce a scale that can be used clinically to measure autobiographical memory loss during and after ECT. To date, no better alternative has been developed.

Possible further criticisms may be that some of the items asked are poorly discriminative in certain people or populations with different customs or lifestyles and that the test may not therefore build up a picture of important memories that are likely to be lost. For example, in our practice, questions about what the patient did at New Year do not elicit a lot of variable information in the elderly people who have depression, who may easily be able to remember that they went to bed, as they do each year.

Subjective impairment

Although more complex questionnaires assessing subjective memory tend to correlate better with mood than objective impairment, simpler questions regarding whether ECT has ‘helped or hurt’ memory correlates well with objective measurement of memory impairment.^{Reference Semkovska, Keane, Babalola and McLoughlin3,Reference Brakemeier, Berman, Prudic, Zwillenberg and Sackeim15} Therefore, this simple question or a simple Likert scale may be very helpful in alerting clinicians to the onset of memory problems.

Individualising questions about autobiographical memory risks losing standardisation. This means that information regarding whether a patient has reached a threshold of memory loss that is likely to become clinically concerning is not available. However, where resources are scarce, a practical alternative to the CUAMI-SF is for clinicians to discuss with patients and their relatives what they have experienced in the previous 1 year and elicit and record some details that the clinician can attempt to elicit again at intervals during a course of ECT. This might involve relevant items from the CUAMI-SF and other items specific to the patient. Examples might include: have they attended a wedding? If so, record details of who was getting married, who attended, where it was held. Or have they attended a sporting event, and if so, who was playing. Suitable events could be suggested by family or friends.

Recent data suggests that although many patients have a subjective sense of memory problems following ECT, when patients perceptions of memory problems before the course of ECT are taken into account, a much lower percentage have a change in this.^{Reference Sigström, Nordenskjöld, Juréus, Clements, Joas and Pålsson40} Therefore it is helpful to ask about and record perceived memory gaps both before and after treatment.

Family and friends are frequently asked about memory performance of patients receiving ECT but it is important to be specific about what is meant by ‘memory’ and what specific aspects are being enquired about.