ECT - Core

Journal of Affective Disorders 185 (2015) 67–75

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Research report

Baseline cognitive function does not predict the treatment outcome of electroconvulsive therapy (ECT) in late-life depression Tor Magne Bjølseth a,n, Knut Engedal b, Jūratė Šaltytė Benth c,d, Gro Strømnes Dybedal a, Torfinn Lødøen Gaarden a, Lars Tanum e a

Department of Geriatric Psychiatry, Diakonhjemmet Hospital, Pastor Fangens vei 18, 0854 Oslo, Norway Norwegian Centre for Aging and Health, Vestfold Health Trust, Tønsberg, Norway c Institute of Clinical Medicine, Campus Ahus, University of Oslo, Norway d HØKH, Research Centre, Akershus University Hospital, Norway e Department of Research and Development in Mental Health, Akershus University Hospital, Norway b

art ic l e i nf o

a b s t r a c t

Article history: Received 23 February 2015 Received in revised form 15 June 2015 Accepted 15 June 2015 Available online 23 June 2015

Background: No prior study has investigated whether impairment of specific cognitive functions at baseline may predict the short-term treatment outcome of electroconvulsive therapy (ECT) in elderly non-demented patients with major depression (MD). Methods: This longitudinal cohort study included 65 elderly patients with unipolar or bipolar MD, aged 60–85 years, treated with formula-based ECT. Treatment outcome was assessed using the 17-item Hamilton Rating Scale for Depression (HRSD17). Cognitive function at baseline was assessed using nine neuropsychological tests or subtests measuring information processing speed, verbal learning and memory, and aspects of executive function. Results: A poorer performance on the word reading task of the Color Word Interference Test rendered higher odds of achieving remission during the ECT course (p¼0.021). Remission was defined as an HRSD17 score of 7 or less. There were no other significant associations between the treatment outcome of ECT and cognitive performance parameters assessed at baseline. Limitations: The limited number of subjects may have reduced the generalizability of the findings. Multiple statistical tests increase the risk for making a type I error. Conclusions: How well patients perform on neuropsychological tests at baseline is most likely not a predictor of, or otherwise not significantly associated with the treatment outcome of formula-based ECT in elderly patients with MD. & 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Electroconvulsive therapy Treatment outcome predictors Baseline cognitive function Major depression Old age

1. Introduction Late-life depression (LLD) is associated with a high risk for disability, physical morbidity, and premature death (Erlangsen et al., 2006; Schulz et al., 2002), which also lead to high levels of burden on caregivers and health services (Luppa et al., 2013; Zivin et al., 2013). Antidepressant medication is the first choice of treatment for major depression (MD) among the elderly (American Psychiatric Association, 2010). However, about two thirds of the elderly patients included in a meta-analysis of 51 double-blind Abbreviations: BF, bifrontal; CWIT, Color Word Interference Test; ECT, electroconvulsive therapy; FAS, Letter Fluency Test; HRSD, Hamilton Rating Scale for Depression; HVLT-R, Hopkins Verbal Learning Test – Revised; LLD, late-life depression; MD, major depression; MMSE, Mini Mental State Examination; RCT, randomized controlled trial; RUL, right unilateral; TMT, Trail Making Test n Corresponding author. Tel.: þ 47 22 45 85 00; fax: þ 47 22 45 85 01. E-mail address: [email protected] (T.M. Bjølseth).

randomized controlled trials (RCTs) did not achieve remission after a single antidepressant trial, whereas 52% were non-responders (Kok et al., 2012). When strictly defined, remission is a more clinically relevant and valid end-point than response because it has stronger implications for good function and prognosis (Rush et al., 2006). In comparison, 52–55% of the patients treated with electroconvulsive therapy (ECT) did not meet the criteria for shortterm remission in two recently published studies (Bjølseth et al., 2014; Oudega et al., 2011). In these trials, formula-based ECT was administered to clinical samples of elderly patients, referred to inward treatment of MD. The proportion of elderly depressed patients who do not achieve remission is a topic of major concern in the field of geriatric psychiatry. One question that has been raised is whether cognitive impairment may predict a poorer treatment outcome. Research has focused primarily on to what extent core cognitive deficits in LLD, specifically slowed information processing and reduced executive functioning (Butters et al., 2004; Elderkin-

http://dx.doi.org/10.1016/j.jad.2015.06.021 0165-0327/& 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

68

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

Thompson et al., 2003; Dybedal et al., 2013) can predict non-response to antidepressant drugs (Pimontel et al., 2012). Similar patterns of results have emerged, indicating that slowed information processing and impairment in aspects of executive function are predictive of a poorer outcome of antidepressant treatment (McLennan and Mathias, 2010). However, methodological flaws such as small sample sizes, large numbers of statistical tests, inadequate adjustments for possible confounders, and outcome measures of questionable validity, give reason to doubt the generalizability of findings (Basso et al., 2013). In contrast to the numerous studies on predictors of response to antidepressants, little is known about the neurocognitive characteristics of elderly patients with MD who achieve remission after ECT and those who do not. Global cognitive function at baseline, as measured by the Mini Mental State Examination (MMSE: Engedal et al., 1988; Folstein et al., 1975), has not been shown to discriminate between remitters and non-remitters (Hausner et al., 2011; Oudega et al., 2011). Although being widely used as a test of cognition in ECT research, it provides only an overall score which cannot be localized to particular cognitive domains (Dunne and McLoughlin, 2012). Hence, it remains to be established whether specific cognitive functions are predictive of remission after ECT. In case neurocognitive characteristics of ECT non-remitters can be identified, such knowledge can lead to better tailored treatment, thus reducing the risk of exposing patients to cognitive side-effects to no avail. Historically, melancholic features and psychomotor retardation have been considered predictors of a good treatment outcome after ECT (Rasmussen, 2009). In contrast to older studies that corroborated this clinical notion (Hickie et al., 1990, 1996), Fink et al. (2007), in a large trial of patients treated with bitemporal ECT, showed that those with melancholic features had a significantly lower remission rate than those without them. Diverging findings may have resulted from changes in diagnostic terminology over the decades (Rasmussen, 2007). Hence, information processing speed may be positively associated with remission after ECT. Additionally, an observational study of magnetic resonance imaging and ECT treatment outcome (Oudega et al., 2011) showed that severely depressed elderly patients without medial temporal lobe atrophy had a three-fold greater chance of remitting, compared to patients with it. This finding implies that difficulties with verbal learning and memory, which are mediated by the temporal lobes, may be associated with poorer improvement after ECT. Corresponding with the aforementioned studies, we hypothesize that elderly depressed patients showing more impairment of information processing speed, verbal memory, and executive function before ECT are less likely to improve from depression after an acute course of ECT, compared to those performing better on such neuropsychological tests at baseline. To our knowledge, no study has previously investigated this relationship. To test our hypothesis, this study was designed to assess whether slower information processing, deficits in verbal learning and memory, and lower executive skills were predictive of, or were significantly associated with, a less favorable time trend for symptom severity when treated with formula-based ECT.

2. Methods 2.1. Study design This longitudinal cohort study originates from a single-site, assessor-blinded, controlled, parallel-group trial with balanced 1:1 randomization comparing the efficacy of formula-based bifrontal (BF) vs. right unilateral (RUL) ECT. Because the analyses did not demonstrate significant differences in efficacy between the

patients allocated to BF or RUL electrode placement, the results will be presented for the participants as one group, consisting of 65 patients. The study is registered at the online clinical trial database Clinicaltrials.gov (identifier: NCT01559324). The randomized phase of the study took place between September 1, 2009 and May 1, 2013, and the results of the RCT have recently been published (Bjølseth et al., 2014). 2.2. Study population Norwegian-speaking in-hospital patients aged 60–85 years with a major depressive episode as defined by the DSM-IV-TR criteria (American Psychiatric Association, 2000), either unipolar or bipolar, and referred for ECT were eligible for inclusion. Additional inclusion criteria were a minimum baseline score of 18 on the 17-item Hamilton Rating Scale for Depression (HRSD17; Hamilton, 1960; Williams, 1988) and the ability to co-operate in testing and to give voluntary informed consent. Exclusion criteria were an MMSE score of 23 (of 30) or less, schizophrenia or schizoaffective disorder, serious brain injury (e.g., dementia, head injury, or moderate- to large-size stroke), alcohol or substance abuse in the three weeks prior to ECT, medical conditions contradicting ECT, and having received ECT in the six months preceding the study. All patients were recruited from the Department of Geriatric Psychiatry at the Diakonhjemmet Hospital, a public hospital serving approximately 230,000 inhabitants of the western and central parts of Oslo, of which 28,000 were aged 65 years or older. 2.3. Treatment 2.3.1. Psychotropic medication at baseline Fifty-six (86.2%) patients had not responded to at least one adequate trial of an antidepressant prior to admission. For those individuals, antidepressants were reduced or withdrawn 3–10 days before starting ECT. Forty (61.5%) patients continued receiving antidepressants at baseline. Thirty-nine (60.0%) were in need of benzodiazepines for anxiety. The dosages of these drugs were converted to a defined daily dosage and are listed in Table 1. In 5 (7.7%) patients, anti-epileptic drugs or mood stabilizers were discontinued one week prior to ECT. Olanzapin (2.5–20 mg/day) or quetiapin (50–200 mg/day) were prescribed to 27 (41.5%) patients with psychotic symptoms or agitation. 2.3.2. ECT All treatments were administered after hyper-oxygenation for at least 1 min with 100% oxygen under mask anesthesia with atropine (0.5–1.0 mg), thiopental (2.5–3.5 mg/kg), and succinylcholine (0.7 mg/ kg), all of which were given intravenously. For ECT using the BF position, each electrode was placed 5 cm above the outer angle of the orbit on a line parallel to the sagittal plane (Letemendia et al., 1993). The d'Elia placement was used for RUL ECT (D’Elia, 1970). Treatment was administered twice a week using a Thymatron system IV (Somatics, LLC, Lake Bluff, Illinois, USA), which delivered a square-wave brief-pulse (0.5–1.0 ms pulse width) bidirectional current at 0.9 A and 10–70 Hz, depending on the stimulus intensity. The initial stimulus dose was determined using the age method (Abrams, 2002) in cases of unilateral stimulation and the half-age method (Petrides et al., 2009) in cases of bifrontal stimulation. The dose was adjusted for sex. Women were administered five energy percent (25 mC) less than the amount derived from the age or half-age formula, and men were administered five energy percent more. At subsequent sessions, the stimulus dosage was adjusted according to the duration, amplitude, and coherence of electroencephalogram delta waves (American Psychiatric Association, 2001), post-ictal suppression (Azuma et al., 2007), peak post-ictal heart rate (Swartz, 2002), time to recover orientation (Sobin et al., 1995), and

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

Table 1 Demographic and clinical characteristics of major depressed elderly patients treated with electroconvulsive therapy (ECT) (n¼ 65). Characteristics

Demographic characteristics Age, mean (SD), years Female, n (%) Years of education, mean (SD)

Clinical characteristics Unipolar depression, n (%) Bipolar depression, n (%) Psychotic symptoms, n (%) Melancholic symptoms, n (%) Co-morbid anxiety disorder, n (%) Previous alcohol abuse, n (%) Failed trials of Z 2 ADa s of different classes, n (%) History of ECT, n (%) First episode before the age of 60, n (%) Time since onset of first episode (median, range), months Duration of index episode (median, range), weeks Total CIRS-Gb score, mean (SD) DDDc of antidepressants at baseline (median, range) DDD of benzodiazepines at baseline (median, range)

Neurocognitive characteristics at baseline Global cognitive function MMSEd, mean (SD) Information processing speed TMT Ae, mean (SD), s CWITf 1, color naming, mean (SD), s CWIT 2, word reading, mean (SD), s Verbal learning and memory HVLT-Rg total learning, mean (SD) HVLT-R delayed recall, mean (SD) Executive function Tower Test, mean (SD) CWIT 3, inhibition, mean (SD), s Letter Fluency Test, mean (SD) Animal Naming Test, mean (SD)

Treatment outcome of ECT HRSD17 score at baseline, mean (SD) HRSD17 score after ECT, mean (SD) Decline in HRSD17 score, mean (SDh) No. of ECT sessions, mean (SD)

Elderly patients with major depression (n¼ 65)

74.9 (6.5) 36 (55.4) 13.5 (2.9)

58 7 15 62 16 7 27

(89.2) (10.8) (23.1) (95.4) (24.6) (10.8) (41.5)

21 (32.3) 35 (53.8) 192 (0–720) 24 (3–288) 6.9 (3.6) 0.5 (0–3.0) 0.8 (1–1.5)

27.6 (1.9) 65.8 (22.7) 42.9 (11.3) 28.0 (6.1) 20.7 (4.7) 6.8 (2.9) 12.5 104.5 35.6 16.7

(4.1) (36.0) (12.1) (5.1)

23.4 9.7 13.7 9.3

(4.5) (6.2) (7.2) (3.3)

a

AD, antidepressant. CIRS-G, Cumulative Illness Rating Scale. c DDD, daily defined dosage. d MMSE, Mini Mental State Examination. e TMT A, Trail Making Test A. f CWIT, Color Word Interference Test. g HVLT-R, Hopkins Verbal Learning Test – Revised. h SD not adjusted for intra-correlations due to repeated measurements. b

Clinical Global Impression Improvement score (Leon et al., 1993) on the following day, as assessed by the patient and the ward nurse. Treatments were continued until the patient achieved remission or until a plateau in the patient's benefit was reached up to a maximum of 16 sessions. 2.4. Assessment 2.4.1. Diagnostic procedures Using all available information from the in-ward observation,

69

the patient interviews, the next of kin, and the referring GP, the diagnosis of MD according to the DSM-IV-TR criteria was confirmed by consensus between two independent and experienced senior consultants in geriatric psychiatry. In addition, the diagnosis of MD was supported by the Mini-International Neuropsychiatric Interview, specifically the MINI-Plus (Mordal et al., 2010; Sheehan et al., 1998), which was also used when screening for psychiatric co-morbidity. Scoring on the Cumulative Illness Rating Scale for Geriatric Patients was performed, with the exception of the psychiatric item (Miller et al., 1992), to document the overall medical burden. 2.4.2. Evaluation of treatment outcome Two assessors, a study psychologist and a trained test assistant, conducted the assessments of depressive symptoms and cognitive function. The HRSD17 was used to assess the decline in symptom severity. The first assessment of symptom severity was carried out at baseline, on an average of four days before the first ECT treatment. Thereafter, patients were assessed every Wednesday between every second session and on average four days after the last ECT treatment. Inter-rater reliability between the study psychologist and the test assistant was tested and found to be high, with an intra-class correlation of 0.90 with the HRSD17. 2.4.3. Definition of treatment outcome and drop-outs The primary outcome was the longitudinal profile of continuous HRSD17 scores over the treatment course. Additionally, we used as a primary outcome, the development of the odds of not meeting the HRSD17 remission criterion at each visit with respect to baseline. Remission was defined as an HRSD17 score of 7 or less (Rush et al., 2006; Zimmerman et al., 2013). Patients who terminated the treatment course prematurely were considered to be drop-outs and were included in the analyses. 2.4.4. Neurocognitive assessment Global cognitive function was assessed using a revised version of the MMSE, and mild cognitive impairment was defined as an MMSE score of 27 or less (Strobel and Engedal, 2008). Information processing speed refers to the time used to complete simple psychomotor tasks. It was assessed using the Trail Making Test Part A (TMT A; Reitan and Wolfson, 1993) and the Delis Kaplan Executive Function System (D-KEFS) Color Word Interference Test, Parts 1 and 2 (CWIT; Delis et al., 2005). In the TMT A, the patient was asked to draw a line in the right order between circles enclosing numbers from 1 to 25. In the CWIT Part 1, the patients had to state the color of 50 small colored squares (red, blue, and green) as quickly as possible, whereas in the CWIT Part 2, words were read that denoted the same colors printed in black ink. Anterograde memory for verbal material, defined as the ability to form and store new information, was measured using the official Norwegian research version 3 of the Hopkins Verbal Learning Test – Revised (HVLT-R; Brandt and Benedict, 2001). In this test, a list of 12 words was presented orally to the patient. Results from total recall across three trials and free delayed recall after 20– 25 min were recorded. The umbrella concept of executive functions encompasses a variety of cognitive abilities, mediated primarily by the prefrontal cortex, that allow adaptive and goal-directed behavior, such as planning, organizing, self-monitoring, response inhibition, and strategy generation (Miyake et al., 2000). Aspects of executive function were assessed using the D-KEFS Tower Test, the CWIT Part 3, the Letter Fluency Test from the D-KEFS battery (FAS; Delis et al., 2005), and the Animal Naming Test (Borowski et al., 1967). The Tower Test was used to assess the ability to solve problems and plan by having the participant move five disks across three

70

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

pegs to build a specific tower in as few movements as possible. In CWIT Part 3, used to measure inhibition of an over-learned response, the participant was asked to state the color in which color words were printed. The FAS and the Animal Naming Test were used to measure the ability to organize thinking (Lezak et al., 2012). In the FAS, the participant was asked to say words that begin with F, A, and S within a time limit of 60 s for each letter, whereas in the Animal Naming Test, the participant was asked to say as many words as possible that belong to the category of animals within a time limit of 60 s. Patients were given maximum scores on the TMT A and the CWIT Part 3 if these tasks were not completed within their time limits of 95 and 180 s, respectively. These ceiling values were given to 14 (21.5%) patients for the TMT A and to 5 (7.7%) patients for the CWIT Part 3. 2.5. Ethics The study was approved by the Regional Committee for Research Ethics in Norway, including the permission to use ECT in both moderate and severe MD and as a first-line treatment in selected cases. In the week following admission, eligible patients and their next of kin were given thorough written and oral information about ECT, the purpose of the trial, and the procedures involved. Inclusion was based strictly upon informed consent with the patient's signature. 2.6. Statistical analyses Descriptive and outcome analyses were conducted using SPSS version 22.0 and SAS version 9.3, respectively. Results with p Values of 0.025 or less were considered statistically significant. All tests were two-sided. 2.6.1. Descriptive analyses Demographic and clinical characteristics were presented as means and standard deviations (SDs) for continuous symmetrically distributed data, whereas skewed data were presented as medians and ranges. Categorical variables were presented as frequencies and percentages. To explore whether neurocognitive function was associated with demographic and clinical characteristics, test scores were compared between males and females, between those who had and had not previously received ECT, and between those who were and were not in need of benzodiazepines at baseline. Comparisons of scores were also performed between patients who did and did not present with psychotic symptoms, previous alcohol abuse, early-onset depression, bipolar disorder, melancholic symptoms, anxiety disorder, and resistance to pharmacotherapy. These comparisons were performed using either an independent sample t test or the Mann–Whitney test, as appropriate. 2.6.2. Missing data The HRSD17 scores were imputed using inclusion values for seven patients that lacked baseline data. This was assumed to be valid because symptom severity did not likely change significantly during the five days (on average) that passed between inclusion and baseline assessment. Four patients did not complete either the TMT A (1), the CWIT Part 1 (2), the CWIT Part 2 (2), the Tower Test (2), the CWIT Part 3 (2), the Animal Naming Test (1), or the Letter Fluency Test (2), resulting in 12 (2.1%) missing neurocognitive values. Because cognitive performance was somewhat more impaired in psychotic patients, the missing values were imputed using the average score from the group that displayed psychotic symptoms or did not, as appropriate.

2.6.3. Outcome analyses Time-trend in continuous treatment outcome was assessed using a linear mixed model with fixed effects for first-, second-, and third-order time components and with random effects for intercepts. A logistic regression model for hierarchical data was estimated to assess time-trend in dichotomous treatment outcome. The model contained fixed effects for first- and second-order time components and random effects for intercepts. The nine time-trend models, one for each neurocognitive test used as a predictor, were estimated for each of the two outcomes. The models were further adjusted for four a priori chosen covariates, the presence of psychosis, symptom severity at baseline using the Montgomery and Aasberg Depression Rating Scale (MADRS; Montgomery and Aasberg, 1979), time since onset, and polarity. Each model was reduced by applying Akaike's Information Criterion, where a smaller value represents a better model. The adjustment variables were chosen on clinical grounds, and have been previously shown to correlate with either ECT treatment outcome or cognitive function. Finally, the reduced models were adjusted for the confounders, age and years of education. The interactions between two neuropsychological tests (HVLT-R total learning and the Tower Test) and psychosis were assessed when estimating the models, but were not significant and therefore excluded.

3. Results 3.1. Participant flow In the original RCT, a total of 79 patients were randomized, and six had no post-baseline assessments (Bjølseth et al., 2014). An additional eight patients were excluded from this observational study because of their inability to co-operate during neuropsychological tests as a result of psychosis (n ¼4) or because they had suffered serious brain injury (n ¼ 4). This yielded a sample of 65 patients. The inclusion of seven patients who prematurely terminated the ECT course did not introduce a bias to the analysis. Drop-outs, apart from being somewhat older than completers (80.1 vs. 74.3 years, p ¼0.023), did not perform significantly worse on any neuropsychological test. 3.2. Characteristics Table 1 lists demographic and clinical characteristics of the sample. Fifteen (23.1%) patients presented with psychotic symptoms, and 27 (41.5%) were resistant to pharmacotherapy in that they failed at least two trials of different-class antidepressants. Twenty-seven (41.5%) displayed mild cognitive impairment at baseline. A considerable improvement in cognition was observed in fifteen (55.6%) of these patients because they no longer met the criterion for mild cognitive impairment after treatment. When comparing neuropsychological test results across demographic and clinical categories, no significant group differences were found except in the categories of sex and presence of psychotic symptoms. Female patients had a lower mean score on the Tower Test (p¼ 0.024). Delusional patients performed worse than non-delusional patients on the HVLT-R total learning subtest (p ¼0.014) and the Tower Test (p ¼0.012). Hence, post-hoc analyses were performed to assess whether psychosis had a confounding effect on the associations between the results of these two tests and treatment outcome. 3.3. ECT treatment outcome The mean (SD) change in HRSD17 scores from baseline to the end of ECT was 13.7 (7.2) points (Table 1). The trajectories of the

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

71

Fig. 1. Trajectories of observed and estimated mean scores for the 17-item Hamilton Rating Scale for Depression (HRSD17) and odds ratios for not meeting the HRSD17 remission criterion during the electroconvulsive therapy course. The number of patients assessed at each time point was: baseline (65) and visits 1 (65), 2 (65), 3 (63), 4 (50), 5 (38), 6 (27), 7 (17), and 8 (6).

observed and estimated HRSD17 mean scores as well as the estimated odds ratios for not having achieved remission during treatment are shown in Fig. 1. According to the linear mixed model for HRSD17, there was a significant non-linear downward tendency during the ECT course (p o0.001). Similarly, there was a significant non-linear trend during the ECT course in the development of odds of not meeting the remission criterion, according to the logistic regression model for hierarchical data (po 0.01). In the study population, 37 (56.9%) patients did not achieve remission, whereas 28 (43.1%) did. 3.4. Treatment outcome of ECT in relation to specific neurocognitive functions We assessed the associations between performance on the neuropsychological tests at baseline and the longitudinal profiles of continuous and dichotomized HRSD17 scores over the ECT course using a multivariate regression analysis. Adjusting for covariates and confounders minimally altered the results for the time components of these profiles. Table 2 shows the crude and adjusted regression coefficients and the odds ratios for the effects of each test or subtest on the trajectories of estimated HRSD17 mean scores and the odds of not having achieved remission during treatment (Fig. 1). We found no significant association between performance on any of the neuropsychological tests and treatment outcome before adjusting for a priori chosen covariates (symptom severity, psychosis, time since onset, and polarity) and confounders (age and years of education). In the final model, however, higher scores on the CWIT Part 2 were significantly negatively associated with the longitudinal profile of the odds of not meeting the remission criterion, implying that patients with poorer performances on this subtest had higher odds of achieving remission during the ECT course. 3.5. Post-hoc analyses We explored whether the presence of psychosis had a confounding effect on associations between performance on two tests (HVLT-R total learning subtest and Tower Test) and treatment outcome because delusional patients displayed lower mean scores on those tests compared with non-delusional patients. There was

no significant interaction between the presence of psychosis and test performance in relation to the time-trend for HRSD17 mean scores or the development of the odds of not meeting the HRSD17 remission criterion. Thus, ECT treatment outcome was predicted by test performance in a similar way in patients with and without psychotic symptoms.

4. Discussion To the best of our knowledge, this is the first study investigating whether specific cognitive functions at baseline are predictive of the short-term treatment outcome of ECT. Previous studies have reported major difficulties in recruiting frail elderly depressed patients displaying psychomotor retardation and indecisiveness to take part in neuropsychological testing prior to ECT (Gardner and O’Conn, 2008; O’Connor et al., 2010; Stek et al., 2007). Thus, the small sample sizes of these studies did not make it possible to assess the predictive value of test results. Our data offered no convincing evidence for our hypothesis, given its power to support it. This result implicates that non-demented elderly patients who present with slower information processing and deficits in verbal memory and executive function are as likely to improve from depression after an acute course of ECT as those performing better on neuropsychological tests at baseline. When adjustments for possible confounders were made, a poorer performance on the CWIT Part 2 rendered higher odds of meeting the remission criterion during the ECT course. This finding must be interpreted with caution because there were no significant associations between the longitudinal profile of continuous HRSD17 scores and the results of this subtest or between treatment outcome and performance on the other tests, used to measure information processing speed. Moreover, a single statistically significant association can be coincidental, because using multiple statistical tests increases the risk for a type I error. It has been suggested that elderly depressed patients who receive ECT and take part in a controlled trial are clinically too homogenous to be included in a study that aims to examine whether neurocognitive performance can predict treatment outcome (Hickie et al., 1996). Such an explanation seems more unlikely in our study because we included patients with mild

72

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

Table 2 Analyses of associations between continuous and dichotomous treatment outcomes of the ECT course, and neuropsychological test results at baseline, in elderly patients with major depression (n ¼65). Models for continuous HRSD17 scores contain time components up to third-order. Models for dichotomized HRSD17 scores contain time components up to second-order. Test effect on continuous HRSD17 mean scores Crude effect a

Information processing speed TMT Ad CWITe1, color naming CWIT 2, word reading Verbal learning and memory HVLT-Rf total learning HVLT-R delayed recall Executive function Tower CWIT 3, inhibition Letter Fluency Animal Naming

Adjusted effect b

R.c. (95% CI)c

pc

0.372 0.682 0.356

0.02 (  0.02; 0.06)  0.04 (  0.13; 0.04)  0.09 (  0.24; 0.05)

0.413 0.337 0.204

0.418 0.164

0.05 (  0.17; 0.26)  0.05 (  0.38; 0.28)

0.656 0.770

0.950 0.367 0.649 0.177

0.02 0.01 0.03  0.12

(  0.22; (  0.02; (  0.04; (  0.32;

0.863 0.648 0.366 0.214

R.c. (95% CI)

p

0.02 (  0.03; 0.07)  0,02 (  0.12; 0.08)  0.08 (  0.26; 0.10)  0.09 (  0.33; 0.14)  0.27 (  0.64; 0.11) 0.01 0.01 0.02  0.15

(  0.28; (  0.02; (  0.07; (  0.36;

0.26) 0.04) 0.11) 0.07)

b

0.26) 0.03) 0.11) 0.07)

Test effect on odds ratios for not having met the HRSD17 remission criteriong Crude effect

Adjusted effect

h

OR (95% CI) Information processing speed TMT A CWIT 1, color naming CWIT 2, word reading Verbal learning and memory HVLT-R total learning HVLT-R delayed recall Executive function Tower CWIT 3, inhibition Letter Fluency Animal Naming

p

i

pj

OR (95% CI)

1.01 (0.98; 1.04) 0.98 (0.93; 1.03) 0.96 (0.88; 1.05)

0.433 0.425 0.369

1.01 (0.98; 1.03) 0.97 (0.93; 1.01) 0.92 (0.86; 0.99)

0.607 0.143 0.021

0.99 (0.88; 1.12) 0.89 (0.72; 1.09)

0.928 0.248

1.03 (0.93; 1.15) 0.94 (0.80; 1.11)

0.544 0.461

1.04 1.00 1.01 0.98

0.588 0.957 0.554 0.792

1.02 1.00 1.03 1.00

0.761 0.878 0.125 0.960

(0.90; (0.98; (0.97; (0.87;

1.20) 1.02) 1.07) 1.11)

(0.90; (0.99; (0.99; (0.90;

1.15) 1.01) 1.07) 1.11)

a

R.c., regression coefficient. p Value from a bivariate linear mixed model. c p Value from a multivariate linear mixed model (adjusted for test result, psychosis, symptom severity at baseline, the presence of unipolar and bipolar disorders, age and years of education). d TMT A, Trail Making Test A. e CWIT, Color Word Interference Test. f HVLT-R, Hopkins Verbal Learning Test – Revised. g Remission defined as an HRSD17 score of 7 or less. h OR, odds ratio. i p Value from a bivariate logistic regression model for hierarchical data. j p Value from a multivariate logistic model for hierarchical data (adjusted for test result, psychosis, symptom severity at baseline, time since onset, age and years of education). b

cognitive impairment and those referred to ECT as part of a regular clinical practice. The lack of association between poorer ECT treatment outcome and slower information processing and more deficits in verbal memory and executive function, compared to findings using antidepressants (Alexopoulos et al., 2005; Morimoto et al., 2011; Potter et al., 2004; Sneed et al., 2007; Story et al., 2008), can be explained by ECT working better than antidepressants in elderly patients with psychomotor retardation, memory problems, and impaired executive skills. Alternatively, it may be that the symptom severity in the aforementioned studies was less grave. Hence, test performance may have been hampered by inertia, inattentiveness, and lack of motivation to a lesser degree, thus yielding scores that more accurately described the optimal cognitive capacity of the patients, compared to the underachievement seen among some participants in our study group. At baseline, indications of mild cognitive impairment were found in 27 (41.5%) patients, whereas 15 (23.1%) were no longer categorized as impaired after the ECT course. Similarly, a misdiagnosis of

dementia has been reported in elderly patients with MD before they were successfully treated with ECT (Pier et al., 2012; Wagner et al., 2011). However, several studies have demonstrated that despite the improvement observed in some patients after treatment (Dybedal et al., 2015), cognitive functioning remained impaired at the group level in elderly depressed patients who presented with cognitive impairment at baseline (Butters et al., 2000; Sarapas et al., 2012). Consequently, changes in the intensity of depressive symptoms appear to have only a small influence on cognitive performance. Inclusion of patients in need of benzodiazepines between ECT sessions, with anxiety disorders and recent alcohol/substance abuse rendered our clinical sample representative of an elderly population normally referred for in-ward treatment for MD in Norway. Nevertheless, it is likely that such conditions may have contributed to a relatively low remission rate of 43.1%; remission rates of 67–90% have been reported in clinical trials that included more highly selected elderly patients (O’Connor et al., 2001; Tew

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

et al., 1999). In these studies, ECT was administered with stimulus titration, and remission was defined as an HRSD24 score of 10 or less among completers. Our relatively low remission rate can also be explained by the fact that we incorporated the whole sample, including drop-outs, who met strict remission criteria and that formula-based ECT provides a suboptimal stimulus dose for elderly patients with a high seizure threshold (McCall, 2009; Tiller and Ingram, 2006). Moreover, this study did not necessarily refute that the depression executive dysfunction model (Alexopoulos, 2001) is applicable to ECT because we did not map the extent of white matter abnormalities in our sample. The model predicts that depressed patients with impaired executive skills are less likely to respond to pharmacotherapy, whereas other types of cognitive function are largely non-related to treatment response. This theory is based on the premise that disruption of frontal–subcortical neural circuits contributes both to the development and perpetuation of LLD and to executive dysfunction. Neither did a meta-analysis of 17 antidepressant trials (McLennan and Mathias, 2010) lend strong support to the model. The authors emphasized that Type I error and the confounding effects of age and depression severity at baseline could possibly account for the observed associations between cognition and response to medication. Therefore, they questioned the role of neuropsychological tests at baseline in predicting antidepressant outcome in the clinical setting. We share the same doubt when it comes to ECT. Our findings indicate that cognitive performance at baseline does not offer a reliable basis for making clinical judgments about the likely treatment outcome of formulabased ECT.

5. Limitations A limited number of subjects may have reduced the generalizability of our findings. However, 65 of the 73 participants (89%) in our original sample took part in neuropsychological testing at baseline. This yielded a relatively large and representative study population for non-demented elderly patients with MD receiving ECT. Multiple comparisons may have resulted in spurious findings (type I error). However, we chose not to adjust p Values other than for primary outcomes for multiple testing because reducing the chance of incorrectly producing a difference on an individual test will increase the chance of falsely retaining the null hypothesis (type II error) for all tests. The use of HRSD17 as a primary outcome measure may have led to an under- or over-estimation of the efficacy of ECT in delusional patients. The sum of its item scores has recently been found to have limited validity for assessing the severity and improvement of psychotic depression (Ostergaard et al., 2014a) because only delusions and hallucinations with very specific themes can be rated. In future research, these shortcomings can be overcome by using a new composite rating scale, the Psychotic Depression Assessment Scale, which has demonstrated good content and clinical validity (Ostergaard et al., 2014b). The dual confounding effects of psychosis on both treatment outcome and neurocognitive predictors may have led to biased estimates of the associations of interest, because patients with psychotic symptoms had better odds of achieving remission (Bjølseth et al., 2014) and because delusional patients performed worse than non-delusional patients on tests measuring verbal learning (HVLT-R total learning subtest) and aspects of executive function (Tower Test). Thus, the negative association between performance on the CWIT Part 2 word reading task and the odds of not meeting the remission criterion may have been confounded by the interaction between psychosis and information processing

73

speed. However, we found no significant interaction between the presence of psychosis and neuropsychological test results in our models. Consequently, cognitive performance at baseline predicted treatment outcome in a similar way in delusional and nondelusional patients, implying that our estimates of associations between cognition and ECT treatment outcome were not biased. As previously discussed, it is unlikely that our primary finding was confounded by state effects. To minimize state and treatment effects on cognitive performance, Simpson et al. (1998) deferred assessment until six months after the ECT course when investigating the influence of cognitive functions on treatment outcome in LLD. Out of 75 patients aged 65–85 years, 43 (57%) were treated with ECT or lithium augmentation. The authors did not find significant differences in verbal learning, memory, or aspects of executive function between ECT/lithium remitters and non-remitters. Moreover, our aim was not to investigate the association between the cognitive capacity of optimally treated elderly patients with MD and ECT treatment outcome because neuropsychological test results at follow-up do not have predictive value.

6. Conclusion In the context of a relatively small sample of elderly non-demented patients with MD, the treatment outcome of an acute course of ECT was not predicted by, or otherwise significantly associated with, performance on baseline neuropsychological tests that measure information processing speed, verbal learning and memory, and aspects of executive function. One single statistically significant finding, indicating that poorer performance on the word reading task of the CWIT rendered higher odds of meeting the remission criterion during the ECT course, can most probably be attributed to Type I error. Hence, there is no reason to expect that non-demented elderly patients with impaired cognitive function are less likely to improve from depression after ECT, compared with those who show no cognitive impairment. Further research is warranted to determine whether ECT works better than antidepressants in elderly depressed patients who exhibit psychomotor retardation and deficits in memory and executive function, as our findings imply. Role of funding source This study was performed without external funding sources.

Conflict of interest No conflict declared.

Acknowledgments The authors thank chief psychiatrist B. Lorentzen at the Department of Geriatric Psychiatry, Diakonhjemmet Hospital, for his generous support. Senior anesthesiologist S.O. Mollestad contributed with his skills and calm, thus reassuring anxious and severely depressed patients in the minutes before sleep induction. We thank M. Gulliksrud, W.J. Lindberg, P. Mikkelsen, A.L. Røstad and, in particular, study nurse M. Larsen for data collection, quality control and their engagement in the study. Statistician F. Dahl was helpful in performing the computed randomization.

References Abrams, R., 2002. Electroconvulsive Therapy, fourth ed. Oxford University Press, New York. Alexopoulos, G.S., 2001. The depression-executive dysfunction syndrome of late life. Am. J. Geriatr. Psychiatry 9, 22–29. Alexopoulos, G.S., Kiosses, D.N., Heo, M., Murphy, C.F., Shanmugham, B., GunningDixon, F., 2005. Executive function and the course of geriatric depression. Biol.

74

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

Psychiatry, 58; , pp. 204–210. American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Disorders. Text Revision (DSM-IV-TR), fourth ed. American Psychiatric Association, Washington. American Psychiatric Association, 2001. Practice of Electroconvulsive Therapy: Recommendations of Treatment, Training and Privileging: A Task Force Report of the American Psychiatric Association, second ed. American Psychiatric Association, Washington. American Psychiatric Association, 2010. Practice Guideline: Treatment of Patients With Major Depressive Disorder, third ed.. American Psychiatric Association, Washington. Azuma, H., Fujita, A., Sato, K., Arahata, K., Otsuki, K., Hori, M., Mochida, Y., Uchida, M., Yamada, T., Akechi, T., Furukawa, T.A., 2007. Postictal suppression correlates with therapeutic efficacy for depression in bilateral sine and pulse wave electroconvulsive therapy. Psychiatry Clin. Neurosci. 61, 168–173. Basso, M.R., Miller, A., Estevis, E., Combs, D., 2013. Neuropsychological deficits in major depressive disorder: correlates and conundrums of cognitive functioning. In: Arnett, P.A. (Ed.), Secondary Influences on Neuropsychological Test Performance. Oxford University Press, New York, pp. 39–66. Bjølseth, T.M., Engedal, K., Benth, J.Š., Dybedal, G.S., Gaarden, T.L., Tanum, L., 2014. Clinical efficacy of formula-based bifrontal versus right unilateral electroconvulsive therapy (ECT) in the treatment of major depression among elderly patients: a pragmatic, randomized, assessor-blinded, controlled trial. J. Affect. Disord. 175, 8–17. Borowski, J.G., Benton, A.L., Spreen, O., 1967. Word fluency and brain damage. Neuropsychologia 5, 135–140. Brandt, J., Benedict, R.H.B., 2001. The Hopkins Verbal Learning Test – Revised. Professional Manual. Psychological Assessment Resources Inc., Odessa, FL. Butters, M.A., Becker, J.T., Nebes, R.D., Zmuda, M.D., Mulsant, B.H., Pollock, B.G., Reynolds 3rd, C.F., 2000. Changes in cognitive functioning following treatment of late-life depression. Am. J. Psychiatry 157, 1949–1954. Butters, M.A., Whyte, E.M., Nebes, R.D., Begley, A.E., Dew, M.A., Mulsant, B.H., Zmuda, M.D., Bhalla, R., Meltzer, C.C., Pollock, B.G., Reynolds 3rd, C.F., Becker, J. T., 2004. The nature and determinants of neuropsychological functioning in late-life-depression. Arch. Gen. Psychiatry 61, 587–595. D’Elia, G., 1970. Unilateral electroconvulsive therapy. Acta Psychiatr. Scand. Suppl. 215, 1–8. Delis, D.C., Kaplan, E., Kramer, J.H., 2005. Delis-Kaplan Executive Function System (D-KEFS), Norwegian Manual. Pearson Assessment, Stockholm. Dunne, R.A., McLoughlin, D.M., 2012. Systematic review and meta-analysis of bifrontal electroconvulsive therapy versus bilateral and unilateral electroconvulsive therapy in depression. World J. Biol. Psychiatry 13, 248–258. Dybedal, G.S.D., Tanum, L., Sundet, K., Gaarden, T.L., Bjølseth, T.M., 2013. Neuropsychological functioning in late-life depression. Front. Psychol. 4, 381. Dybedal, G.S.D., Tanum, L., Sundet, K., Bjølseth, T.M., 2015. The role of baseline cognitive function in the neurocognitive effects of electroconvulsive therapy in depressed elderly patients. Clin. Neuropsychol., doi.org/10.1080/ 13854046.2015.1050457. Elderkin-Thompson, V., Kumar, A., Bilker, W.B., Dunkin, J.J., Mintz, J., Moberg, P.J., Mesholam, R.I., Gur, R.E., 2003. Neuropsychological deficits among patients with late-onset minorand major depression. Arch. Clin. Neuropsychol. 18, 529–549. Engedal, K., Haugen, P.K., Gilje, K., Laake, P., 1988. Efficacy of short mental tests in the detection of mental impairment in old age. Compr. Gerontol. A 2, 87–93. Erlangsen, A., Zarit, S.H., Tu, X., Conwell, Y., 2006. Suicide among older psychiatric Inpatients: an evidence-based study of a high-risk group. Am. J. Geriatr. Psychiatry 14, 734–741. Fink, M., Rush, A.J., Knapp, R., Rasmussen, K., Mueller, M., Rummans, T.A., O’Connor, K., Husain, M., Biggs, M., Bailine, S., Kellner, C.H., 2007. DSM melancholic features are unreliable predictors of ECT response: a CORE publication. J. ECT 23, 139–146. Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12, 189–198. Gardner, B.K., O’Connor, D.W., 2008. A review of the cognitive effects of electroconvulsive therapy in older adults. J. ECT 24, 68–80. Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry 23, 56–62. Hausner, L., Damian, M., Sartorius, A., Frölich, L., 2011. Efficacy and cognitive side effects of electroconvulsive therapy in depressed elderly inpatients with coexisting mild cognitive impairment or dementia. J. Clin. Psychiatry 72, 91–97. Hickie, I., Parsonage, B., Parker, G., 1990. Prediction of response to electroconvulsive therapy: preliminary validation of a sign-based typology of depression. Br. J. Psychiatry 157, 65–71. Hickie, I., Mason, C., Parker, G., Brodaty, H., 1996. Prediction of ECT response: validation of a refined sign-based (CORE) system for defining melancholia. Br. J. Psychiatry 169, 68–74. Kok, R.M., Nolen, W.A., Heeren, T.J., 2012. Efficacy of treatment in older depressed patients: a systematic review and meta-analysis of double-blind randomized controlled trials with antidepressants. J. Affect. Disord. 141, 103–115. Leon, A.C., Shear, M.K., Klerman, G.L., Portera, L., Rosenbaum, J.F., Goldenberg, I., 1993. A comparison of symptom determinants of patient and clinical global ratings in patients with panic disorder and depression. J. Clin. Psychopharmacol. 13, 327–331. Letemendia, F.J.J., Delva, N.J., Rodenburg, M., Lawson, J.S., Inglis, J., Waldron, J.J., Lywood, D.W., 1993. Therapeutic advantage of bifrontal electrode placement in

ECT. Psychol. Med. 23, 349–360. Lezak, M.D., Howieson, D.B., Bigler, E.D., Tranel, D., 2012. Neuropsychological Assessment, fifth ed. Oxford University Press, Oxford. Luppa, M., König, H., Heider, D., Leicht, H., Motzek, T., Schomerus, G., Riedel-Heller, S.G., 2013. Direct costs associated with depressive symptoms in late life: a 4.5 year prospective study. Int. Psychogeriatr. 25, 292–302. McCall, W.V., 2009. Stimulus dosing. In: Swartz, C.M. (Ed.), Electroconvulsive and Neuromodulation Therapies. Cambridge University Press, New York, pp. 447–467. McLennan, S.N., Mathias, J.L., 2010. The depression-executive dysfunction (DED) syndrome and response to antidepressants: a meta-analytic review. Int. J. Geriatr. Psychiatry 25, 933–944. Miller, M.D., Paradis, C.F., Houck, P.R., Mazumdar, S., Stack, J.A., Rifai, A.H., Mulsant, B., Reynolds, C.F., 1992. Rating chronic illness burden in geropsychiatric practice and research: application of the cumulative illness rating scale. Psychiatry Res. 41, 237–248. Miyake, A., Friedman, N.P., Emerson, M.J., Witzki, A.H., Howerter, A., Wager, T.D., 2000. The unity and diversity of executive functions and their contributions to complex frontal lobe tasks: a latent variable analysis. Cogn. Psychol. 41, 49–100. Montgomery, S., Aasberg, M., 1979. A new depression rating scale designed to be sensitive to change. Br. J. Psychiatry 134, 382–389. Mordal, J., Gundersen, Ø., Bramness, J.G., 2010. Norwegian version of the MINI-International Neuropsychiatric Interview: feasibility, acceptability and test–retest reliability. Eur. Psychiatry 25, 172–177. Morimoto, S.S., Gunning, F.M., Murphy, C.F., Kanellopoulos, D., Kelly, R.E., Alexopolous, G.S., 2011. Executive function and short-term remission of geriatric depression: the role of semantic strategy. Am. J. Geriatr. Psychiatry 19, 115–122. O’Connor, D.W., Gardner, B., Eppingstall, B., Tofler, D., 2010. Cognition in elderly patients receiving unilateral and bilateral electroconvulsive therapy: a prospective, naturalistic comparison. J. Affect. Disord. 124, 235–240. O’Connor, M.K., Knapp, R., Husain, M., Rummans, T.A., Smith, G., Mueller, M., Snyder, K., Bernstein, H., Rush, A.J., Fink, M., Kellner, C., 2001. The influence of age on the response of major depression to electroconvulsive therapy: a C.O.R.E. report. Am. J. Geriatr. Psychiatry 9, 382–390. Ostergaard, S.D., Meyers, B.S., Flint, A.J., Mulsant, B.H., Whyte, E.M., Ulbricht, C.M., Bech, P., Rothschild, A.J., 2014a. Measuring psychotic depression. Acta Psychiatry Scand. 129, 211–220. Ostergaard, S.D., Meyers, B.S., Flint, A.J., Mulsant, B.H., Whyte, E.M., Ulbricht, C.M., Bech, P., Rotschild, A.J., 2014b. Measuring treatment response in psychotic depression: the Psychotic Depression Assessment Scale (PDAS) takes both depressive and psychotic symptoms into account. J. Affect. Disord. 160, 68–73. Oudega, M.L., van Exel, E., Wattjes, M.P., Comijs, H.C., Scheltens, P., Barkhof, F., Eikelenboom, P., de Craen, A.J.M., Beekman, A.T.F., Stek, M.L., 2011. White matter hyperintensities, medial lobe atrophy, cortical atrophy, and response to electroconvulsive therapy in severely depressed elderly patients. J. Clin. Psychiatry 72, 104–112. Petrides, G., Braga, R.J., Fink, M., Mueller, M., Knapp, R., Husain, M., Rummans, T., Bailine, S., Malur, C., O’Connor, K., Kellner, C., 2009. Seizure threshold in a large sample. J. ECT 25, 232–237. Pier, K.S., Briggs, M.C., Pasculi, R.M., Kellner, C.H., 2012. Successful electroconvulsive therapy for major depression misdiagnosed as Alzheimer dementia. Am. J. Geriatr. Psychiatry 20, 909–910. Pimontel, M.A., Culang-Reinlieb, M.E., Morimoto, S.S., Sneed, J.R., 2012. Executive dysfunction and treatment response in late-life depression. Int. J. Geriatr. Psychiatry 27, 893–899. Potter, G.G., Kittinger, J.D., Wagner, H.R., Steffens, D.S., Krishnan, K.R.R., 2004. Prefrontal neuropsychological predictors of treatment remission in late-life depression. Neuropsychopharmacology 29, 2266–2271. Rasmussen, K.G., 2007. Attempts to validate melancholic depression: some observations on modern research methodology. Bull. Menn. Clin. 71, 150–163. Rasmussen, K.G., 2009. Evidence for electroconvulsive therapy efficacy in mood disorders. In: Swartz, C.M. (Ed.), Electroconvulsive and Neuromodulation Therapies. Cambridge University Press, New York, pp. 109–123. Reitan, R.M., Wolfson, D., 1993. The Halstead-Reitan Neuropsychological Test Battery: Theory and Clinical Interpretation, second ed. Neuropsychology Press, Tucson, AZ. Rush, A.J., Kraemer, H.C., Sackeim, H.A., Fava, M., Trivedi, M.H., Frank, E., Ninan, P.T., Thase, M.E., Gelenberg, A.J., Kupfer, D.J., Regier, D.A., Rosenbaum, J.F., Ray, O., Schatzberg, A.F., 2006. Report by the ACNP task force on response and remission in major depressive disorder. Neuropsychopharmacology 31, 1841–1853. Sarapas, C., Shankman, S.A., Harrow, M., Goldberg, J.F., 2012. Parsing trait and state effects of depression severity on neurocognition: evidence from a 26-year longitudinal study. J. Abnorm. Psychol. 121, 830–837. Schulz, R., Drayer, R., Rollman, B., 2002. Depression as risk factor for non-suicide mortality in the elderly. Biol. Psychiatry 52, 205–225. Sheehan, D.V., Lecrubier, Y., Sheehan, K.H., Amorim, P., Janavs, J., Weiller, E., Hergueta, T., Baker, R., Dunbar, G.C., 1998. The MINI-International Neuropsychiatric Interview (MINI); the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J. Clin. Psychiatry 59, 22–33. Simpson, S., Baldwin, R.C., Jackson, A.J., Burns, A.S., 1998. Is subcortical disease associated with a poor response to antidepressants? Neurological, neuropsychological and neuroradiological findings in late-life depression. Psychol. Med. 28, 1015–1026. Sneed, J., Roose, S.P., Keilp, J.G., Krishan, K.R.R., Alexopoulos, G.S., Sackeim, H.A., 2007. Response inhibition predicts poor antidepressant treatment response in very old depressed patients. Am. J. Geriatr. Psychiatry 15, 553–563.

T.M. Bjølseth et al. / Journal of Affective Disorders 185 (2015) 67–75

Sobin, C., Sackeim, H.A., Prudic, J., Devanand, D.P., Moody, B.J., McElhiney, M.C., 1995. Predictors of retrograde amnesia following ECT. Am. J. Psychiatry 152 995-001. Stek, M.L., van der Wurff, F.B., Iutdehaag, B.M.J., Beekman, A.T.F., Hoogendijk, W.J.G., 2007. ECT in the treatment of depressed elderly: lessons from a terminated clinical trial. Int. J. Geriatr. Psychiatry 22, 1052–1054. Story, T.J., Potter, G.G., Attix, D.K., Welsh-Bohmer, K.A., Steffens, D.C., 2008. Neurocognitive correlates of response to treatment in late-life depression. Am. J. Geriatr. Psychiatry 16, 752–759. Strobel, C., Engedal, K., 2008. MMSE-NR. Norwegian Revised Mini Mental Status Examination. Revised and Expanded Manual. Norwegian Centre for Aging and Health, Oslo. Swartz, C.M., 2002. ECT dosing by the benchmark method. German J. Psychiatry 5, 1–4. Tew, J.D., Benoit, B.A., Mulsant, H., Haskett, R.F., Prudic, J., Thase, M.E., Crowe, R.R., Dolata, D., Begley, A.E., Reynolds III, C.F., Sackeim, H.A., 1999. Acute efficacy of ECT in the treatment of major depression in the old–old. Am. J. Psychiatry 156,

75

1865–1870. Tiller, J.W., Ingram, N., 2006. Seizure threshold determination for electroconvulsive therapy: stimulus dose titration versus age-based estimations. Aust. N. Z. J. Psychiatry 40, 188–192. Wagner, G.S., McClintock, S.M., Rosenquist, P.B., McCall, W.V., 2011. Major depressive disorder with psychotic features may lead to misdiagnosis of dementia; a case report and review of the litterature. J. Psychiatr. Pract. 17, 432–438. Williams, J.B.W., 1988. A structured interview guide for the Hamilton Depression Rating Scale –. Arch. Gen. Psychiatry 45, 742–747. Zimmerman, M., Martinez, J.H., Young, D., Chelminski, I., Dalrymple, K., 2013. Severity classification on the Hamilton Depression Rating Scale. J. Affect. Disord. 150, 384–388. Zivin, K., Wharton, T., Rostant, O., 2013. The Economic, public health and caregiver burden of late-life depression. Psychiatr. Clin. North Am. 36, 631–649.