Design and setting
The study was conducted at the Department of Psychiatry of the Ersta Hospital psychiatric clinic in Stockholm, Sweden. The clinic specialized in treating health care professionals, and all patients were working in various healthcare units in the larger Stockholm catchment area. Patients were referred by their general practitioners or were self-referred. Typically, a diagnosis was formulated after three sessions with a senior psychiatrist, an M.D. together with a psychiatric nurse. Patients diagnosed with SED received written information about the present study and were invited to participate. Due to unforeseen external factors, data collection had to be terminated earlier than planned, limiting the number of participants.
In total, 49 subjects with SED as the primary diagnosis participated in the study. After controlling anamnestic data, nine subjects with a history of cerebral trauma or illness, a history of psychosis, bipolar disorder, or substance abuse were excluded. One additional subject was excluded due to a technical error in the registered data. Thus, the study population comprised of 39 subjects (35 females, four males). The mean age was 46,8 years (SD 10.1; range 30–60 yrs.). All participants were healthcare professionals, with a college or university degree, doctors, registered nurses, and psychologists. Diagnostic criteria are presented in Table 1.
Regarding treatment with psychotropic drugs, specific data for the included subjects were lacking. However, they were all participating in the clinic’s SED program. In this program, 33% of the subjects (n = 153) were treated with SSRI, 11% with SNRI, 29% with anxiolytics, and 33% were treated with hypnotics.
Before starting the neuropsychological assessment, the participants were asked to rate their symptoms on a VAS scale, ranging from 1 -to -10, of their a) current level of exhaustion b) perceived level of exhaustion at its highest level.
A psychologist provided detailed information about the test session. The session started with a short anamnestic interview regarding possible earlier brain injury or disease. The clinical neuropsychological tests were administered in the following fixed order: Coding, MapCog Spectra, Ruff 2&7, Symbol search, Digit Span, and Matrix Reasoning (see description below). They were all conducted in one session with an average duration of 75 min.
The MapCog Spectra (MCS) standardized measurement procedure was performed on a tablet (iPad), and commenced as follows: First, a short training session was conducted wherein the subject was asked to name four different colour and shape combinations (e.g., red heart, black ball, blue cube, and yellow star) displayed on the tablet. As the methodology is based on sound recordings detecting pause time durations between verbal responses, the subjects were instructed to avoid extraneous sounds (e.g., coughing, laughing, etc.). After a successful training session, a grid of 40 (5 × 8) randomly ordered colour-shape combinations appeared on the screen (i.e., the same colours and shapes as in the training session, but re-randomized in different combinations). The subjects were asked to name the stimulus combinations beginning from top left to the bottom right of the screen and were also instructed to name the colour first and then the shape. After each one of the 40 combinations was named, a second trial was performed. The time between the two trials was less than 30 s. The stimulus combinations were automatically re-randomized between the two trials to avoid learning effects. After completing the second trial, the results automatically appeared on the screen in a graph (Fig. 1).
The graph in Fig. 1 shows 60 consecutive pause times (horizontal X-axis), and their duration transformed to z-scores (vertical y-axis). The red bars denote +/− 2 standard deviations (S.D.) of z-scores. Note that all pause times of this individual (blue dots) lie below 2 SD of normo-typical performance. The performance is denoted below the curve as 0.000% (i.e., the number of aberrant pauses above 2 SD is 0/60 = 0.000%).
Only the occurrence of pauses longer than 100 milliseconds was recorded due to an increasing inaccuracy of identifying the borders of sound onset and offset of pause time durations shorter than this time limit. The identification of the boundaries between speech offset and onset was automatically performed by Fast- Fourier analysis of the wave-files.
After 60 successive pauses had been recorded, the identification of additional pauses was automatically stopped.
Thereafter, the duration of each individual pause time was automatically recalculated into z-values. This recalculation was based on the total number of typical reference values included in the database of the test (n = 347, ages 10–87 years). This database of reference values comprises normo-typical school children (n = 164), university students (n = 53), and healthy elderly who had responded to local advertisements (n = 130). Subgroups of these reference values are included in several publications [28, 31, 32].
The reason for the recalculation into standardized z-values was that the means of the successive pause time durations differed between each other, as did the standard deviations. Based on this recalculation, 60 z-values were plotted on the resulting graphical presentation of pauses obtained during a subject’s performance. In addition, the graphical presentation also includes an upper limit of 2 standard deviations for each z-value. An aberrant pause time means that the pause time duration lies above 2 standard deviations from the reference value. Each pause time is also depicted in a scroll list, which shows at which time the registered pause occurred, the corresponding z-value, and if the pause was aberrant or not.
The number of aberrant pauses of a performance was defined as follows: During a normal performance, an upper limit of 6 aberrant pauses was allowed. This was motivated with reference to Sonuga-Barke and Castellanos  who hypothesized that the brain’s default mode network could interfere with test performance with a frequency of 0.01 < Hz < 0.1. Therefore, a total of 6 aberrant pauses (of the total number of pauses) were included in a normal performance.
Clinical neuropsychological tests
The selection of clinical neuropsychological tests, containing predominantly processing speed and attention, was based on results from earlier studies [13,14,15, 17] which used more extensive neuropsychological test batteries and consequently found impairment in these cognitive areas. Due to the core symptom of exhaustion, we tried to shorten the testing period. Matrix Reasoning was included as an estimate of premorbid IQ.
Tests from the Wechsler adult intelligence scale-IV [34, 35]
Digit Span (DS). The task is a measure of working memory capacity, i.e., the cognitive ability to store and manipulate information transiently.
Matrix Reasoning (M) is a test of nonverbal abstract problem solving and inductive reasoning. The task is to view a series of increasingly difficult visual patterns and select the picture that fits the array from five options. The test was used as a test of inclusion to ensure an adequate level of reasoning skills and abstract thinking for the remaining tasks.
Symbol search (S.S.) The task is to determine whether a target symbol appears among various simple symbols. The subject is required to mark either the yes or no checkbox with a pencil in response to as many items as possible within 2 min. The test is designed to assess information processing speed and visual perception.
Coding is a pencil and paper test of psychomotor performance in which the subject is given a key grid of numbers and matching symbols and a test section with numbers and empty boxes. The test consists of filling as many empty boxes as possible with a symbol matching each number. The test demands attentional control, attentional switching, graphomotor control, associative learning, visuospatial processing, and executive control/cognitive control.
Test scores are presented both as raw scores and as age corrected scaled scores according to the manual.
The Ruff 2 & 7 selective attention test
This test was developed to assess sustained and selective visual attention . The test consists of a series of 20 trials of a visual search and cancellation task. Each trial lasts 15 s (in total 5 min). The respondent detects and marks through all occurrences of the two target digits: 2 and 7. Two types of trials are presented in the test: Automatic detection trials, in which target numbers are presented among distractor letters, and Controlled Search trials, in which target numbers are presented among distractor numbers. Selecting targets from different stimulus categories represent parallel search or automatic information processing. In contrast, selecting targets from the same stimulus category requires a serial search or controlled information processing. Thus, working memory and effortful processing of stimulus characteristics are needed to select targets from distractors effectively. The following scores are used: Automatic Detection Speed, the raw score representing the sum of correct letter trials; Automatic Detection Accuracy (ADA), where the number of errors modifies the raw ADS score; Controlled Search Speed (CSS), i.e., the total number of hits in the number trials; Controlled Search Accuracy (CSA), where the number of errors modifies the raw CSS score. According to the T-scores manual, raw scores were corrected for age and educational level .
Further, composite scores were created for Total Speed and Total Accuracy by summing the T- scores for ADS and CSS, and ADA and CSA respectively. Finally, a Discrepancy Analysis was conducted based on the differences between the two speed and the two accuracy T-score. The absolute value of this difference constitutes the Total Difference score with higher values indicating increasing difficulties in coping with issues of selective attention. The manual provides significance levels associated with Speed, Accuracy, and Total Difference scores, indicating impaired performance compared to the normative group. Test scores are presented both as raw scores and as corrected scaled scores for age and educational level according to the manual.
The means and standard deviations of the raw scores of each test were used in the analyses. Regression analyses and correlation analyses (Pearson’s correlations) were used to compare the mean number of aberrant pauses of the two MCS trials (dependent variable) and the WAIS subtests and the RUFF 2&7 (independent variables). A stepwise regression analysis was also performed to assess which predictor variables accounted independently for MCS variance. Sensitivity, specificity, the ROC curve, and Area Under the Curve (AUC) when using MCS to discriminate between SED patients and healthy controls was calculated employing R-4.0.2 statistical software  and the pROC package . The Youden index  (= sensitivity + specificity - 1) was defined for each point on the ROC curve and the point giving the largest index value was selected as a cut-off. Power, i.e., the probability to receive a significance level = 0.05, was calculated for the significant findings assuming that the population effect size equals observed sample effect size.
Patients were informed about the study by a written letter. Potential participants were provided with detailed information by the research psychologist before they signed informed consent. All participants could withdraw their consent to participate at any time. The Regional Ethical Board in Stockholm, Sweden approved the study (Dnr: 2017/2298–31/2).