Neuropsychological and psychiatric outcomes in encephalitis: A multi-centre case-control study

Objectives Our aim was to compare neuropsychological and psychiatric outcomes across three encephalitis aetiological groups: Herpes simplex virus (HSV), other infections or autoimmune causes (Other), and encephalitis of unknown cause (Unknown). Methods Patients recruited from NHS hospitals underwent neuropsychological and psychiatric assessment in the short-term (4 months post-discharge), medium-term (9–12 months after the first assessment), and long-term (>1-year). Healthy control subjects were recruited from the general population and completed the same assessments. Results Patients with HSV were most severely impaired on anterograde and retrograde memory tasks. In the short-term, they also showed executive, IQ, and naming deficits, which resolved in the long-term. Patients with Other or Unknown causes of encephalitis showed moderate memory impairments, but no significant impairment on executive tests. Memory impairment was associated with hippocampal/medial temporal damage on magnetic resonance imaging (MRI), and naming impairment with left temporal and left frontal abnormalities. Patients reported more subjective cognitive complaints than healthy controls, with tiredness a significant problem, and there were high rates of depression and anxiety in the HSV and the Other encephalitis groups. These subjective, self-reported complaints, depression, and anxiety persisted even after objectively measured neuropsychological performance had improved. Conclusions Neuropsychological and psychiatric outcomes after encephalitis vary according to aetiology. Memory and naming are severely affected in HSV, and less so in other forms. Neuropsychological functioning improves over time, particularly in those with more severe short-term impairments, but subjective cognitive complaints, depression, and anxiety persist, and should be addressed in rehabilitation programmes.


Results
Patients with HSV were most severely impaired on anterograde and retrograde memory tasks. In the short-term, they also showed executive, IQ, and naming deficits, which resolved in the long-term. Patients with Other or Unknown causes of encephalitis showed moderate memory impairments, but no significant impairment on executive tests. Memory impairment was associated with hippocampal/medial temporal damage on magnetic resonance imaging (MRI), and naming impairment with left temporal and left frontal abnormalities. Patients reported more subjective cognitive complaints than healthy controls, with tiredness a significant problem, and there were high rates of depression and anxiety in the HSV and the Other encephalitis groups. These subjective, self-reported complaints,

Introduction
Encephalitis-which refers to inflammation of the brain caused by infection or autoimmune disease-can have debilitating long-term consequences [1]. Infectious causes, in particular herpes simplex virus (HSV), have long been considered the main cause of encephalitis [2,3]. The relatively recent discovery of cell-surface antibodies has enabled a broadening of the search for aetiologies of encephalitis. This has allowed the accurate diagnosis of an increasing number of patients with autoimmune encephalitis [4,5], including those involving the n-methyl-d-aspartate (NMDA) receptor [6] and the leucine-rich glioma 1 (LGI-1) protein [7]. However in 30-50% of encephalitis patients, the aetiology still remains unknown, despite intensive investigations [2,3,8,9].
There have been few studies comparing neuropsychological outcomes across different encephalitis aetiologies. Before the identification of cell-surface antibodies, several studies were published on outcomes in patients recovering from HSV encephalitis [10][11][12]. Compared with other infectious aetiologies, Hokkanen et al. [13] found that HSV patients showed greater difficulty on verbal memory, semantic, and visuo-spatial functions than the other aetiological groups. Studies focusing specifically on infectious causes of encephalitis other than HSV found impairments in executive function [14], retrograde memory [13], visuo-spatial processing [15], and working memory [16]. Pewter et al. [17] compared HSV and 'non-HSV' patients, who included infectious and autoimmune aetiologies. They found significant and widespread cognitive deficits in HSV patients, whereas 'non-HSV' patients showed more isolated disorders of executive function [17]. Research focusing on autoimmune aetiologies indicates neuropsychological impairments in attention, executive function, working memory, visuospatial and language abilities, and memory [18][19][20][21]. However, these various studies did not use uniform neuropsychological assessments to measure outcome, nor did they investigate how outcomes related to patients' self-reported perception of their cognitive abilities, depression, or anxiety. In most cases, they also did not examine changes in outcomes over time.
In terms of neuroimaging, HSV encephalitis causes persisting hyper-intensities on T2-weighted MRI scans and substantial volume loss in the medial temporal lobes [10,22]. Lateral temporal and widespread cortical damage is also common [10,22,23]. The severity of anterograde amnesia after HSV encephalitis appears to be correlated with the extent of pathology in the medial temporal regions, with bilateral damage predictive of severe amnesia [10,11]. Limbic encephalitis with autoimmune aetiologies has been specifically associated with hippocampal atrophy, which is reported to correlate with disease severity and memory deficits [22][23][24].
In this study, we have investigated (i) neuropsychological outcomes; (ii) patients' perceptions of their cognition, and (iii) psychiatric outcomes in terms of depression and anxiety at three time-points after hospital discharge. We examined how this was related to clinical variables and neuroimaging findings. The study was a component of a NIHR programme grant on encephalitis (ENCEPH-UK, www.encephuk.org), examining predictors of outcome [25,26].

Aims
The purpose of the current study was to examine: 1. The presence of distinct patterns of neuropsychological impairment across different subgroups of encephalitis patients, e.g. whether memory impairments were more evident in specific diagnostic subgroups, whether executive function impairments were more pronounced in other causes of encephalitis, and whether there were differences in IQ and naming; 2. The patients' perceptions of the impact of encephalitis on their lives, with respect to subjective, self-reported cognitive complaints, depression and anxiety; 3. The changes in neuropsychological performance across encephalitis subgroups when measured at different time-points; and 4. The relationship between clinical variables, neuroimaging, and neuropsychological/psychiatric outcomes.

Methods
Two cohorts of encephalitis patients were recruited and investigated between 2012 and 2015.
The first cohort was assessed 4 months after their hospital discharge to measure 'short-term' outcomes, and then again 9-12 months after the first assessment to examine 'medium-term' outcomes. This will be referred to as the "short-and medium-term outcome cohort". The second cohort was recruited at least 1 year after hospital discharge in order to examine the 'longterm' outcomes. This will be referred to as the "long-term outcome cohort". In both cohorts, patients with current or prior neurological conditions or brain injuries impacting on cognitive function were excluded in order to ascertain the specific association between neuropsychological impairment and encephalitis. Migraine was not an exclusion criterion. A healthy control group without neurological or psychiatric illness, subsequently referred to as "healthy controls" or "healthy control group", was recruited from the public. Fig 1 shows the recruitment details of both patient cohorts as well as the healthy control group. The study received ethical approval by the NRES Committee East Midlands-Nottingham 1 REC (ref 11/EM/0442). Written informed consent was obtained from all participants. For the main analyses, patients in each cohort were categorised into three diagnostic groups: (i) HSV encephalitis (referred to as 'HSV'), (ii) Encephalitis of Other Identified Causes (referred to as 'Other'), and (iii) Encephalitis of Unknown Cause (referred to as 'Unknown'). This was done according to predetermined case definitions for HSV and other infectious causes [25][26][27] and autoimmune encephalitis [28]. Medical notes and discharge letters were additionally reviewed by the study's panel of infectious disease and neurology specialists.
For further exploratory analyses, the Other group was split into two subgroups: encephalitis resulting from non-HSV infection (referred to as 'Infections') and autoimmune encephalitis (referred to as 'Autoimmune').

Participants
Short-and medium-term outcome cohort. Patients in the short-and medium-term outcome cohort were identified by research nurses on hospital wards at ENCEPH UK sites. Fortyfive participants with encephalitis were recruited and assessed approximately 4 months after hospital discharge (4.3±1.9 months) to assess short-term outcomes. When available, they were assessed again 9-12 months after the first assessment (N = 28, 10.5±1.9 months) for mediumterm outcomes. There were 20 patients included in the HSV group, 10 patients in the Other group, and 15 patients in the Unknown group. Table 1 shows the means, standard deviations and frequencies for age, gender and Glasgow Coma Scale (GCS) scores at admission across these three diagnostic groups.
As mentioned above, the Other group was split into two subgroups for further exploratory analyses. In the short-and medium-term cohort, the Other group thus comprised 4 'Infections' patients and 6 'Autoimmune' patients.
Long-term outcome cohort. Patients in the long-term outcome cohort were identified by consultants across ENCEPH UK hospital sites, from the Public Health England study database [2] or they self-referred in response to a notice by the Encephalitis Society [27]. There were 81

PLOS ONE
encephalitis patients, consisting of (i) 30 HSV patients, (ii) 24 Other patients, and (iii) 27 Unknown patients (see Table 1 for the means, standard deviations and frequencies of age and gender).
For further exploratory analyses, the Other group in this long-term outcome cohort comprised 10 'Infections' patients and 14 'Autoimmune' patients.
Healthy controls. Healthy controls without neurological or psychiatric history were recruited via community and website advertisements. From responders to these advertisements, we recruited and assessed a total of 78 participants. Based on whole-group demographic variables (mean age, gender distribution, and mean premorbid IQ), 45 healthy controls were matched to the short-and medium-term outcome cohort and 70 healthy controls were matched to the long-term outcome cohort. Table 1 shows the means, standard deviations and frequencies for age and gender in each matched healthy control group.
Anterograde memory: Doors and People battery [31], including verbal and visual, recall and recognition tests.
Retrograde memory: Autobiographical Memory Interview [AMI, [32]], assessing personal semantic and autobiographical incidents memory in terms of total scores.
Subjective self-report measures were used to assess psychiatric outcomes in terms of depression and anxiety, and self-perceived cognitive complaints: Self-perceived cognitive complaints: The A-B Neuropsychological Assessment (self-report) Schedule: ABNAS [42] which examines tiredness, mental speed, memory, concentration, motor skills, and language.
Information on clinical variables was also collected in the short-and medium-term outcome cohort: Time between hospital admission and commencement of appropriate drug treatment (antiviral, immunosuppressant), duration of hospital stay, and time between hospital admission and neuropsychological testing. This information was not consistently available in the (retrospectively collected) long-term outcome cohort.

Procedure
Neuropsychological testing was conducted at the participant's home or one of the research sites (King's College London, or Walton Centre, Liverpool).
The same battery of neuropsychological tests was used in the short-and medium-term outcome cohort as well as the long-term outcome cohort.

Imaging
In the short-and medium-term outcome cohort, MRI scans were obtained during the course of clinical care [27]; the scan closest in date to the first neuropsychological assessment was analysed.
MRI analysis involved a qualitative, systematic assessment by two neuroradiologists. They knew the patients had encephalitis but were blinded to aetiology and neuropsychological test results. The brain was evaluated in pre-specified regions-of-interest: hippocampi (head, body and tail), amygdalae, para-hippocampal gyri, occipito-temporal gyri, lateral temporal lobes (temporal stem, superior, middle and inferior temporal gyri), insulae, cingulate gyri, fornices, mammillary bodies, and frontal lobes (inferior and dorsolateral areas). These regions were examined for oedema (signal change on T2-weighted or fluid-attenuated inversion recovery [FLAIR] images), parenchymal loss (including cystic encephalomalacia), or gliosis. We focused on the site of damage, rather than its nature (inflammatory change, signal alteration, volume loss), and we determined whether 'damage' was present or absent at each anatomical location. The inter-rater reliability between the two neuroradiologists was assessed using a Kappa coefficient score on this binary evaluation (presence/absence of damage) across all brain regions.

Statistical analysis
Shapiro-Wilk tests were used to assess for normality. One-way ANOVAs (parametric) or Kruskal-Wallis tests (non-parametric) were conducted. ANCOVAs were used when neuropsychological scores correlated significantly with either BDI or BAI. For parametric data, Hochberg GT2 (homogeneity assumed) or Games-Howell (homogeneity not assumed) post-hoc tests were applied to compare performance in individual encephalitis groups with that of healthy controls. For non-parametric data, Mann-Whitney U tests were used for pairwise comparisons; here, a corrected alpha of p = 0.02 was applied. In the short-and medium-term outcome cohort, Spearman correlations were used to analyse associations between clinical and neuropsychological variables, and where these were significant, linear regression analyses were conducted, which included age and premorbid IQ as confounding variables. To increase the sample size, all patients were merged into one group for correlation and regression analyses.
As described above, the Other group in both cohorts was subdivided into other Infections and Autoimmune aetiological subgroups. Thus, for the further exploratory analyses there were a total of four patient groups in each cohort: HSV, Infections, Autoimmune, and Unknown.
Each of these four patient groups per cohort was compared with its own healthy control group, matched for mean age, gender distribution, and mean premorbid IQ. T-tests or Mann-Whitney U comparisons were then employed to identify specific neuropsychological impairments within these aetiologies.
Rate of change between short-and medium-term timepoints was analysed by conducting one-way ANOVAs or Kruskal-Wallis tests on difference scores. These analyses were conducted only in the 28 participants who were assessed at both timepoints. Those lost to followup were not included.
In the short-and medium-term outcome cohort, imaging (damage/no damage) comparisons between groups were analysed using chi-square test of independence, and associations between imaging and neuropsychological findings were examined using point-biserial correlations. Where correlations were significant, linear regression analyses were conducted and included age and premorbid IQ as confounding variables.

Short-and medium-term outcome cohort
To examine short-term outcomes, 45 patients were assessed and compared with 45 healthy controls (see Table 1 for demographic characteristics and Table 2 for neuropsychological/psychiatric outcome means and standard deviations). The recruitment flowchart outlines the specific aetiologies within each patient group (Fig 1).
For further exploratory analyses the Other group was split into two subgroups. The 4 'Infections' patients comprised 2 males and 2 females and had a mean age of 59.90 (SD = 8.81). The 6 'Autoimmune' patients comprised 1 male and 5 females and had mean age of 37.00 (SD = 14.64).   (Table 2) with striking impairments in the HSV group on both components (p<0.001), and in the Other group on autobiographical incidents (p<0.001).

Language & visual semantic access
On Graded Naming, the HSV group showed severe impairment [F(3, 85) = 6.93, p<0.001; post-hoc versus healthy controls, p<0.01]. On visual semantic access, all mean scores were in the normal range.

Perception
All patient groups scored within the non-impaired range and Kruskal-Wallis and ANOVA analyses did not reveal significant differences (p's>0.19, Table 2).   Table 2).

Rate of change between short-and medium-term outcome
To examine change between short-and medium-term outcomes, 28 of the 45 patients studied in the short-and medium-term outcome cohort were reassessed 9-12 months after their first assessment. Difference scores (T2-T1) indicating rate of change were compared across encephalitis groups.
We compared mean age, GCS score at hospital admission as an indicator of illness severity, and aetiology in the participants who dropped out versus those who did not. T-tests revealed no significant differences in age or admission GCS score (p's > 0.60). A Fisher's exact test showed no significant differences in aetiology (p = 0.10). The reasons for drop-out most commonly cited by our participants were tiredness and physical illness. Table 3 shows that the time between hospital admission and commencement of appropriate treatment correlated negatively with measures of intelligence, retrograde and anterograde memory, and executive function (i.e. longer time until treatment was associated with worse performance). Older age was correlated with lower scores on naming (rho = -0.30), and more errors on the Hayling test (rho = 0.44) (both p<0.05). Treatment was instigated earlier if the GCS score was low (rho = 0.34, p<0.05). To further examine these correlations, we also carried out linear regression analyses, including time until treatment, premorbid IQ and age as potential predictors of neuropsychological outcome. On inputting these variables, we obtained significant overall regressions in predicting verbal IQ [R 2 = .42, F(3, 37) = 8.95, p < .001], verbal recall [R 2 = .34, F(3, 37) = 6.28, p = .001] and rule detection [R 2 = .42, F(3, 36) = 8.69, p < .001] scores. However, of our specific predictors, only premorbid IQ (β = .60, p < .001; β = .54, p = .001; β = -.54, p < .001, respectively) and age (for rule detection only: β = .34, p = .01) were statistically significant. Time until appropriate treatment did not contribute significantly to the regression. Table 3 shows that length of hospital stay also correlated significantly with measures of memory and executive function (i.e. longer stay associated with worse performance). On the other hand, it correlated with fewer patient-perceived language and motor complaints. On linear regression, increased length of stay remained a significant predictor of reduced patientperceived language complaints when covarying for premorbid IQ and age (β = -.55, p < .001) [R 2 = .39, F(3, 34) = 7.36, p = .001].

Clinical correlations
Imaging. Table 4 summarises the imaging findings on clinical MRI scans obtained closest to the time of the first neuropsychological assessment. The kappa coefficient score for inter-rater reliability between the two individuals assessing the binary presence/absence of damage was 0.98. There were significant differences between the three patient groups for hippocampal damage (particularly on the left) and medial temporal damage; both were most common in the HSV group. In contrast, dorsolateral frontal damage was most common in the Unknown group, and there were significant differences in comparing the distribution of left or right medial temporal versus dorsolateral frontal changes across groups (left, X 2 = 19.21, p<0.001; right X 2 = 19.18, p<0.001). Table 5 shows that left hippocampal and left medial temporal damage were correlated with lower IQ; and hippocampal and medial temporal damage with lower visual recall memory as well as with poorer response initiation and suppression. Right hippocampal, medial temporal, and lateral temporal damage were all correlated with poor visual recognition memory. There were also significant correlations between left temporal or frontal damage and graded naming.

Long-term outcome cohort
Eighty-one patients completed neuropsychological and psychiatric assessments more than one year after acute encephalitis and were compared with the scores of 70 healthy controls matched on age and premorbid IQ. Table 1 summarises the demographic characteristics across groups and Table 6 shows the means and standard deviations of the neuropsychological and psychiatric outcome measures.

Retrograde memory
Personal semantic memory did not differ significantly across groups (p = 0.18, Table 6). There was a significant group effect on autobiographical incidents memory [H(3) = 13.15, p = 0.004], reflecting impaired HSV and Unknown performance relative to healthy controls (post-hoc p<0.001 and p = 0.03, respectively). Table 6 shows all executive function scores were in the non-impaired range (p's>0.11).
On visual semantic access, all scores approached ceiling. Perception. Performance approached ceiling on all perception measures (Table 6).  Table 6).
Patient-perceived (subjective) cognitive function. Patients reported more complaints than healthy controls [Fig 3(B) Fig 3(B) shows that the HSV group expressed significantly more complaints than healthy controls across all domains (post-hoc p's<0.01). The Unknown group also reported more difficulty than healthy controls on memory, language, and tiredness (post-hoc p's<0.01). The Other group reported greater tiredness than healthy controls (post-hoc p = 0.03).

Discussion
The current study investigated neuropsychological and psychiatric outcomes across three different encephalitis groups in the short-term (4 months post-discharge), medium-term (9-12-months follow-up after first assessment), and the long-term (>1 year post-discharge). We found that that neuropsychological and psychiatric outcomes after encephalitis varied according to the aetiology. Memory and naming impairments were most severely affected the HSV group, and this was associated with more severe medial temporal lobe damage on MRI. Across groups, cognitive performance improved over time, but subjective cognitive complaints, depression, and anxiety persisted. We will discuss these findings with respect to our original aims in turn.
In line with our first aim, which was to examine neuropsychological profiles across encephalitis aetiological groups, differential patterns of neuropsychological impairment were found across the patient groups. In the short-term, both HSV and patients with Other or Unknown causes of encephalitis were impaired on anterograde memory, but this was more severe in the HSV group, consistent with previous findings [15,17]. The Other group was also impaired across anterograde memory tests, statistically significant for verbal recognition memory. In terms of autobiographical memory, HSV patients showed short-term (<1 year) impairments in the recall of personal semantic facts, and short-and long-term impairments (>1 year) in recalling autobiographical episodes. The Unknown aetiology group also showed long-term impairment in recalling autobiographical memories. Other authors have also found mild/ moderate memory impairments in 'non-HSV' encephalitis patients [13,17].
With respect to other cognitive functions, we did not find more pronounced executive impairment in either the Other or the Unknown groups than the HSV patients, contrary to Pewter et al.'s [17] finding of isolated executive impairment in 'non-HSV' patients. In the short-term, only the HSV group showed impairments on executive function, naming, and fullscale IQ. The naming impairment occurred in the presence of normal verbal fluency (FAS) and normal visual semantic access. In the short-term, both the Other and the Unknown groups were unimpaired on executive function and naming, and only the Other group was impaired on verbal and full-scale IQ. In the long-term, the HSV group showed only a trend towards impairment on naming, but otherwise there were no long-term impairments across the aetiological groups on executive, naming, or IQ tasks. In other words, executive function, naming, and IQ impairments appeared to resolve after the first year.
With respect to our second aim, which was to assess the patients' subjective, self-reported perceptions of their cognitive abilities, there were differences in tiredness, mental speed, concentration, memory, and language in the encephalitis groups compared with the healthy controls. These occurred both in the short-and long-term after the illness (Fig 3), with HSV the most, and the Other group the least, severely affected. Tiredness was a particularly prominent complaint, as has also been reported elsewhere [43]. With respect to psychiatric outcomes, there were significantly higher short-term rates of depression in the HSV and Other groups, and anxiety in the Unknown group, than in healthy controls. In the Unknown group, this anxiety might reflect the negative impact of receiving an uncertain diagnosis. In the long-term, all patient groups continued to show raised levels of depression and anxiety. This is in line with findings from another study which examined psychiatric sequelae of encephalitis, suggesting depression and anxiety can arise both from neurobiological changes and from the psychological adjustment to encephalitis [44]. It is notable that, in terms of patients' subjective perceptions of cognitive function and mood state, complaints remained severe after the first year. This appears particularly at odds with current and previous suggestions of improved objectively measured cognitive function over time [45,46], and may reflect the dawning realisation that further improvement after severe neurological illness was likely to be slow and limited [43,[47][48][49]. Correlations between subjective complaints and objective performance were generally low.
The third aim was to examine the rate of change between the short (4 months)-and medium-term (9-12 months later) post-discharge. Rates of change were broadly equivalent across the patient groups during the first year, except that the Other group showed significant improvement on verbal IQ, whereas the Unknown group showed a decline in verbal IQ. During this first year, the HSV group showed a widespread pattern of cognitive impairment (memory, executive functioning, verbal IQ performance, and naming). After the first year, these had narrowed down to a predominant deficit in anterograde and retrograde memory. In the Other and Unknown cause groups, cognitive deficits which had been evident during the first year were no longer present, except for the Unknown group's performance on autobiographical event memory.
Fourthly, we examined the relationship between clinical variables and short-term neuropsychological and psychiatric outcomes. Time between admission and commencement of appropriate treatment, as well as a longer hospital stay, were both correlated with poorer memory and executive outcomes, emphasizing the importance of timely treatment [2,21]. Timely treatment may be particularly important in older adults and those with lower premorbid intellectual abilities. Furthermore, despite the association with poorer memory and executive outcome on objective assessments, longer hospital stay was also correlated with fewer selfreported motor and language complaints, possibly because these had been more fully addressed during a more extensive stay. In line with this, a longer hospital stay significantly predicted lower self-reported language complaints even after accounting for age and premorbid IQ. It is possible that reducing self-reported complaints by providing thorough clinical care may improve mood, which in turn might reduce complaints further [50] and enhance objectively measured cognitive performance-this requires further investigation.
With regard to the MRI findings across all three patient groups, there were significant correlations between short-term visual recall and recognition scores and the presence of hippocampal and medial temporal lobe pathology. Naming impairment was associated with the presence of left-sided damage on MRI. Linear regression analyses accounting for premorbid IQ and age revealed that hippocampal and temporal damage remained significant predictors, directly supporting the relationship between brain lesion and short-term neuropsychological impairments.

Study limitations
Even though the present study was multi-centre across UK-wide NHS departments, recruitment to the Other group, particularly the autoimmune aetiologies was difficult. At the time of recruitment, cell-surface antibodies aiding autoimmune diagnosis were not always consistently being tested in hospitals. In this neuropsychological sub-study, HSV cases comprised 33.3% of the short-and medium-term cohort, and 55% of the long-term cohort. The ENCEPH-UK parent study comprised 20% of HSV cases, which is very similar with the 19% HSV cases recruited in the HPA study [2]. The increased representation of HSV cases in the neuropsychological sub-study may have resulted from a response bias since participants could opt in to take part in the neuropsychological assessments if they wished. Overall, it is therefore important to state that the diagnostic proportions reported here are not reflective of true prevalence rates, especially with recent advances in diagnostic tools for autoimmune encephalitis. Secondly, there were relatively high rates of attrition between the initial assessment and the follow-up 9-12 months later. This could not be attributed to differences in aetiology, illness severity (based on GCS at admission) or age. In fact, the reasons for drop-out most commonly cited by our participants were tiredness and physical illness, consistent with subjective complaints reported across all three encephalitis subgroups. Finally, the patient participants in the short-and medium-term outcome cohort were not the same individuals as in the long-term outcome cohort. While this does not affect the interpretation of the findings within each cohort, it is nonetheless worth noting that the 'longitudinal' conclusions of this study should be interpreted cautiously.

Summary and conclusions
In summary, the HSV group was the most impaired, and memory was the most severely affected cognitive function within this group. In the short-and medium-term assessments (4 months to 1-year post-discharge), the HSV group also showed deficits in executive function, verbal IQ, and naming. In the long-term (>1 year), these deficits had narrowed down to impaired anterograde and autobiographical memory. The Other and Unknown groups also showed most impairment in memory relative to other neuropsychological functions, which is contrary to some previous observations [17]. On subjective evaluations, the HSV group reported self-perceived problems in memory, language, tiredness, concentration, and mental speed, as well as depressed mood; and these were still present at >1 year. In the Other and Unknown groups, there were also patient-perceived, subjective cognitive complaints, anxiety, and depression, with tiredness being identified as a common problem. In particular, the Other group reported high rates of depression, whereas the Unknown group described anxiety, possibly indicating that receiving an uncertain diagnosis increases anxiety. In terms of associated clinical factors, delayed time until appropriate treatment, and prolonged hospital stay, were correlated with poorer outcomes across the patient groups, underlining the importance of timely treatment [1].
These findings have implications for neuropsychological rehabilitation. Outcomes for patients with encephalitis differed across the aetiological groups, and rehabilitation should be tailored to this. It is important that psychiatric conditions including depression and anxiety are treated appropriately as this may positively influence the success of neurorehabilitation for cognitive deficits [44]. Compensatory strategies, aiming to capitalize on preserved executive function, could be used to support memory [51], and self-perceived cognitive complaints, fatigue, depression, and anxiety are particularly important to address.
Supporting information S1 Data. The supporting information provided with this manuscript contains 3 SPSS database files. These files contain the raw data. The files labelled "short-and medium-term outcome cohort and matched controls" and "long-term outcome cohort and matched controls" contain the demographic information, neuropsychological assessment scores, psychiatric assessment scores, the self-rated cognitive impairment scores and the clinical variables where appropriate. The file labelled "short-and medium-term outcome cohort_MRI data" contains the data extracted as binary damage/no damage from clinical scans made available to the researchers. (ZIP) 14