The Nottingham Fatigue after Stroke (NotFAST) study: factors associated with severity of fatigue in stroke patients without depression

Objective: To identify factors associated with post-stroke fatigue in a sample of stroke survivors without depression. Design: Cross-sectional cohort study. Setting: Recruitment was from four stroke units in the UK. Subjects: Participants were assessed within four to six weeks of first stroke; those with high levels of depressive symptoms (score ⩾7 Brief Assessment Schedule Depression Cards) were excluded. Main measures: Participants were assessed after stroke on the Fatigue Severity Scale of the Fatigue Assessment Inventory, the Rivermead Mobility Index, Nottingham Extended Activities of Daily Living scale, Beck Anxiety Index, Sleep Hygiene Index, 6m walk test, and measures of cognitive ability. Results: Of the 371 participants recruited, 103 were excluded and 268 were assessed. Of the latter, the mean age was 67.7 years (SD 13.5) and 168 (63%) were men. The National Institutes of Health Stroke Scale mean score was 4.96 (SD 4.12). Post-stroke fatigue was reported by 115 (43%) of participants, with 71 (62%) reporting this to be a new symptom since their stroke. Multivariate analysis using the Fatigue Severity Scale as the outcome variable found pre-stroke fatigue, having a spouse/partner, lower Rivermead Mobility Index score, and higher scores on both the Brief Assessment Schedule Depression Cards and Beck Anxiety Index were independently associated with post-stroke fatigue, accounting for approximately 47% of the variance in Fatigue Severity Scale scores. Conclusions: Pre-stroke fatigue, lower mood, and poorer mobility were associated with post-stroke fatigue.


Introduction
Fatigue is one of the most distressing symptoms after stroke and its treatment is an important unmet need for stroke survivors and their carers. 1 Post-stroke fatigue adversely affects daily occupational performance and roles, return to work, participation in rehabilitation programmes, and quality of life. [2][3][4] Many studies have highlighted associations between depressive symptoms and fatigue after stroke, leading to the suggestion that post-stroke fatigue is a symptom of depression. 5,6 However, post-stroke fatigue has also been found independently of low mood. 7,8 No previous studies have excluded participants with depressive symptoms, allowing fatigue to be investigated independently of this factor. Aside from the link with depression, evidence identifying other factors associated with post-stroke fatigue is limited and often conflicting.
Given the impact of fatigue on outcomes, it is important that post-stroke fatigue is appropriately identified and managed in clinical practice. The aim of the Nottingham Fatigue After Stroke (NotFAST) study was to identify factors associated with post-stroke fatigue in a sample of stroke survivors without depression, in order to inform clinical practice and the development of effective interventions.

Methods
NotFAST was a multi-centre, longitudinal cohort study.

Stroke inpatients were identified by research assistants and by UK Clinical
Research Network clinical trials researchers. Those eligible for inclusion had a clinical diagnosis of stroke, were aged ≥18 years, able to give informed consent, and had no previous history of stroke. Potential participants were excluded if they were unable to read or speak English, or had a documented diagnosis of dementia.
Participants were identified within four weeks of stroke and, after obtaining informed consent, screening for dysphasia and depressive symptoms was completed. The Sheffield Screening Test for Acquired Language Disorders was used for dysphasia, and those scoring below the age-recommended thresholds 9 were excluded in order to ensure that subsequent assessments could be completed. Depressive symptoms were assessed using the Brief Assessment Schedule Depression Cards; those who scored ≥7, consistent with a diagnosis of a depressive disorder, 10 were excluded.
Participants were assessed on the following:  Self-reported fatigue using the nine item Fatigue Severity Subscale of the Fatigue Assessment Inventory 11 (score range 7-49, with higher scores indicative of greater fatigue). Participants were also asked to recall their pre-stroke fatigue level using the Fatigue Severity Subscale. A score >36 was used to indicate clinically significant fatigue, an approach used by others. 12  Mobility and activities of daily living (ADLs) using the Rivermead Mobility  Univariate analyses were used to identify factors significantly associated with post-stroke fatigue. Linear regression explored the unadjusted relationships between fatigue (Fatigue Severity Subscale score) and continuous variables, and differences in Fatigue Severity Subscale scores for categorical variables were investigated using one-way ANOVA or t-tests. extreme outliers using scatter plots, and residuals were computed to test assumptions of normality and constant variance. To determine factors independently associated with fatigue, using the Fatigue Severity Subscale score as the dependent variable, an explanatory model using multivariable linear regression analysis was developed whereby those variables that were statistically significant in univariate analysis (p ≤ 0.05) were entered into a multivariable model and a step-wise modelling procedure followed to obtain a final model of only statistically significant (p ≤ 0.05) variables.

Results
Three hundred and seventy-one participants were recruited. After exclusion on screening variables and losses (n=103), further assessments were conducted with 268 participants.
Details of study recruitment and retention are presented in Figure 1.
The demographic and clinical characteristics of participants are presented in Table 1. Participants comprised mainly men (63%) and had a mean age of 67.7 years (SD 13.5, range 24 -94 years). Most (91%) had had an infarction, with a high proportion of lacunar strokes (41%). National Institutes of Health Stroke Scale (NIHSS) 23 scores indicated mild or/moderate stroke severity 24 (mean 4.96, SD 4.12). The mean days post-stroke at which assessments were completed was 23.61 (SD 13.15), with 164 assessments (61%) completed at home, and the remainder completed in hospital (30%) or in clinic/other settings (9%). [Table1] The distributions of scores are shown in Table 2. Mean Fatigue Severity Subscale scores were significantly higher post-stroke than pre-stroke (mean difference 10.4, 95% C.I. 8.7, 12.2, p < 0.001).

Factors associated with fatigue
The correlation between fatigue and other variables showed 12 statistically [ Table 6]

Discussion
Fatigue was common at four to six weeks post-stroke in a sample of stroke survivors without depression. Clinically significant levels of fatigue were reported by 43% of participants. Fatigue was associated with self-reported prestroke fatigue, being in a relationship, lower levels of functional mobility, and higher levels of anxiety and depressive symptoms.
Our findings on the frequency of fatigue lie within the range of proportions of stroke survivors with fatigue reported by Choi-Kwon et al., 8 however the large range in frequency of fatigue (23-75%) in the latter review probably reflects the methodological diversity of the included studies. Our figure is considerably higher than that reported by Duncan et al. 25 at one month post-stroke (33%), a study which used a case definition interview, rather than a screening questionnaire, to identify fatigue.
Our sample was relatively unimpaired physically and cognitively, probably due to the predominance of those with lacunar syndrome strokes. The prevalence of fatigue was higher than that reported by Radman et al. (30%) 26 , although the assessment of those with minor stroke (NIHSS 6) in the latter study was conducted rather later, at six months post-stroke. Our findings thus suggest that minor stroke is associated with fatigue frequently, and early, in the recovery process, and therefore has the potential to impact on the early stages of rehabilitation. 3 The association of post-stroke fatigue and retrospectively reported pre-stroke fatigue has been noted previously. 12,27,28 Our sample reported a lower prevalence of pre-stroke fatigue than both Choi-Kwon et al. 27 (38%) and Lerdal et al. 12 (30%), although direct comparison is problematic as fatigue was assessed differently by visual analogue scale and dichotomous question, respectively. Our findings are broadly comparable with those of Chen et al. 28 who also used the Fatigue Severity Subscale to determine the prevalence (22%) of pre-stroke fatigue.
The majority (62%) of those who reported post-stroke fatigue had not experienced fatigue previously, suggesting that the stroke event may be linked to the development of fatigue. However, a high proportion (86%) of our participants who reported fatigue pre-stroke also reported fatigue after stroke, suggesting that some survivors experienced a continuation or exacerbation of pre-existing fatigue following stroke. The causes of pre-stroke fatigue are not clear, and are likely to be multifactorial. 28 Pre-stroke fatigue is, however, difficult to measure accurately, relying on recollection of events prior to stroke, which can result in recall bias.
It has been reported that post-stroke fatigue is more common in single people than in married/cohabiting people, 29 while others have found no significant association with relationship status. 12,30 In contrast, we found that being in a relationship was significantly associated with fatigue, with higher levels of fatigue reported by those with a spouse or partner compared to those who were single, widowed or divorced, and explained the most variation in Fatigue Severity Subscale scores. However, we did not find an association between fatigue and living arrangements, suggesting that the influence of a personal relationship may be different from the potential availability of practical support at home. The reasons for the differences between these findings are unclear, but may result from heterogeneous samples, chance findings, or a complex interplay of social factors warranting further investigation.
Poorer functional mobility was a significant independent predictor of post-stroke fatigue. Reduced mobility may be a consequence of fatigue, although the direction of this relationship is unclear. However, lower physical activity levels at one month after stroke have been shown to predict fatigue at six and 12 months. 25 This suggests that interventions to promote early activity may have the potential to modify fatigue later in recovery. There is a need for longitudinal intervention studies to explore the effect of early activity levels on the development of fatigue.
The results support the suggestion that fatigue may arise due to several factors. This includes neurophysiological factors, such as corticomotor excitability, physical factors, such as increasing level of disability leading to fatigue, and psychological factors, such as mood leading to increased reports of fatigue. It is not possible to differentiate these factors in people with stroke and, indeed, this multifactorial nature is consistent with fatigue also being a problem in people with other disabling conditions, such as multiple sclerosis 31 and rheumatoid arthritis. 32 The association between fatigue and depression has been established previously. 5,6 However, despite having excluded participants with high levels of depressive symptoms, we found that the association between fatigue and depressive symptoms remained. We found a similar association between anxiety symptoms and fatigue. Our findings are consistent with a meta-analysis of psychological factors in relation to fatigue which also found an association between fatigue and mood in those without clinically significant depression or anxiety. 33 Therefore low levels of distress may impact on fatigue, and this would offer justification to address the emotional impact of stroke, even in the absence of clinically significant levels of anxiety or depression.
A limitation of this study is representativeness of the stroke sample. A substantial proportion of participants were recruited from acute stroke wards, where people with more severe strokes were not medically stable, and therefore less likely to participate. This may partially explain the predominance of lacunar syndrome and partial anterior circulation strokes, and the relatively low levels of physical and cognitive impairment in our sample. We also excluded those with aphasia for pragmatic reasons, and accept that our results would more accurately represent the stroke population if these participants had been included.
We had some losses after screening (13%) and participants withdrawing consent or being uncontactable (14%). Not all tests were fully completed, due to interruptions during assessments or participants being unable or declining to complete particular questions or tasks (e.g. 6m walk test, cognitive tests).Diagnosis of stroke classification and NIHSS scores were unavailable in some cases (NIHSS scores available for 203/268 cases (i.e. 76%)), affecting the overall estimation of stroke severity.
We recruited a greater proportion of men (68%) than women, consistent with observations that women are underrepresented in cardiovascular clinical trials. 34 We excluded people with communication problems for pragmatic reasons, but accept that studies investigating fatigue in this population are needed.
One of the key limitations may be our main outcome measure. The Fatigue Severity Subscale is commonly used in stroke research, however there is no validated 'cut-off' score to define clinically significant fatigue after stroke.
Nevertheless, the approach we used is consistent with that used in other studies of fatigue in neurological conditions. 12

Clinical Messages
 Self-reported pre-stroke fatigue, being in a relationship, and poor mobility were all associated with higher levels of fatigue.
 Mood was associated with fatigue, even though those with high levels of depressive symptoms were excluded from the study.
 The factors found to be associated with fatigue require further exploration, but are potential targets for interventions to treat post-stroke fatigue.

Author contributions
AD, NL, NS, NW, AM, PT and GM were responsible for the study conception and design.
LH was responsible for the co-ordination of the data collection.
AD, NL, LH and EW input and analysed the data.
AD, NL and LH wrote the initial draft of the article and all authors contributed to the revisions.

Conflict of interest statement
The authors declare that there are no conflicts of interest.

Funding support
This research was funded by the Stroke Association (reference: TSA 2012/04).