Use of the Fatigue Severity Scale to assess clinically reliable temporal changes in post-stroke fatigue by stroke type and subtype

Introduction

Fatigue management is a critical unmet need that survivors of stroke identify as a high-priority area for research (Hill et al. 2022). This is unsurprising given that approximately half of survivors of stroke experience post-stroke fatigue (pooled prevalence estimate across 31 studies = 48% (95% confidence interval (CI) = 42–53%)) (Alghamdi et al. 2021). Fatigue is a major disabling condition and is a substantial barrier to engaging in rehabilitation and other activities that improve quality of life (Glader et al. 2002; van de Port et al. 2007; Duncan et al. 2012). The World Stroke Organization recommends providing people with stroke and their families/carers with information, education, and strategies to manage fatigue (Johansen Skogestad et al. 2021). Yet, there are no proven treatments for post-stroke fatigue (Ponchel and Bordet 2015).

A recent consensus statement on post-stroke fatigue (English et al. 2024) addressed central questions to research, including its biological causes, how fatigue and its potential causes should be identified in clinical practice, the most promising interventions for management, and the best available outcome measure for research. The Fatigue Severity Scale-7 item version (FSS-7) was recommended as the primary outcome measure for fatigue in all stroke research. The 7-item version of the scale was suggested for research purposes as Lerdal and Kottorp (2011) found that items 1 and 2 in the FSS-9, which reflect the impact of fatigue on motivation and the impact of physical exercise on fatigue, should not be used in a mean score because the FSS-7 shows better validity and reliability and is likely more sensitive for measuring change in fatigue. Although the FSS-7 did show better validity, in Lerdal and Kottorp (2011) baseline Cronbach’s alpha for the FSS-9 and FSS-7 did not vary (0.86 and 0.87, respectively), though alpha was better for the FSS-7 over time. Furthermore, it is noted that the FSS-7 is suggested as an outcome for research purposes. As the aim of this study was to provide data required to calculate reliable change within clinical settings, and the first two items of the FSS provide important information for clinical rehabilitation, the FSS-9 was retained.

The consensus statement also noted that data required to calculate minimum clinically reliable change on the FSS are not available in stroke. The issue of minimum clinically reliable differences in outcomes acknowledges that a statistically non-significant outcome does not automatically imply that a treatment has not been clinically effective, particularly as small sample sizes and measurement variability in this area of research can influence statistical results (Batterham and Hopkins 2006). Also noted by Page (2014), statistically significant differences (or lack thereof) may not always result in an appropriate change in clinical practice. Clinical interpretation of research using calculation of clinically reliable change to describe treatment outcomes is important because of its more direct translation to clinical decision making. As such, Page recommends that authors provide results relative to clinically meaningful differences or clinically reliable changes (Page 2014). The data required to calculate clinically reliable changes in post-stroke fatigue would support clinical decision making and measures of effectiveness for interventions designed to reduce fatigue.

The aim of this study was to describe fatigue as measured by the FSS at 1 month and 12 months post stroke using data from the population-based fifth Auckland Region Community Outcomes of Stroke Study (ARCOS-V), compare fatigue progression on the basis of the FSS by stroke subtype over time, and provide tabulated data for clinicians to determine if changes in fatigue in their own patients are clinically reliable by stroke type (i.e. ischaemic stroke (IS), intracerebral haemorrhage (ICH) and subarachnoid haemorrhage (SAH)), and IS subtype (using Trial of Org 10172 in Acute Stroke Treatment (TOAST) classifications; Adams et al. 1993).

Methods

ARCOS-V is the fifth in a series of prospective population-based studies of stroke incidence, prevalence, and outcomes in the Auckland region of New Zealand. The methods are described in detail elsewhere (Krishnamurthi et al. 2014). Each study included all incident strokes in residents of the Auckland region aged ≥16 years, with complete case ascertainment assured through multiple overlapping sources of information on all new hospitalised or non-hospitalised cases. The ARCOS-V study follows the recommendations for ideal stroke studies presented by Sudlow and Warlow (1996).

All public hospital admissions data were searched daily, with weekly checks of hospital discharge and outpatient clinic data. Regular checks of private hospitals, rest homes, and community health services were also conducted. New Zealand Ministry of Health data of all fatal and nonfatal stroke cases in the study population and residents meeting the inclusion criteria but who had their stroke while outside the Auckland region were also included.

Stroke was defined according to the World Health Organization as ‘rapidly developing clinical signs of focal (or global) disturbance of cerebral function, lasting more than 24 hours or leading to death, with no apparent cause other than of vascular origin’ (Aho et al. 1980, p. 114). A diagnostic review committee comprising neurologists confirmed the diagnosis and classification of all stroke cases using medical history, hospital discharge summaries, clinical and laboratory findings, imaging (including vascular and cardiac), or necropsy results when available. Cases without imaging or pathological necropsy confirmation of subtype were classified as stroke of undetermined type. Stroke type (i.e. IS, ICH, and SAH) was recorded, and those with IS subtypes were subclassified using the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classifications (i.e. large artery atherosclerosis (LAA), cardioembolism, small-vessel occlusion, stroke of other determined aetiology, and stroke of undetermined aetiology) (Adams et al. 1993).

The analyses described in this paper included all ARCOS-V cases who: (1) provided written informed consent for follow-up; and (2) had outcome data from 1 month assessments, the majority of whom also had 12 month data (see Fig. 1). Individuals who consented to the study but who were too unwell to complete the baseline assessment were approached at 12 months to determine willingness and ability to participate in the follow-up assessment. Each assessment took approximately 1 hour to complete and was conducted via telephone by a trained research assistant. Assessments reported here occurred 1 month and 12 months post stroke.

Fig. 1.

Flow of participants into the study sample.

Results

Overall, 358 of the original ARCOS-V sample of 2515 participants had FSS data at 1 month, and 338 had FSS data at 12 months (Fig. 1). Descriptive data for the sample at 1 month are provided in Table 1. These are contrasted with those who did not consent or provide data at 1 month. As can be seen in the table, those who participated were significantly younger and more likely to be male, to be European, and to have experienced an IS. The two samples did not differ in terms of stroke severity (NIHSS) or marital status.

Moderate-to-severe fatigue was present in 60.3% of participants at 1 month and 51.1% of participants at 12 months. At 1 month post stroke, people with ICH had the highest FSS scores; the stroke categories with the next highest scores were SAH and then IS (Table 2). Within IS subtypes, those with small vessel occlusion had the highest initial levels of fatigue. The greatest reductions in FSS mean scores between 1 and 12 months seen in Table 2 were for IS of other aetiology and SAH. Increases in fatigue over time were found for those with IS from undetermined aetiology (mean increase 0.01) and small vessel occlusion (mean increase 0.68), whereas all other groups had reductions. Within-subject improvements in fatigue for those with SAH reached statistical significance (mean reduction 0.81). Although within-subject change for IS also reached significance, the group mean only reduced by 0.26. As can be seen in Table 2, of those comparisons that were statistically significant, only that for SAH was quite large, whereas those for total stroke and all IS were small-to-moderate. Though not statistically significant, the effect size for LAA was large, and that for undetermined IS was quite large.

We observed overall significant within-subject improvements for the following items: ‘My motivation is lower when I am fatigued’, ‘My fatigue prevents sustained physical functioning’, ‘Fatigue is among my most disabling symptoms’, and ‘Fatigue interferes with my work, family, or social life’ (Table 3). None of the other items on the FSS showed detectable changes over time.

The greatest proportion of individuals in each stroke type and IS subtype (72.7%–90%) had no reliable change in fatigue (Table 4). Of those who did experience reliable change, more people had reliable reductions in fatigue (2–14 fold) than reliable increases in fatigue. The exception to this was for those with IS of undetermined aetiology, where the proportion of those whose scores reliably improved was the same as the proportion whose scores deteriorated. Reliable change was not significantly correlated with baseline stroke severity as measured by the NIHSS (r = −0.140, P > 0.05).

Discussion

The findings of this study provide the data needed to calculate minimum clinically reliable differences on the recommended FSS post stroke, thereby filling a gap in knowledge noted by a recent consensus statement (English et al. 2024). Between 1 and 12 months post stroke, the majority of patients (72%–90%) did not experience any reliable change in fatigue, 10%–20% experienced improvements in fatigue, and only 0–11% experienced reliable increases in fatigue. Although most individuals experienced no clinically reliable change in fatigue, the proportions of individuals who did experience reliable increases or decreases varied widely across stroke type and IS subtypes. The level of fatigue experienced within the sample warrants intervention, with over half of participants experienced moderate-to-severe fatigue at both 1 month and 12 months.

Examinations of changes in post-stroke fatigue should be considered by stroke type/IS subtype. This is supported by our observation that mean levels of fatigue and changes in fatigue over time varied by stroke type and IS subtype. Statistically significant reductions in FSS were only found for IS of other aetiology and SAH, and not any other stroke type/subtype. Of those that were statistically significant, only the SAH group had a large effect size. Though not statistically significant, the effect size for LAA was large, and the effect size for undetermined IS was quite large. These discrepancies were likely affected by different sample sizes and lack of power.

Use of minimum clinically reliable differences acknowledges that a statistically non-significant outcome does not automatically mean a treatment has not been clinically effective, particularly as research in this area often has small sample sizes and large measurement variability (Batterham and Hopkins 2006). The findings presented here support this assertion with individuals experiencing clinically significant change where statistically significant changes were absent at a group level. The absence of statistically significant changes in a group may mask the effectiveness of an intervention present in a sub-sample (Page 2014). The calculation of a ‘clinically reliable change’ to describe treatment outcomes is important because of its more direct translation to clinical decision making. Reliable change should therefore be reported in clinical trials (Page 2014).

The data presented here will allow clinicians and researchers who examine this issue to calculate clinically reliable change by stroke type and by IS subtype. For example, if a clinician were working with an individual who had experienced an SAH who had been assessed with the FSS and received scores of 4.9 at 1 month and 3.88 at 12 months post stroke and wanted to know if that drop was clinically reliable, this would be calculated as (X2 − X1) − (M2 − M1)/SDD or ((3.88 − 4.9) − (3.89 − 4.7))/1.40 = −0.157 where M1, M2, and the SDD in Table 2 are 4.7, 3.89, and 1.40. Thus, although the individual has experienced a drop in FSS score (−1.02), it does not meet criteria (±1.96) for a reliable reduction. For clinicians, looking at Table 4, this drop of −1.02 falls between the critical values for a reliable increase/decrease, so would not be considered a reliable change. In applying the data, it must be remembered that the degree of change presented here is for a period from 1 month to 12 months post stroke. A limitation to the clinical utility of the data is that only these two measurement points are presented, and a clinician might not have access to FSS scores for these timepoints. It should be noted that the extend of change within this sample is likely to exceed that for clinical data for shorter timeframes. We therefore suggest caution if applying these data to clinical cases that have data from differing timeframes. Recent work on adaption of reliable change calculations to allow for more flexibility in applications related to timing of the data is underway (Helmich 2024).

Conclusion

Although most participants had no clinically reliable change in fatigue between 1 and 12 months post stroke, statistically significant reductions in FSS were identified in patients who had experienced IS and SAH. Across stroke type and IS subtypes, the majority of those who did experience reliable change had reductions in fatigue over time. There was wide variability in the proportions of individuals who experienced reductions and increases in fatigue, suggesting that any examination of changes in post-stroke fatigue should be considered by stroke type/IS subtype.