Classifying post-stroke fatigue: Optimal cut-off on the Fatigue Assessment Scale

Highlights

• We identified ≥ 24 as the optimal Fatigue Assessment Scale cut-off in stroke.

• Our data-driven approach used 2 independent, criterion standard-based studies.

• It is possible that a shorter 3-item FAS can be used as a valid screen for fatigue.

Abstract

Objective

Post-stroke fatigue is common and has debilitating effects on independence and quality of life. The Fatigue Assessment Scale (FAS) is a valid screening tool for fatigue after stroke, but there is no established cut-off. We sought to identify the optimal cut-off for classifying post-stroke fatigue on the FAS.

Methods

In retrospective analysis of two independent datasets (the ‘2015’ and ‘2007’ studies), we evaluated the predictive validity of FAS score against a case definition of fatigue (the criterion standard). Area under the curve (AUC) and sensitivity and specificity at the optimal cut-off were established in the larger 2015 dataset (n = 126), and then independently validated in the 2007 dataset (n = 52).

Results

In the 2015 dataset, AUC was 0.78 (95% CI 0.70–0.86), with the optimal ≥ 24 cut-off giving a sensitivity of 0.82 and specificity of 0.66. The 2007 dataset had an AUC of 0.83 (95% CI 0.71–0.94), and applying the ≥ 24 cut-off gave a sensitivity of 0.84 and specificity of 0.67. Post-hoc analysis of the 2015 dataset revealed that using only the 3 most predictive FAS items together (‘FAS-3’) also yielded good validity: AUC 0.81 (95% CI 0.73–0.89), with sensitivity of 0.83 and specificity of 0.75 at the optimal ≥ 8 cut-off.

Conclusion

We propose ≥ 24 as a cut-off for classifying post-stroke fatigue on the FAS. While further validation work is needed, this is a positive step towards a coherent approach to reporting fatigue prevalence using the FAS.

Introduction

Post-stroke fatigue is a widely experienced obstruction for stroke survivors [1]. A recent meta-analysis contained a pooled prevalence estimate of 50% [2]. Fatigue can be defined as a subjective lack of physical or mental energy (or both) that is perceived by the individual to interfere with usual or desired activities. Post-stroke fatigue is significantly related to poorer quality of life, even after adjusting for age, disability and depression [3]. It is important to patients, with 40% reporting fatigue as their worst or one of their worst symptoms [4]. Post-stroke fatigue decreases functional independence, limits social participation and is linked to increased mortality [5], [6].

Despite the high prevalence and importance of fatigue after stroke, assessment tools and procedures remain rudimentary. Most scales have been developed for multiple sclerosis or chronic fatigue syndrome. In post-stroke fatigue studies, the most widely used measure is the Fatigue Severity Scale [2]. Yet this scale focuses on severity of fatigue symptoms and their impact on other activities, rather than whether or not fatigue is present, and the first 2 items have problematic psychometric properties in stroke [7]. Only one attempt has been made to systematically evaluate fatigue scales for stroke [8], and the Fatigue Severity Scale was not included due to its poor face validity. All 4 tested scales were feasible to administer, with the Fatigue Assessment Scale (FAS) having the best test-retest reliability. The FAS also had the poorest internal consistency, but this may be considered a positive given that items cover both mental and physical aspects of fatigue. Scale validity has been difficult to establish, due to the lack of an accepted criterion standard for fatigue diagnosis. The 2007 case definition of fatigue, based on a structured interview and developed specifically for stroke survivors [9], overcomes this problem.

The FAS has promising face validity and reliability properties, but there is no consensus on what FAS score constitutes fatigue. In contrast, the Fatigue Severity Scale has a well-accepted cut-off that is used to classify fatigue, making prevalence comparisons between studies straightforward. In the current study, we aimed to identify an optimal cut-point on the FAS that best discriminates those with and without fatigue after stroke.