Psychometric properties of fatigue severity scale in Chinese systemic lupus erythematosus patients

Systemic lupus erythematosus (SLE) is a multisystem, autoimmune, inflammatory disorder presenting with symptoms from various organs, including joint, skin, kidney, brain, cardiovascular, lung, etc. [1]. In western countries and the United States, SLE prevalence was 4 to 250/100,000 in adult. In China, its prevalence was 97.5 to 100/100,000 [2‐5]. The peak SLE incidence occurs at age 20–29 years, followed by 30–39 years old for females, but in 70–74 years old for males [3], and the female-to-male ratio was about 10:1 in China [6‐8]. The most common symptom of SLE is fatigue, with about 51%~ 90% of SLE patients experiencing fatigue problem [9]. Fatigue can affect the daily lives of SLE patients [10, 11], and is also related to sleep disorders [12], pain, depression, and quality of life [13]. Fatigue is a subjective phenomenon without clinically validated definition [14], and it’s hard for patients with fatigue problem to participate in a long time investigation, so a reliable and simple instrument to measure fatigue problem of SLE is important, which would be conducive to interventions and outcomes monitoring.

Many fatigue instruments have been used in SLE patients. A study conducted a systematic review of the fatigue instruments used in SLE patients. 15 instruments were found, including Visual Analogue Scale [15], Fatigue Severity Scale (FSS) [16], Chalder Fatigue Scale [17], Multidimensional Fatigue Inventory (20 items) [18], etc. The study found that FSS was most widely used and recommended the use of FSS to measure fatigue for SLE patients. First developed and tested by Krupp in SLE and multiple sclerosis patients [16], FSS was used to measure the impact of fatigue on functions [9]. It is a unidimensional fatigue scale with 9 items. FSS has also been used in diseases other than SLE, such as cancer [19], chronic hepatitis C [20], central neural system disease [21], chronic neck pain [22], major depressive disorder [23], stroke [24, 25], as well as general population [26]. Though widely used in other countries, past properties test of FSS in SLE enrolled relatively small number of participants [27‐29], and the FSS hasn’t been tested in Chinese SLE patients. The properties of a scale are related to specific patient group and cultural factors, so the aim of our study was to test the psychometric properties of FSS in Chinese SLE.

FSS has been translated into Chinese version but hasn’t been applied in SLE patients. As fatigue is one of the mostly referred subjective symptoms in SLE, there is a need for a reliable instrument to assess this problem. This was the first study to test the psychometric properties of FSS in Chinese SLE patients. Our study showed that original 9-item FSS had acceptable psychometric properties, in accordance with findings from other language versions in different populations [28, 33, 34]. And the 7-item FSS also have good psychometric properties for Chinese SLE patients.

When more than 15% of the samples have either the lowest or the highest score, the instrument demonstrates floor and ceiling effects [35]. In our study, 4.5% of the samples had the lowest score and 4% had the highest score, indicating the ceiling and floor effects were minimal. However, Ferentinos’ study [36] demonstrated a ceiling effect of FSS in major depression subjects. Possible explanation may be that depressed patients are more sensitive to negative feelings and experiences so they are more likely to choose higher score of FSS. Such result indicated that the acceptability of the FSS differs in different patient group, which further confirmed the importance to test the psychometric properties of an instrument before applying it to a specific disease.

Exploratory factor analysis results were in agreement with previous studies using FSS in SLE patients [27‐29]. Only one factor was extracted in our sample, explaining 61.80% of the total variance. All the item loadings were higher than 0.40. The first and second item showed the lowest correlation with the rest of the items, in line with Lerdal’s report [26]. Ottonello’s study [37] reported that deletion of item 1 could improve the unidimensionality of the FSS in Italian MS patients. Lerdal’s study concluded that the removal of item 1 and item 2 could improve FSS’s psychometric properties in MS and stroke patients [14, 38], while Mill’s study [39] concluded that a 5-item FSS scale could measure the impact of the fatigue better than the original 9-item scale in MS patients. Wang’s study indicated that the deletion of item 2 and item 3 showed better psychometric properties in major depressive disorder participants [23]. When we deleted item 1 and item 2, 7 items explained 69.54% of the total variance.

Past studies demonstrated that the Cronbach’s Alpha of FSS ranged from 0.87 to 0.96 [34, 40‐43] in different language versions and different disease subgroups. The original FSS had good internal consistency. Our study found that the Cronbach’s Alpha showed a minor increase from 0.92 to 0.93 after the deletion of item 1 and item 2. Possible reason may be that the description of item 1 and item 2 can’t measure the severity of fatigue, and they lack the discrimination ability for healthy and disease-suffered groups since healthy people can also have such experiences. The test-retest reliability was good, indicating that it was a stable instrument to use at different interval.

The original FSS showed acceptable construct validity. Significant correlation between FSS score and depression score (r = 0.52, P < 0.01) was found, in line with past studies showing that the correlation coefficient ranged from 0.46 to 0.75 [40, 44, 45]. What’s more, correlation was also found between FSS score and score of Vitality (r = − 0.55, P < 0.01), in accordance with previous reports, in which the correlation coefficient ranged from − 0.32 to − 0.72 [27, 34, 46]. After the deletion of the item 1 and 2, the concurrent validity seemed better since the correlation coefficient between 7-item FSS score and Vitality score increased (r = − 0.58, P < 0.01), while the correlation coefficient between 7-item FSS score and depression score remained the same (r = 0.52, P < 0.01) (Table 4).

Known-groups validity was tested by comparing fatigue difference between patients with and without depression. We hypothesized that patients with depression would show higher degree of fatigue. And our result verified such hypothesis, showing FSS’s suitable ability to reflect differences between two groups both for original FSS and 7-item FSS (Table 5).

Our study did have some limitations. Firstly, we only investigated SLE patients in one general hospital with a small number of male patients, and the female to male ratio was less than the reported ratio in Chinese SLE patients, so the representativeness of the samples may be reduced. Secondly, the overall disease activity of SLE was mild, so further investigation of SLE with different level of disease activity are necessary to increase study’s generalizability. Finally, depression can have some impacts on the fatigue problem of SLE patients. Our study didn’t exclude the depressive patients as depression and fatigue can be interactive in SLE patients and it may be impractical to distinguish these two symptoms in SLE patients.

In summary, 9-item FSS is a reliable instrument and can be used to assess fatigue problem in Chinese SLE. The 7-item FSS also demonstrated good psychometric properties. As the deletion of two items simplify the content of FSS, it may make patients more willing to participate in study. However, another study about whether the 7-item FSS is more easy to use from patients’ perspectives is needed.