Background
Fatigue is the state of weariness that may result from excessive physical and mental effort and psychological distress [
1]. In general, fatigue is classified as acute, prolonged or chronic by the duration of the symptom, and chronic fatigue (fatigue > 6 months) can be considered a medical problem [
2]. Among types of chronic fatigue, chronic fatigue syndrome (CFS) is the most debilitating illness, characterized by post exertional malaise (PEM), sleep disorder, cognitive dysfunction, orthostatic intolerance and a seven-fold higher suicide rate [
3]. Studies have found that up to 30–50% of the general population experienced fatigue [
4,
5], approximately 10% experienced chronic fatigue [
6], and a recent review study reported that 1% experienced CFS [
7].
On the other hand, the recognition of fatigue has been expanded to various dimensions due to the complexity of fatigue [
2,
8]. Fatigue could be a physiological response as well as a disorder; however, there is no objective biological parameter to assess fatigue, which raises problems in the diagnosis and management of fatigue [
9]. Accordingly, various fatigue measurement tools have been introduced to assess fatigability [
2]. To date, diverse patient-reported outcome (PRO) measurements have been developed and used to assess fatigue status in clinics. Some are fatigue-nonspecific instruments, such as the 36-item Short Form Health Survey (SF-36) [
10], Clinical Global Impression (CGI) [
11], and Sickness Impact Profile-8 (SIP-8) [
12], while fatigue-specific tools include the Checklist Individual Strength (CIS) scale [
13], Chalder Fatigue Questionnaire (CFQ) [
14], and Multidimensional Fatigue Inventory (MFI) [
15].
We recently reviewed the trend of fatigue-assessment instrument application in clinical trials for CFS and found that both MFI and CFQ were the most commonly employed instruments [
16]. These instruments reflect the clinical features of chronic fatigue divided into physical and mental domains and further into five dimensions: general, physical, mental, reduced activity, and reduced motivation [
14,
15]. In fact, several clinical studies have adopted MFI and CFQ as primary assessment tools for fatigue and CFS patients [
17‐
21]. The selection of an appropriate assessment tool is crucial in fatigue-related clinical studies, while the Korean version of the MFI (MFI-K) and modified CFQ (mKCFQ) were designed for assessing the therapeutic process of Koreans complaining of fatigue and CFS, and each of them was clinically validated [
22,
23]. However, no studies have been conducted comparing the characteristics of tools, particularly the most commonly used tools: MFI and CFQ.
The present study aims to evaluate the correlations and find the similarities and differences between the two instruments in identifying those with fatigue to determine their optimal usefulness.
Discussion
The CFQ and MFI were initially developed to measure the severity of various fatigue types, including CFS, in England in 1993 [
14] and to measure cancer-related fatigue in the Netherlands in 1995 [
15]. These two PRO-based instruments have been translated into as many languages and are commonly used for assessments of the extent and severity of fatigue in patient and nonpatient populations [
27]. This study analyzed the similarities and differences of fatigue-assessment data from the Korean versions of MFI (MFI-K) and modified CFQ (mKCFQ) [
22,
23] and aimed to produce important information regarding clinical choice and applications.
To evenly compare these instrument-derived fatigue scores, we converted mKCFQ (maximum 99 points) scores to 100 point-fitted scores to ensure that it was assessed on the same scale as the MFI-K score (maximum 100 points). As we expected, the total MFI-K and mKCFQ scores were significantly correlated (
r = 75%,
p < 0.001, Table
2; Fig.
2), but they showed some differences regarding severity groups and dimensions of fatigue for a single group comprising 70 participants. The MFI-K score was significantly higher than the mKCFQ score by 1.3-fold for 70 participants (45.8 ± 11.3 vs. 36.1 ± 16.2,
p < 0.001), 40 male and 30 female participants (Table
2). This high-score result in MFI-K mainly came from the score difference (17.6 points) for the ‘mild’-fatigue group (40.4 ± 9.7 vs. 22.8 ± 10.7, Table
3), which is shown clearly in the Box-Whisker plot (Fig.
3a). These results may explain the possibility of a tendency of MFI-K to be scored high, especially for the general population with a very low level of fatigue. In fact, this pattern of the MFI-K tool was seen in its initial validation study, which showed the comparison between the MFI-K total score (maximum 100) and VAS (maximum 100). In this study, compared to the ‘severe’ (5.5 points) and ‘moderate’ (19.9 points) groups, the ‘mild’ fatigue group (24.5 points) had the largest difference [
22]. This can be due to MFI-K having more number of questions than mKCFQ, thus MFI-K may be sensitive to certain aspects of fatigue questions.
In general, fatigue is grouped into two categories, so-called physical and mental fatigue, in most studies [
28,
29]. The two instruments included in this study are also structured to distinguish between physical and mental fatigue assessments, and the responses to physical-mental fatigue-focusing questions were well differentiated from each other [
22,
23,
30]. When we analyzed the correlation between the MFI-K and mKCFQ in the aspect of the physical and mental fatigue categories, each of the ‘physical’ and ‘mental’ fatigue scores correlated well between the two instruments, while the physical fatigue dimensions (‘general’, ‘physical’, and ‘activity’) were more highly correlated than the mental fatigue dimensions (‘mental’ and ‘motivation’, Table
2). We confirmed this feature in both box-whisker plots (Fig.
3b, c) and EDM analysis (Fig.
4). This would reflect the similarity of ‘physical’ fatigue assessments between two instruments but further dissimilarity of ‘mental’ fatigue assessments. In fact, previous studies using MFI reported similar results, showing a higher correlation of ‘general’ and ‘physical’ fatigue scores with VAS-based overall fatigue levels than scores of ‘activity’ and ‘motivation’ [
15,
31]. These results may mean that the mKCFQ further sharply differentiates ‘mental’ fatigue severity from low-to-severe levels compared to the MFI-K.
Among the three fatigue groups and five dimensions, ‘general’ fatigue (
r = 91%, MFI-K vs. mKCFQ = 15.4 ± 2.3 vs. 14.7 ± 2.6) and ‘reduced activity’ (
r = 19%, MFI-K vs. mKCFQ = 9.5 ± 2.1 vs. 13.3 ± 2.4) showed the highest and lowest correlations in the ‘severe’ fatigue group (Table
3). One possible reason might be the misunderstanding of questions such as “I feel very active”, “I think I do a lot in a day”, and “I feel like doing all sorts of nice things” in the MFI-K, which may seem less likely related to fatigue-related symptoms but rather more general-behavioral questions (Additional file
1: Table S1). This is reflected in the two-dimensional matrix in Fig.
4, which, unlike the others, reduced ‘activity’ in the mental dimension (Fig.
4). From the results above, we suspect that the questions for reduced ‘activity’ and ‘motivation’ in MFI-K possibly misled the responses. In fact, a study of the reliability and validity of the MFI-K using outpatients of the Department of Family Medicine showed very low correlations with the VAS (‘reduced activity’ 0.087, and ‘motivation’ 0.159) [
22]. The present study showed a relatively better ability of the mKCFQ to appropriately assess the severity of fatigue. In fact, CFQ has been criticized as an operational method that asks one to choose among “less than usual”, “no more than usual”, “more than usual”, and “much more than usual”, which may lead the answers to the extreme end of the scale [
32]. Extreme scoring is likely to cause an inability to discriminate between different groups [
32]. Thus, the mKCFQ was designed on a 10-point Likert scale, which allows for the assessment of small fatigue differences among participants and changes after therapeutic interventions [
21,
23].
In summary, the MFI-K and mKCFQ are likely to be sensitive in discriminating the ‘severe’ fatigue group, while the mKCFQ seems to be more suitable for participants with low levels of fatigue symptoms. In addition, clinicians or researchers may need to be aware of the low sensitivity of ‘reduced activity’ and ‘motivation’ relative to ‘general’ and ‘physical’ fatigue. For a certain illness such as CFS, the optimal fatigue scale should accurately identify specifications with multiple dimensions of fatigue in the process of both the diagnosis and assessment of therapeutics [
33]. The MFI and CFQ and their Korean versions (MFI-K and mKCFQ) have been adapted in clinics and clinical trials for CFS patients [
21,
23]; however, we still need to increase the specificity to differentiate patients by adding specific fatigue dimensions for postexertion malaise (PEM), one of the primary symptoms in patients with CFS [
33]. This study has some limitations, such as a relatively small number of participants, especially for the ‘severe’ fatigue group and the inclusion of exclusively university personnel. This study was performed in Korea using Korean versions of fatigue scales; thus, limits in generalization with other languages. The correlations of the two instruments were compared based on the scores of the total and each domain. Further studies are needed among larger-scale populations with diverse fatigue severities.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.