- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
Rheumatoid arthritis (RA) is an autoimmune, systemic, inflammatory condition causing pain, disability, and psychological distress (1). Fatigue is experienced by up to 90% of patients with RA and its causality is likely to be multidimensional (2–4). Fatigue has far-ranging consequences on patients' lives and is an important outcome for many patients (5–8), but is currently not among the 7 internationally agreed core outcome measures in RA clinical trials (9). However, the Outcome Measures in Rheumatology Clinical Trials (OMERACT) group has been discussing the importance of fatigue as a key outcome and has given a mandate to pursue further work on identifying valid measures (10).
If treatments are to be developed and tested, fatigue needs to be accurately assessed using scales with adequate validation properties of comprehensiveness, accuracy, biologic sense, reliability, and sensitivity to change. In common with many other chronic illnesses, there is no agreed definition of fatigue in RA. However, it is widely accepted by clinicians that there is a subjective element to RA fatigue that goes beyond physiologic muscle fatigue. Qualitative studies suggest that RA fatigue may incorporate not only physical but also cognitive and emotional elements (2, 5). The nature of fatigue as experienced by patients with RA may be different from that in other long-term conditions (5); therefore the application of generic scales, or the creation of new scales without patient involvement, may have limited validity. Generic scales may contain items that in RA could reflect inflammatory disease or disability rather than fatigue (as seen in some generic depression scales) (11).
The use of inappropriate or unvalidated scales for outcomes can result in unreliable or misleading results. The Association of Rheumatology Health Professionals has recently reviewed a wide range of arthritis measurement scales, but this extensive and valuable resource was limited by space constraints and was not intended to examine the specific validity of scales measuring RA fatigue (12). The goal of the present study was therefore to systematically identify scales that have been used to measure fatigue in patients with RA in published studies, and to examine the evidence of the scales' validity to measure RA fatigue against recognized criteria (12–14).
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
The searches identified abstracts from 166 initial articles containing the search terms, 59 of which appeared likely to provide separately identifiable RA fatigue data in the full article. Two further RA fatigue articles known to the authors were not identified by the searches (one used a scale acronym rather than the word scale, one did not mention fatigue in the abstract), giving a total of 61 index articles. Upon review of the full articles, 50 of 61 index articles reported identifiable RA fatigue data. Twenty-three different fatigue scales were used, with some studies utilizing more than 1 scale, resulting in 71 occasions on which fatigue was measured (Table 2). Validation references were cited on only 35 occasions (49%). Overall, 118 articles were reviewed in the search for validation data, but only articles contributing substantial evidence are reported here.
Table 2. Fatigue scales identified as used in rheumatoid arthritis (RA)*
|Scale and subscale||Description||References reviewed||RA studies|
|Binary question: fatigue||Yes/no||16||1|| |
|Binary question: tiredness||Yes/no||17||1|| |
|CFS||11 items, 4 points||22,23||1|| |
|CIS/SF||8 items, 7 points||24–27|| ||1|
|CFI||16 VAS (mean)||28||1|| |
|CIFI||11-point scale + NHP fatigue items||29||1|| |
|Fatigue hours||Unknown||18|| ||1|
|FTC||10 descriptors, unobtainable||30||1|| |
|Five Items Modeled After Tack||Mean of 5 items||19|| ||1|
|FACIT-F||13 items, rated 0–4||51–54|| ||1|
|Morning fatigue||Lack vigor/fatigued on waking||20||1|| |
|MAF||15 items plus change question||44–50||5||3|
|MFI||5 scales of 4 items, 5 points||31–33||2|| |
|NIH ACTRE||Log every 30 minutes × 48 hours (0–4)||30,41||1||1|
|NHP: energy subscale||3 weighted items||23,34–39||6|| |
|Numerical Rating Scale: degree||0–5||21||1|| |
|Numerical Rating Scale: impact||0–10||6||1|| |
|Ordinal degree scales||E.g., none, mild, moderate, severe||6,55,56||3|| |
|POMS: fatigue/inertia subscale||7 items||44–46,57–60||3||3|
|Psychasthenia||10 items||40||1|| |
|SF-36: vitality subscale||4 items||23,52,61–75,80,87,89,90||16||5|
|Time to onset of fatigue||Hours||42,43||1||1|
|VAS||See Table 5||3,23,74–97||23||3|
Table 5. Visual analog scales (VAS) currently used to measure fatigue in patients with rheumatoid arthritis
|VAS||Start descriptor||End descriptor||Timescale||Length||Scoring||Comment in article|
|Lwin (23), Gilboe (74, 75)||No fatigue||Fatigue as bad as it could be||Month||100 mm||0–100|| |
|Currey (76)||Fatigue is no problem||Fatigue is a major problem||Week||10 cm||0–10||Unusual fatigue or tiredness|
|Tack (77)||No fatigue||Fatigue as bad as it could be||Week||100 mm||0–100|| |
|Tack (77)||No distress||Distress as bad as it could be||Week||100 mm||0–100||Fatigue distress|
|Scharloo (78)||No tiredness||Very severe tiredness||Week||10 cm||0–100|| |
|Mengshoel (79)||No fatigue||Total exhaustion||Week||100 mm||0–100|| |
|Kvien (80)||Fatigue no problem||Fatigue major problem||Month||100 mm||0–100|| |
|Crosby (81)||No fatigue||Extremely fatigued||Evening before||10 cm||0–100%||Vertical line|
|Wolfe (3, 82, 83)||Fatigue no problem||Fatigue major problem||Week||*||0–3||Fatigue or tiredness|
|Riemsma (84, 85)||Not tired at all||Very tired||Week||*||0–100|| |
|Barlow (86)||No fatigue||Fatigue as bad as it could be||*||10 cm||0–10|| |
|Hagen (87)||Fatigue is no problem||Fatigue is a major problem||*||100 mm|| || |
|Brekke (89, 90)||*||*||*||100 mm||0–100||Fatigue|
|Gudbjornsson (91)||*||*||Time of investigation||*||10 grades||Degree of fatigue|
|Heiberg (92)||*||*||*||100 mm||0–100||Fatigue|
|Jensen (93)||*||*||*||100 mm||0–100||Fatigue|
|Uhlig (94)||*||*||*||100 mm||0–100||Fatigue|
|Chiang (95)||*||*||*||15 cm||0–150||Degree of fatigue|
|Gerber (96)||*||*||*||*||0–3|| |
Fatigue scales where limited evidence of validation for RA could be identified.
On systematic examination of the available validation data, for 7 of the 23 scales being used to measure fatigue in RA, either few validation studies could be identified, evidence of validation was limited, or the scales had been designed for use in other populations and did not perform well in RA (Table 3). Seven scales had been created for single RA studies and did not appear to have previously been validated (Table 3). Binary questions can only indicate the presence or absence of fatigue, although they may be useful as a screening question (16, 17). A measure of fatigue hours was sensitive to change in a randomized trial of nonsteroidal antiinflammatory drugs (NSAIDs), but the article neither described the question nor provided validity or reliability data (18). The Five Items Modeled after Tack scale showed evidence of sensitivity to change in a cognitive-behavioral therapy (CBT) intervention, but no criterion or reliability data were provided (19). The Morning Fatigue scale (20) may have reflected the patient's experience of waking unrefreshed (5), but no validation data were provided. Numerical rating scales showed some evidence of construct validity, but no data were identified on criterion validity, reliability, or sensitivity (6, 21).
Table 3. Scales currently used to measure fatigue in rheumatoid arthritis (RA): little or limited evidence of validity*
|Scale and subscale||Face validity||Content validity||Criterion validity, gold standard||Construct validity||Reliability||Sensitivity||Feasibility|
|Source||Inclusive||No misleading items||Converge||Diverge||Internal consistency||Stability|
|Created for single study|| || || || || || || || || || || |
| Binary: fatigue||3||1||3||3||0||0||0||NA||0||0||3|
| Binary: tired||1||0||1||1||0||0||1||NA||0||0||3|
| Fatigue hours||0||0||0||0||0||0||0||NA||0||2||3|
| 5 Items after Tack||1||1||2||2||0||1||0||0||0||2||2|
| Morning fatigue||1||0||2||1||0||0||1||NA||0||0||3|
| Numerical degree||3||0||3||3||0||1||0||NA||0||0||3|
| Numerical impact||3||0||1||3||0||1||0||NA||0||0||3|
|Scales applied in RA|| || || || || || || || || || || |
| CFS|| || || || || || || || || || || |
| CFI total||2||2||3||2||0||1.5||0||1.5||0||0||2|
| CFI chronicity||3||2||3||3||0||0||0||1.5||0||0||0|
| CFI interference||1||2||1||1||0||0||0||1.5||0||0||0|
| MFI|| || || || || || || || || || || |
| NHP: energy||2||2||1||2||0||2||1.5||1||1.5||0||2|
|RA-specific scales|| || || || || || || || || || || |
| NIH ACTRE||1||0||1||3||0||1||0||0||0||1||1|
| Time to onset||1||1||1||1||0||0||0||NA||1||1||3|
Eight generic scales used to measure fatigue had limited data on validation in RA (Table 3). Some scales had been designed to measure cognition or personality. For example, 3 of the 11 items on the Chalder Fatigue Scale address mental clarity (22), and the scale does not differentiate between patients with RA and controls (23). Three of the 8 items on the Checklist of Individual Strengths Subjective Fatigue subscale (CIS/SF) may reflect inflammation rather than fatigue (feeling fit, in good shape, in bad condition) (24–26). The CIS/SF demonstrated sensitivity to change in its single study of patients with RA undergoing CBT (27) but it has not been tested for reliability or construct or criterion validity in RA. The Chronic Fatigue Index was developed from interviews of patients with chronic disease and was reviewed by women with RA, showing content validity. However, it was not retested after item reduction, and little evidence of testing for sensitivity or construct or criterion validity could be identified (28). The Composite Index of Fatigue Impairment is an 11-point numerical rating scale plus the Nottingham Health Profile (NHP) energy subscale, but how these are combined was not explained (29), and the validation reference that was cited was inappropriate. The Feeling Tone Checklist (FTC) was validated in healthy airforce personnel in 1956, which limits its applicability to RA fatigue, and the single RA study that tested the FTC alongside an activity record did not report any data (validation reference and scale unobtainable) (30).
Generic fatigue scales can include items that might lead to contamination in RA from outcomes such as disability. The Multidimensional Fatigue Inventory contains such items (“Physically I feel only able to do a little,” “Physically I am in bad condition”) (31). It does not differentiate between individuals with RA and those with ankylosing spondylitis on 4 of its 5 scales (32), nor between patients with RA and healthy controls on 2 of its scales (33). The NHP is a well-validated tool for surveying population health (34) and although RA studies have demonstrated that the energy subscale has construct validity and stability and differentiates between persons with RA and other populations, sensitivity to change and internal consistency have not been thoroughly examined in RA (23, 35–39). The 3-item energy subscale contains 1 item on fatigue, 1 item on energy (an absence of fatigue may not translate to the presence of energy), and 1 item on effort, which in RA may be the result of disability. The Psychasthenia scale is a personality scale and only 3 of 10 items assess physical fatigue. Although it differentiates between patients with RA and those with fibromyalgia, no other validation data could be identified (40).
Two RA-specific scales had limited data on validation. The National Institutes of Health Activity Record was developed to measure outcomes of an energy conservation program and was reported as having no standard scoring system and did not show sensitivity to change (30, 41). The time to onset of fatigue question showed that fatigue did not improve in an NSAID trial, but little information on construct or criterion validity could be found (42, 43).
Only 2 of these 17 scales had evidence for stability in RA or had been tested against another fatigue measure, and only 5 have been used in RA intervention studies. On the limited evidence available to date, it seems uncertain whether the 17 scales in Table 3 would be the first choice for providing robust, valid, accurate, and sensitive measures of fatigue in patients with RA.
Scales with reasonable evidence of validation for measuring fatigue in RA.
Multidimensional Assessment of Fatigue scale.
The Multidimensional Assessment of Fatigue scale (MAF) comprises 16 questions concerning the quantity, degree, distress, impact, and timing of fatigue (44). Questions 1–15 form the final score (Global Fatigue Index, 0–50) whereas question 16 concerns change over the past week. Questions 1–14 are 10-point items, whereas 15 and 16 are 4-point items. The MAF is an RA-specific revision of the Piper Fatigue Scale developed in oncology, giving it face and content validity (Table 4), and it has been tested against another fatigue scale (44, 45). Construct validity was demonstrated through moderate convergence with disease activity and mood (0.45–0.54) (44). Higher MAF scores were related to increased depression and reduced sleep, with 61% of variance in fatigue explained by disease activity, sex, and psychosocial status (44, 46). Higher MAF scores differentiated between patients with RA with and without previous depression, between different levels of disease activity, and between patients and controls (44, 47–49). Reliability was reflected by good internal consistency (interitem correlations 0.53–0.83, Cronbach's alpha 0.91–0.96) (44, 46–48). Sensitivity to change was demonstrated following drug therapy (50) and after exercise, where the MAF showed initial worsening of fatigue followed by improvement beyond baseline and showed a difference between individuals with low levels of exercise and those with high levels of exercise (45). Although designed to be scored as a global fatigue scale, there is some evidence for the validity of the individual sections of the MAF (44, 45).
Table 4. Scales currently used to measure fatigue in rheumatoid arthritis: reasonable or good evidence of validity*
|Scale||Face validity||Content validity||Criterion validity, gold standard||Construct validity||Reliability||Sensitivity||Feasibility|
|Source||Inclusive||No misleading items||Converge||Diverge||Internal consistency||Stability|
|MAF: Global Fatigue Index||3||2||2.5||3||2||2.5||2.5||3||1||3||3|
|Ordinal scales: best scores||3||0||3||3||0||1||2||NA||2||3||3|
| Stone (ref.55): not at all–extremely||3||0||3||3||0||1||2||NA||2||3||3|
| Katz (ref.6): none versus severe†||2||0||3||3||0||1||0||NA||0||0||3|
| Pinals (ref.56): none–severe||2||0||3||3||0||1||0||NA||0||0||3|
|SF36|| || || || || || || || || || || |
| Vitality (month)||2||1||2||2||2||2||2||3||0||3||3|
| Vitality (week)||2||1||2||2||0||2||0||2||0||1||3|
Functional Assessment of Chronic Illness Therapy fatigue scale.
The index article using the Functional Assessment of Chronic Illness Therapy fatigue scale (FACIT) in RA did not report fatigue data (51). The article reporting RA validation was published shortly after the final search date but is nonetheless reviewed here (52). The FACIT-F is a 13-item scale originally developed to measure fatigue in patients with cancer (53, 54). In RA, the FACIT-F has demonstrated convergent validity with disease activity, good internal consistency (Cronbach's alpha 0.86–0.87), and evidence of sensitivity to change (effect size 0.19–1.13) (52). However, no information on divergent validity or stability in RA has been reported. The source of the scale items was oncology patients, and in RA several items may be confounded by disability (e.g., needing help to do usual activities). One item measures energy rather than fatigue and items applicable to patients with cancer may hold less relevance for patients with RA. For example, feeling too tired to eat is not reported in qualitative RA fatigue studies (2, 5).
Three studies measured RA fatigue using ordinal scales such as “none” to “very severe” with responses ranging from 4 to 7 points (6, 55, 56). Overall, ordinal scales showed reasonable content and construct validity (Table 4), differentiating between patients with RA with and without inflammation and showing associations with other symptoms, reduced perceived ability to cope with fatigue, and poor sleep (6, 55, 56). When measured 7 times per day for 7 days, the ordinal scale appeared stable and consistently demonstrated that fatigue was lowest at noon and worst in the evenings, supporting the ability of ordinal scales to capture variation in fatigue (55). However, there are no data from intervention studies.
Profile of Mood States.
The Profile of Mood States (POMS) (57) fatigue/inertia scale contains 7 items (worn out, listless, fatigued, exhausted, sluggish, weary, bushed). It was designed to measure mood, but may address some of the cognitive elements and overwhelming fatigue experienced by RA patients (5). The POMS had criterion validity in RA (MAF r = 0.84) (48), although little information on construct validity could be identified. The POMS differentiated between patients with RA with and without a fear of falling and had good internal consistency (Cronbach's alpha 0.88) (58). Three dance- or exercise-based interventions revealed change in POMS, although this only approached significance (45, 59, 60), whereas in one of these interventions the MAF showed significant change (45).
Short Form 36 vitality subscale.
The vitality subscale of the Short Form 36 (SF-36) comprises 4 items (full of life, energy, worn out, tired) with 6 responses from “all of the time” to “none of the time” (61). Many studies reported RA data (52, 61–75). Three studies reported that patients with RA had less vitality than healthy controls, suggesting good construct validity (23, 66, 67), but another study reported that patients with RA had more vitality than controls (63). Although the vitality scale was associated with measures of disease activity, correlations with erythrocyte sedimentation rate (ESR) varied (r = 0.18–0.34) and the vitality scale did not always reflect change in ESR (62, 65, 67). The SF-36 vitality subscale was shown to correlate strongly with mood in one study (r = 0.67) (65), suggesting that it may not differentiate well between fatigue and depression. The SF-36 demonstrated good internal consistency (Cronbach's alpha 0.84–0.88) and only a small ceiling and floor effect (Table 4) (63–65, 72). Sensitivity to change was demonstrated with biologic agents (effect size 0.25–1.52) (52, 68).
Visual analog scales.
Visual analog scales (VAS) with identifiable RA fatigue data were used in 22 index articles and in 4 articles obtained during the search for validation data (3, 23, 74–96). Validation references for a fatigue VAS were cited on only 4 of 26 occasions and referred to overviews of the ability of the VAS to measure pain (97). Only 10 of 26 VAS were described sufficiently enough for them to be fully reproduced and only 3 could be identified as completely identical for descriptors, timescale, and length (Table 5). Lack of standardization clearly limits comparison between studies.
Where the scales were described, VAS face and content validity were good (Table 6) except that information was rarely given as to whether patients were the source of the descriptors. Construct validity was reasonable, with evidence of convergent validity with pain (r = 0.31–0.8), poor sleep (r = 0.6), and disability (r = 0.33–0.41) (77, 79, 80, 88). VAS fatigue was moderately associated with low mood (r = 0.41–0.47) (77, 80, 84). VAS for fatigue discriminated between patients with RA and fibromyalgia, patients with RA and healthy controls, and patients with RA with and without pain (23, 76, 79, 91, 92). Overall, available evidence for reliability was limited, with one study showing stability for the VAS but not the SF-36, and another showing the opposite (89, 90). Of 3 intervention studies, one only reported combined osteoarthritis/RA data (86), and neither total hip replacement (88) nor an energy conservation program (96) showed change in RA fatigue. Three longitudinal studies demonstrated variation in a fatigue VAS over time (87, 89, 90).
Table 6. Validation of visual analog scales for measuring fatigue in rheumatoid arthritis: best scores for each version*
| ||Face validity||Content validity||Criterion validity, gold standard||Construct validity||Reliability||Sensitivity||Feasibility|
|Source||Inclusive||No misleading items||Converge||Diverge||Stability|
|Best scores overall||3||2||3||3||2||2.5||2.5||0||1.5||3|
|Brekke (89, 90)||0||0||0||0||1||0||0||0||1.5||0|
|Lwin (23), Gilboe (74, 75)||3||0||2||3||1||0||1||0||0||3|
|Riemsma (84, 85)||3||0||3||3||0||2.5||0||0||0||3|
|Wolfe (3, 82, 83)||2||0||3||1.5||1||2||1.5||0||0||2.5|
One article reporting use of a VAS to measure fatigue stated that for scanning purposes, the VAS was formatted as 21 boxes representing a 0–10 scale in 0.5 increments (ClinHAQ) (98). The reduction of a continuous VAS line to a tick-box scale of 21 discrete points may alter its measurement properties. The authors did not state in their earlier VAS studies whether the VAS was presented in box or standard format (3, 82, 83).
Some studies used several scales simultaneously. A comparison of the VAS, MAF, and SF-36 vitality subscale in a large RA cohort yielded mean scores of 4.5, 5, and 5.5 (out of 10), respectively (98). The scales correlated well with each other (r = 0.71–0.8) and correlated moderately with clinical measures (r = 0.5–0.63). The MAF had the smallest floor and ceiling effects at 0.01% and 0.2%, respectively (SF-36 0.4% and 3.4%, respectively; VAS 6.4% and 1.8%, respectively). The VAS had the most variability, with an SE of 0.032 (0.026 for the MAF and SF-36). However, the VAS was presented as 21 boxes, the MAF and SF-36 were rescaled, and associations with mood were not assessed.
Other studies reporting data from both a fatigue VAS and the SF-36 vitality subscale demonstrated slightly lower associations (r = −0.58 to −0.71) and suggested that the scales have different distributions and perform differently (80, 89, 90). The SF-36 vitality subscale was more strongly associated with social and mental health, and unlike the fatigue VAS varied with age. Although one longitudinal study suggested that the SF-36 and VAS change in similar manners (87), other studies reported that changes in fatigue were sometimes shown on the VAS but not the SF-36, and vice versa (89, 90). These differences may reflect the conceptual gap between measuring fatigue and vitality, because a low score on one may not necessarily reflect a high score on the other.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
The use of poorly validated outcome scales would limit the interpretation of study results, yet this systematic review of scales used to measure RA fatigue demonstrates that for 17 of the 23 scales used, only limited validation could be identified. Many scales had been created for use in a single study, whereas for many others, little evidence of attempts at validation in RA could be found. Some of these scales for measuring RA fatigue may be useful and appropriate, but this is difficult to assess in the absence of identifiable validation data. The MAF, SF-36, FACIT-F, ordinal scales, POMS, and VAS do show evidence of validation for measuring RA fatigue. However, even these 6 scales could benefit from further research, particularly concerning content validity for patients with RA, the inclusion of cognitive items, and sensitivity to change.
This study raises several important issues. First, if authors do not give a validation reference for a scale, or do not fully describe the scale, it is difficult for readers to assess or replicate their methodology. This is particularly problematic with the VAS where only 10 of 26 articles fully described the wording, timescale, and length. Although the combined data from many RA studies support reasonable validation for a fatigue VAS, there is no standardized VAS for RA fatigue, making comparison between studies difficult. The most common VAS wording was “problem,” but this infers a combination of severity and consequence rather than pure level of fatigue. There is a pressing need to develop and validate a standardized RA fatigue VAS.
Second, many of these scales were developed ahead of the current systematic, rigorous approach to reporting validation data, making it difficult to assess their validity. In particular, the inclusion of patient opinion in scale development is now considered crucial. The lack of sensitivity to change data is a major drawback and is complicated by the lack of knowledge about which interventions might reduce RA fatigue. A circular argument develops whereby it is difficult to validate scales for sensitivity until there are effective interventions, and it is difficult to prove efficacy of interventions without validated scales. In RA, it is possible that interventions such as exercise may not improve fatigue, but they may allow patients to do more without increasing fatigue, an additional complication. If future intervention studies could simultaneously measure other aspects of change, such as changing levels of activity or impact of fatigue, this might help clarify a difficult area.
Third, some generic scales could be open to contamination because in RA the wording may reflect problems arising from disability or inflammation rather than fatigue. Some scales include phrases that are not reflected in qualitative studies of the nature of RA fatigue (e.g., being too tired to eat) and most omit concepts that may be specific to RA fatigue (e.g., sudden, overwhelming onset or cognitive elements) (2, 5). Several scales measure energy or vitality, but an absence of fatigue does not necessarily mean the presence of energy, therefore such scales may not accurately reflect fatigue. Indeed, the POMS treats fatigue and vigor as different concepts and shows that they are only moderately associated (r = 0.43) and perform differently in their relationships with mood, suggesting that they are not opposite ends of a single concept (57).
This study has limitations, including the lack of an internationally agreed definition of RA fatigue, which is outside the scope of this study. Searching databases for RA and fatigue/tiredness produced an unmanageable data set; therefore, because the purpose of the study was to identify fatigue measures, the terms scale, questionnaire, inventory, and checklist were used. It is therefore possible that some RA fatigue studies that might have contributed validation data were missed. However, a second search was conducted specifically for the measurement of RA fatigue using the 6 scales identified as having stronger validation. The review scoring system was relatively arbitrary but the goal was to create a broad summary of available evidence, using criteria based on a recognized validation framework (12–14), systematically applied by 3 researchers first independently and then in discussion to arrive at agreement.
The treatment and self-management of fatigue in patients with RA is receiving increasing interest as its importance to patients becomes apparent, and this requires accurate measurement of fatigue. The present study helps clarify a difficult issue by identifying stronger validation in certain fatigue scales. However, it also indicates that further work is urgently required to develop and validate a standardized VAS, to consider the addition of cognitive fatigue items to the MAF, to compare questionnaire items with qualitative descriptions of RA fatigue to ensure they capture the essence of the concept in RA, and to provide more sensitivity to change data for several scales. A measure of the impact of fatigue, in addition to the presence or severity of fatigue, would be useful.