Relationship Between Objectively Assessed Physical Activity and Fatigue in Patients With Rheumatoid Arthritis: Inverse Correlation of Activity and Fatigue

Authors


Abstract

Objective

Fatigue is generally associated with low physical activity in patients with various chronic medical conditions. However, such an association has not been reported among patients with rheumatoid arthritis (RA). The objectives of this study were to investigate whether daily activity level is associated with fatigue in patients with RA, and whether pain, disability, coping, and/or cognition are associated with the level of daily activity.

Methods

Patients with RA who visited our outpatient clinic were recruited consecutively. Fatigue severity was measured using the Checklist Individual Strength (CIS20). Physical activity was measured for 14 consecutive days using an ankle-worn actometer. The daily activity level of each patient was calculated, and each patient was classified as having a low or high activity level with respect to the group average. Data were analyzed by linear regression.

Results

A total of 167 patients were included in the analysis; 25% had a low activity level and 75% had a high activity level. A regression analysis revealed that higher activity levels were associated with reduced fatigue (P = 0.008). The mean ± SD CIS fatigue score was 30.9 ± 12.3 among the patients with a high activity level and 35.7 ± 12.8 among the patients with a low activity level (P = 0.03). Pain, disability, coping, and cognition were not associated significantly with daily activity level.

Conclusion

Among patients with RA, a higher level of daily physical activity was associated with reduced levels of fatigue. This relationship was not explained by differences in sex, age, disease duration, pain, disability, or other fatigue-related factors.

INTRODUCTION

Fatigue is a common symptom among patients with rheumatic diseases, including rheumatoid arthritis (RA) ([1]). As many as 40% of patients with RA experience severe fatigue (defined as a Checklist Individual Strength [CIS] fatigue score ≥35), and fatigue often causes debilitating and restricted daily functioning ([2-6]).

In patients with RA, fatigue can be associated with pain, disability, depressive thoughts, anxiety, worrying, feelings of helplessness, reduced self-efficacy, sleep disturbances, and limitations in social functioning ([7-9]). Based on these findings, pain and disability, and not inflammation per se, and psychosocial factors may be important contributors to the presence and persistence of fatigue. In addition, it remains unclear which treatments are effective for treating fatigue in RA patients. This uncertainty can contribute to fatigue being neglected during health care visits. However, some randomized controlled trials (RCTs) suggest that both cognitive–behavioral therapy and physical exercise can be effective for treating fatigue in patients with RA ([10, 11]).

With respect to chronic fatigue syndrome (CFS), graded exercise training and cognitive–behavioral therapy are the only 2 interventions that are considered to be effective ([12-15]). Patients with CFS are significantly less physically active compared to age-matched healthy controls ([16-18]), and CFS patients with “persistently passive” daily activity patterns are more inclined to avoid physical exertion and experience more physical dysfunction ([18]). In this respect, “persistently passive” patients were defined as patients with an activity level that was lower than the group average for at least 90% of the total observation period; the remaining patients were defined as “active” ([18]).

Decreased levels of physical activity have been reported among patients with RA ([19-26]). For example, an international study reported that the majority of patients with RA did not engage in regular physical exercise (with physical exercise defined as ≥30 minutes of exercise with some shortness of breath and/or perspiration) ([23]). The percentage of Dutch patients with RA who met the recommendation for physical activity was similar to the general population; however, among participants ages 45–64 years, the average number of minutes per week performing physical activity was significantly lower in the RA patient population than in the general population ([19]). Both the belief that physical activity can help manage the disease and increased motivation to engage in physical activity can drive higher levels of physical activity among the RA population ([27]). Among other patient groups, including patients with multiple sclerosis, increased physical activity (measured objectively) has been associated with decreased fatigue ([28]).

With respect to RA, daily physical activity has not been measured objectively, and it is unclear whether fatigue and physical activity are correlated. Daily physical activity can also be influenced by pain, disability, coping, and/or cognition. Clarifying how these factors can influence activity levels may provide important clues for developing an effective treatment for fatigue in patients with RA. Daily physical activity can be measured objectively using actigraphy, providing a reliable, valid measure of human physical activity ([17, 29]).

The objectives of this study were to investigate whether objectively measured activity levels and activity patterns are associated with fatigue levels in patients with RA, and whether pain, disability, coping, and/or cognition are associated with, or influence, the level of activity among patients with RA.

Box 1. Significance & Innovations

  • Whether patients with rheumatoid arthritis (RA) have high or low daily physical activity levels (measured using an actometer) and whether fatigue is correlated with physical activity has not been investigated previously. Understanding this relationship, and the effect of other factors, may provide important clues regarding the treatment of fatigue in patients with RA.
  • In our study, 25% of patients had a low activity level and 75% had a high activity level. Moreover, RA-related fatigue and daily physical activity levels were inversely related, such that higher levels of physical activity were related to reduced fatigue.
  • The relationship between fatigue and daily physical activity level could not be explained by differences in pain, disability, coping, or cognition, or by other factors that can cause fatigue. Whether fatigue causes inactivity, and/or vice versa, is currently unknown and warrants further study.

MATERIALS AND METHODS

Design

This cross-sectional study was part of a cohort study that was designed to determine which factors are associated with fatigue in RA. The study was approved by the Medical Ethics Committee Arnhem-Nijmegen in The Netherlands, and all participants provided written informed consent.

Included patients

From June 2006 through October 2007, consecutive patients ages 18–75 years who visited their rheumatologist at the outpatient clinic of the Radboud University Medical Center at Nijmegen were invited to participate. Patients were informed by either their rheumatologist or nurse specialist, and the information was provided both verbally and in writing.

Inclusion criteria included a diagnosis of RA in accordance with the 1987 American College of Rheumatology classification criteria ([30]), age 18–75 years, and the ability to read and write Dutch. Patients were excluded from the study if they had a second rheumatic disease, a history of malignancies and/or other chronic fatigue–related comorbidity, or a current diagnosis of depression, or were currently receiving psychological or psychiatric treatment. Patients with a comorbidity that was well controlled were eligible to participate; such comorbidities included regulated thyroid disease (free T4 ≥8 pmoles/liter and thyroid-stimulating hormone ≤1.0 unit/liter), controlled diabetes mellitus (with normalized urine and blood glucose values of 45–66.6 mg/dl and 72–100.8 mg/dl, respectively, and glycosylated hemoglobin values <8.0%), mild nonrestrictive chronic obstructive pulmonary disease, and successfully treated nonmetastasized basal cell carcinoma or squamous cell carcinoma in the skin.

Data collection

Patient characteristics (sex, age, and body mass index [BMI]), disease characteristics (disease duration and rheumatoid factor), comorbidity, and medication use were recorded by research nurses at the time of inclusion in the study. Blood samples were obtained and used to determine erythrocyte sedimentation rate, C-reactive protein level, and hemoglobin level. Upon inclusion, disease activity was assessed by the rheumatologist or by a specialized rheumatology nurse using the Disease Activity Score in 28 joints. At inclusion, psychosocial variables, including beliefs regarding the somatic and nonsomatic causes of fatigue, coping strategies, and catastrophizing, were recorded using patient questionnaires that were answered using a computer. Daily activity was recorded objectively using an actometer for 14 consecutive days immediately following inclusion.

Fatigue among patients with RA could have multiple determinants, and previous studies using multidimensional models have documented the importance of variables such as mood, coping processes, and self-efficacy ([4, 31]). Therefore, we collected a set of psychosocial variables that could potentially influence fatigue.

Fatigue severity was assessed using the fatigue severity subscale of the CIS20 ([31]). The CIS fatigue subscale consists of 8 items, each of which is scored using a 7-point Likert scale (yielding a total score of 8–56). The 8 items are designed to measure the patient's fatigue level during the previous 2 weeks, with a higher score on the CIS fatigue indicating a higher level of fatigue. A total score of ≥35 (which is 2 SDs above the mean score for a healthy control group) indicates severe fatigue ([31]). The CIS20 has been validated and is considered reliable under many conditions; this checklist also has been used previously to assess patients with RA ([8, 31]).

Pain was assessed using the bodily pain subscale of the Short Form 36 (SF-36) health survey, which determines the patient's pain severity and pain-related limitations experienced during the previous 4 weeks ([32]). The SF-36 bodily pain subscale score ranges from 0–100, with a higher score indicating less severe pain and fewer pain-related limitations. Pain severity was also assessed using a numerical rating scale regarding pain in the current situation (range 0–10, where 0 = no pain and 10 = extreme pain).

Daily functioning was assessed using the Health Assessment Questionnaire (HAQ) disability index and the physical functioning subscale of the SF-36 ([33]).

The Beck Depression Inventory for primary care (BDI-PC) was used to classify patients for clinical depression ([34]). The BDI-PC is a 7-item self-reporting instrument that is scored by totaling the highest scores from each of the 7 individual items. Each item is rated on a 4-point scale (ranging from 0–3); therefore, the maximum total score for the BDI-PC is 21. A total score of ≥4 on the BDI-PC indicates clinical depression ([34]).

Beliefs regarding the somatic and nonsomatic causes of fatigue were measured using a modified version of the Causal Attribution List. Self-efficacy with respect to fatigue was measured using the Self-Efficacy Scale 28, a questionnaire containing 7 items that are scored on a 4-point Likert scale.

Coping strategies were measured using the Modified Pain Coping Inventory for Fatigue (MPCI-F), which is scored on a 4-point Likert scale and is based on the Pain Coping Inventory. In the MPCI-F, “pain” is replaced with “fatigue.”

To assess catastrophizing, the Fatigue Catastrophizing Scale (FCS) was used. The FCS was derived from the Pain Catastrophizing Scale, with the word “pain” replaced with “fatigue” ([35]).

Daily physical activity, including general daily activities at home, at work, during rest, and during leisure time, was measured using an actometer (Actilog version 4.1), an ankle-worn motion-sensing device that registers and quantifies daily physical activity ([17, 18]). The actometer contains a piezoelectric sensor that is sensitive in 3 directions. Acceleration of the sensor above a predefined threshold (the actometers were all calibrated to have the same threshold) is considered to be activity and is stored in the device's internal memory. The actometer's counter was read and reset each second by the microcontroller, which added the value to an integration counter that was set to 5 minutes. Therefore, the activity score was measured every 5 minutes, with a maximum of 300 scores counted.

Each actometer was calibrated before collecting data. The actometer's output was an activity count that was the weighted sum of the number of accelerations measured during a 5-minute period. To obtain a valid measurement of the patient's daily activity, the patient was instructed to wear the actometer for 14 consecutive days and to remove it only during swimming and bathing; therefore, all activities throughout the day (except swimming and bathing) were recorded.

Actometer data

If a non–actometer-wearing period exceeded 3 hours, the day was recorded as “nonvalid.” Patients with >2 nonvalid registration days were excluded from the analysis; in case of 1 or 2 nonvalid days, data for these days were inserted using the mean values of the remaining 10 or 11 valid days. Individual physical activity was analyzed using the Actilog Analyzer software, version 4.10. The mean physical activity score of all included patients was calculated to determine the average daily physical activity over the entire 12-day period, and this was expressed as the average number of accelerations per 5-minute interval. To subtype activity levels among the patients with RA, we assigned each patient to 1 of 2 groups: patients with low daily physical activity and patients with high daily physical activity.

In accordance with van der Werf et al, the patients with a low activity level were defined objectively as those patients whose activity level was lower than the group average for at least 90% of the total observation period; the remaining patients were defined as having a high daily activity level ([18]). Figure 1 shows examples of the daily activity pattern of a patient with a high daily activity level (Figure 1A) and a patient with a low daily activity level (Figure 1B).

Figure 1.

Graphical depiction of 2 continuous actometer recordings over 12 consecutive days. A, A rheumatoid arthritis (RA) patient with a high activity level. B, An RA patient with a low activity level. The y-axis shows the activity count and the x-axis shows the day number and the mean physical activity during each day (M). The mean activity levels of all valid days and their SDs are shown in the top right-hand corner of each graph. The dark gray box is the first valid registration day (day 2).

Statistical analysis

Baseline differences between the high and low daily activity level groups were analyzed using Student's t-test, the Mann-Whitney U test, or the chi-square test, where appropriate. To determine whether there was an association between fatigue and daily physical activity, continuous and dichotomous (low and high activity level) linear regression analyses were performed, with CIS fatigue score as the dependent variable and daily physical activity as the independent variable, with correction for potential confounders. Age and sex were predetermined for inclusion in the adjusted linear regression model, whereas all other variables were considered to be confounders if the regression coefficient of the main effect (daily physical activity) in the linear regression model changed by ≥10% after adding the variable (one of the baseline characteristics) (Table 1) to the model. To analyze the association between fatigue and activity score divided into equal group sizes, we performed a sensitivity analysis using a linear regression analysis with activity divided into quartiles. The association between activity level (low and high) and the occurrence of fatigue (severe and not severe) was also analyzed in a 2 × 2 table using the chi-square test; for this analysis, fatigue was classified as either severely fatigued (CIS fatigue score ≥35) or not severely fatigued (CIS fatigue score <35) ([6]). All analyses were performed using the PASW statistics package, version 18.0 (SPSS).

Table 1. Baseline characteristics of 167 patients with rheumatoid arthritis with a low or high daily activity level*
 Low activity level (n = 42)High activity level (n = 125)P
  1. Values are the mean ± SD unless indicated otherwise. IQR = interquartile range; COPD = chronic obstructive pulmonary disease; DMARD = disease-modifying antirheumatic drug; DAS28 = Disease Activity Score in 28 joints; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; NRS = numerical rating scale; SF-36 = Short Form 36 health survey; CIS = Checklist Individual Strength; HAQ = Health Assessment Questionnaire; DI = disability index; SCL-90 = Symptom Checklist-90; BDI-PC = Beck Depression Inventory for primary care; CAL = Causal Attribution List; SES28 = Self-Efficacy Scale 28; MPCI-F = Modified Pain Coping Inventory for Fatigue; FCS = Fatigue Catastrophizing Scale.
  2. aStatistically significant.
  3. bCRP levels <5 mg/liter were scored as 0.
Age, years56.78 ± 11.0354.79 ± 10.590.30
Female sex, no. (%)33 (79)67 (54)0.004a
Body mass index, median (IQR) kg/m226.72 (24.34–29.04)24.51 (22.69–26.68)0.004a
Positive rheumatoid factor, no. (%)31 (73.8)92 (73.6)0.99
Disease duration, median (IQR) years9.5 (6–17.25)10 (5.0–16)0.43
Cardiovascular disease, no. (%)6 (14)18 (14)0.99
COPD, no. (%)0 (0)5 (4)0.19
Diabetes mellitus, no. (%)1 (2.4)8 (6.4)0.32
DMARD use, no. (%)33 (78.6)110 (88)0.13
Biologic agent use, no. (%)18 (42.9)43 (34.4)0.33
Corticosteroids, no. (%)11 (26.2)31 (24.8)0.86
Statin use, no. (%)3 (7.1)3 (2.4)0.15
DAS28 (range 0–10)3.28 ± 1.183.07 ± 1.230.32
Swollen joint count 28 (range 0–28), median (IQR)3 (2–6)3 (2–6)0.65
Tender joint count 28 (range 0–28), median (IQR)2 (1–4.25)1 (0–4)0.24
ESR, median (IQR) mm/hour8 (4–16.5)7 (4–13)0.31
CRP, median (IQR) mg/literb0 (0–8.75)0 (0–6)0.26
Hemoglobin, mg/dl147.6 ± 10.27149.37 ± 16.760.89
NRS pain severity (range 0–10)4.76 ± 2.694.26 ± 2.360.25
SF-36 bodily pain (range 0–100)63.94 ± 18.6263.61 ± 19.070.96
CIS fatigue (range 8–56)35.74 ± 12.8430.88 ± 12.270.03a
HAQ DI (range 0–3)0.80 ± 0.560.65 ± 0.600.16
SF-36 physical functioning (range 0–100)56.31 ± 23.5659.96 ± 24.200.40
SF-36 social functioning (range 0–100)73.51 ± 21.6075.60 ± 23.430.66
SCL-90 sleep disturbances (range 3–15)6.31 ± 2.985.90 ± 2.920.24
Depression, BDI-PC ≥4, no. (%)0 (0)9 (7.2)0.56
CAL somatic (range 3–12), median (IQR)6 (5–7)6 (6–7.5)0.24
CAL nonsomatic (range 3–12)9.1 ± 1.828.66 ± 1.950.21
SES28 (range 7–28)18.83 ± 3.7519.73 ± 3.550.17
MPCI-F worrying (range 9–36), median (IQR)14 (11–17.3)14 (11–17)0.72
MPCI-F retreating (range 7–28)12.62 ± 3.9511.84 ± 3.670.24
MPCI-F resting (range 5–20)12.24 ± 2.9911.41 ± 2.930.12
MPCI-F fatigue transformation (range 4–16)8.71 ± 2.458.32 ± 2.530.37
MPCI-F reducing demands (range 3–12)6.14 ± 1.846.67 ± 2.010.13
MPCI-F distraction (range 5–20)11.93 ± 2.4011.06 ± 3.220.06
FCS rumination (range 0–16)3.86 ± 3.754.46 ± 3.770.37
FCS helplessness (range 0–24)3.60 ± 4.424.07 ± 4.270.35
FCS magnification (range 0–12)0.88 ± 1.741.17 ± 1.910.28

RESULTS

Patients

Of the 230 patients who were included in the study, 181 (78%) were willing to wear an actometer for 14 consecutive days. The first and last days of data collection were excluded from the analysis, yielding 12 complete continuous days of data. One hundred sixty-seven (92%) of these 181 patients had at least 10 daily measurements of activity and were included in the analysis (Table 1). We found no statistically significant difference in baseline characteristics between the 49 patients who were unwilling to wear an actometer and the 181 patients who agreed to wear an actometer, nor did we find a difference between the 14 patients with an insufficient number of daily measurements and the 167 patients who were analyzed (data not shown). Based on our definition (see Methods), 42 patients were classified as having a low activity level, and the remaining 125 patients were classified as having a high activity level.

High daily activity level versus low daily activity level

Univariate analyses

Of the 167 patients with RA who were included in the analysis, 44% were severely fatigued (CIS fatigue score ≥35), and the overall mean ± SD CIS fatigue score was 32.1 ± 12.6. The mean ± SD daily physical activity score of 167 patients was 73 ± 27. Therefore, the average activity level of patients with RA lies between healthy controls and patients with CFS, and is similar to other chronic conditions (Table 2). The RA patients with a high activity level had a mean ± SD daily activity score of 83 ± 23, and the patients with a low activity level had a mean ± SD daily activity score of 43 ± 9. Table 1 summarizes separately the clinical characteristics of the 42 low activity (25%) and the 125 high activity (75%) patients with RA. Our analysis revealed no significant difference between the high activity and low activity groups with respect to comorbidity or medication use. Conversely, the high activity patients had significantly lower CIS fatigue scores than the low activity patients. Moreover, the low activity patient group had significantly higher BMI scores and a significantly higher percentage of women. No differences were detected between the 2 groups with respect to other characteristics, including pain, disability/function, and any of the variables that reflect coping and cognition.

Table 2. Overview of studies that measured daily physical activity using an ankle-worn accelerometer, with activity count as the outcome measure*
StudyParticipants (n)Total no. of registered daysActivity count, mean ± SD
  1. RA = rheumatoid arthritis; COPD = chronic obstructive pulmonary disease.
Current studyRA patients (167)1273 ± 27
Van der Werf et al, 2000 ([18])Chronic fatigue syndrome (277)1266 ± 22
 Healthy controls (47)1291 ± 25
Servaes et al, 2002 ([42])Severely fatigued disease-free patients with breast carcinoma (57)1276.1 ± 22.5
 Non–severely fatigued disease-free patients with breast carcinoma (93)1279.1 ± 20.8
 Patients without a history of breast carcinoma (78)1276.9 ± 15.5
Steele et al, 2003 ([41])COPD patients (41)587.4 ± 38.8

Relationship with fatigue

To analyze the relationship between activity level and fatigue, both age and sex were predetermined for inclusion in the adjusted linear regression model, in which BMI, pain severity, and the HAQ were included as confounders. The analysis revealed that the relationship between activity and fatigue was linear (Table 3), with each unit increase in activity correlating with a 0.08 decrease in the fatigue score. To facilitate the interpretation of these results, Figure 2 shows box plots of the fatigue levels of the low and high daily activity level groups. The plot shows that the median CIS fatigue score of the low activity patients was higher than the median CIS fatigue score of the high activity patients (36 versus 31; P = 0.03).

Table 3. Linear regression model of fatigue (continuous) versus daily activity (continuous)*
 BP95% CI
  1. P values and 95% confidence intervals (95% CIs) are based on the regression coefficient. B = regression coefficient; constant = intercept.
  2. aThe adjusted model was corrected for age, sex, the disability index of the Health Assessment Questionnaire, body mass index, and the numerical rating scale for pain.
Unadjusted model   
Constant38.49< 0.00132.94, 44.04
Activity−0.0870.017−0.158, −0.016
Adjusted modela   
Constant49.90< 0.00134.62, 65.19
Activity−0.0820.008−0.14, −0.021
Figure 2.

Box plots of the Checklist Individual Strength (CIS) fatigue scores are shown for the 2 activity groups. The bold horizontal bars indicate the median and the upper and lower boxes indicate the quartiles. The whiskers indicate the minimum and maximum values.

Table 4 summarizes the results of the linear regression analysis between low and high activity, adjusted for age, sex, HAQ, BMI, and pain. The table shows that high activity patients with RA scored an average of 4 points lower for fatigue than low activity patients with RA. The CIS fatigue score is usually divided into severe fatigue (CIS ≥35) and elevated/normal fatigue (CIS <35). A chi-square test revealed a significantly higher percentage of severely fatigued (i.e., CIS ≥35) patients in the passive group than in the active group (60% versus 38%; P = 0.017).

Table 4. Results of the unadjusted and adjusted linear regression models of fatigue (continuous) for low versus high activity*
 BP95% CI
  1. P values and 95% confidence intervals (95% CIs) are based on the regression coefficient. B = regression coefficient; constant = intercept; NA = not applicable.
  2. aThe adjusted model was corrected for age, sex, the disability index of the Health Assessment Questionnaire, body mass index, and the numerical rating scale for pain.
Unadjusted model   
Constant35.74< 0.00131.96, 39.52
High activity level−4.860.03−9.23, −0.49
Adjusted modela   
Constant46.20< 0.00131.94, 60.47
High activity level−4.420.018−8.059, −0.78
Adjusted model (divided into quartiles)a   
Constant41.331< 0.00128.47, 54.19
First quartile5.6980.0141.16, 10.24
Second quartile3.4670.126−0.99, 7.92
Third quartile0.1980.928−4.10, 4.50
Fourth quartile0NANA

Sensitivity analysis

Table 4 also shows the linear regression analysis of sensitivity with activity divided into quartiles; this analysis was adjusted for age, sex, HAQ, BMI, and pain. Significantly higher fatigue scores (P < 0.001) were found between the low activity patients (i.e., the first quartile) and the high activity patients (i.e., the fourth quartile).

DISCUSSION

Based on the results obtained from this study, patients with RA who have a high level of daily physical activity have less fatigue than patients with low daily physical activity; moreover, the level of activity is not associated with pain, disability, coping, or cognition.

Activity level was measured objectively, and patients who had an activity level that was lower than the group average for more than 90% of the observation time were defined as having a low activity level (see Methods). Therefore, the classification of each patient into either the low or high activity group was based on the activity of his/her fellow patients. The majority of patients (75%) were classified as having high activity and the remaining 25% were classified as having low activity. Overall, nearly half of all patients had severe fatigue, as determined using the CIS questionnaire. According to the CIS definition of severe fatigue, this means that many of the RA patients in this study had a fatigue level that was similar to the level in patients with CFS. Based on the regression analysis, higher activity scores were associated with lower fatigue scores, even after correcting for potential confounders. Consistent with this analysis, the average fatigue score among patients with high activity was 5 points lower than that for patients with low activity. This difference is likely relevant, as a trial that examined the effect of cognitive–behavioral therapy in RA patients reported a 6-point difference in CIS fatigue score between the 2 groups ([36]).

In addition to higher fatigue levels, RA patients who have more pain and disability would likely have lower average levels of activity. Moreover, activity level can also be associated with coping and cognition; patients with “passive” or “unproductive” coping styles can also have lower levels of activity. However, our results revealed no strong indication that either coping or cognition is correlated with activity level. Notably, coping, cognition, pain, and disability are all associated with fatigue in patients with RA ([8]).

To the best of our knowledge, this is the first study to objectively measure and relate the level of physical activity and the activity pattern in patients with RA. A few other studies have investigated the association between physical activity and fatigue, and none of these studies included patients with RA. It is important to note that decreased physical activity has been associated with increased fatigue in patients with CFS, Sjögren's disease, and breast cancer ([37, 38]).

Nevertheless, one cannot necessarily conclude that “passiveness” causes fatigue (or vice versa). Indeed, both scenarios are conceivable; fatigue may lead to decreased activity, and decreased activity may lead to fatigue. To address the possibility of a clinically relevant causal relationship, an RCT should be performed. For example, one can test whether increasing activity through exercise and/or training can reduce fatigue in severely fatigued RA patients with low daily activity ([12]).

One question that remains is whether RA patients have low levels of fatigue relative to healthy subjects and/or the general population. Although several studies have investigated the activity levels of RA patients, comparing the findings across studies is problematic because of differences in the assessment methods (e.g., assessment by questionnaire versus objective measurements) ([19-26, 39]). Based on the published literature, patients with RA appear to have a relatively low level of physical activity. In particular, a study comparing RA patients from various countries found that up to 80% of patients in some countries were “physically inactive” ([23]); the authors defined “physically active” as ≥30 minutes of exercise with at least some shortness of breath and/or perspiration ([23]).

The type of accelerometer that we used in this study has been used in other studies, including studies with healthy controls and other patient groups (Table 2). Therefore, the group averages can be compared (albeit indirectly), although groups may not be fully comparable with respect to age or sex. Nevertheless, the activity level of patients with RA lies somewhere between the activity level of patients with CFS and healthy controls, but lower than the activity level of patients with breast cancer or Sjögren's syndrome. Moreover, compared to healthy controls, patients with chronic diseases have decreased daily activity levels, including patients with diabetes mellitus ([40]), Sjögren's syndrome ([37]), chronic obstructive pulmonary disease ([41]), breast carcinoma ([42]), and hereditary motor and sensory neuropathy type 1 ([43]).

The strength of our study is the large number of patients and the measurement of daily activity for 12 consecutive days. In addition, because nonrandomized studies always carry a risk of confounding, the analyses were adjusted for several confounding factors. Because the difference between the adjusted and unadjusted models was negligible, it is unlikely that any residual confounding factors biased the association. Moreover, actigraphy (i.e., the use of a uniaxial actometer) is a reliable and valid instrument for continuously and objectively measuring physical activity ([18, 29, 44]). However, a limitation is that the actometer results cannot be reliably correlated to energy expenditure ([45]). Therefore, patients with a low daily activity level should be advised to increase their physical activity, although we cannot precisely define or recommend how many minutes one should be physically active; thus, current recommendations advise patients to engage in physical activity for ≥30 minutes/day, ≥5 days/week.

Another limitation of our study might be the substitution of missing actometer values with the mean values of the remaining 10 or 11 days. However, because only 7 (4.2%) of the 167 patients had 1 or 2 missing values, the effect of this approach was likely negligible. Another limitation could be the cross-sectional design of this study. To determine whether the relationship between fatigue and daily activity level is causal, it would have been an advantage if longitudinal data could be used. If changes in daily activity level would precede changes in fatigue, this could be interpreted causally. A stronger design to determine causal relationships, however, is an RCT.

In summary, we report that the level of activity and the level of fatigue are associated in patients with RA; however, which factors influence the level of activity in RA patients remains unclear. With respect to fatigue in RA patients, which treatments are effective also remains unclear. Nevertheless, reducing pain and inflammation should have high priority in managing RA, and reducing pain may also reduce fatigue. However, RA patients whose disease is managed well can still develop severe fatigue ([6]). In addition to the association between fatigue and decreased activity revealed here, fatigue can also be associated with pain, disability, coping, and cognition ([7-9]). RCTs have suggested that cognitive–behavioral therapy and physical exercise can be beneficial in treating fatigue in RA patients ([10, 11]). Although several RCTs found that exercise is beneficial for RA patients ([46]), only one relatively small trial found that exercise may be beneficial specifically for fatigue in RA patients ([10]). Evidence from studies of CFS also suggests that exercise may be beneficial for RA patients ([10, 47]). Exercise might be particularly relevant to patients with low daily activity levels. For relatively active patients, other factors, such as changing dysfunctional beliefs with respect to fatigue or reducing their “all-or-nothing” behavior, may be more important ([48]). Future studies should investigate whether fatigue in patients with RA can indeed be treated effectively with exercise or another form of graded activity, particularly among patients who have a low level of physical activity and chronic severe fatigue.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Ms Rongen-van Dartel had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Rongen-van Dartel, Repping-Wuts, van Hoogmoed, Knoop, Bleijenberg, van Riel, Fransen.

Acquisition of data. Rongen-van Dartel, Repping-Wuts, van Hoogmoed, Knoop, van Riel, Fransen.

Analysis and interpretation of data. Rongen-van Dartel, Repping-Wuts, Bleijenberg, Fransen.

Ancillary