Current and Maintained Health-Enhancing Physical Activity in Rheumatoid Arthritis: A Cross-Sectional Study

Authors


Department of Neurobiology, Care Sciences and Society, Division of Physiotherapy, 23100 Karolinska Institutet, SE-141 83 Huddinge, Sweden. E-mail: ingrid.demmelmaier@ki.se

Abstract

Objective

To describe and identify the explanatory factors of variation in current and maintained health-enhancing physical activity (HEPA) in persons with rheumatoid arthritis (RA).

Methods

In this cross-sectional study, current HEPA was assessed with the International Physical Activity Questionnaire and maintained HEPA with the Exercise Stage Assessment Instrument, the latter explicitly focusing on both aerobic physical activity and muscle strength training. Sociodemographic, disease-related, and psychosocial data were retrieved from the Swedish Rheumatology Quality (SRQ) registers and a postal questionnaire. The explained variations in the respective HEPA behaviors were analyzed with logistic regression.

Results

In all, 3,152 (58.5%) of 5,391 persons identified as eligible from the SRQ registers responded to the questionnaire. Current HEPA was reported by 69%, and maintained HEPA by 11% of the respondents. The most salient and consistent factors explaining variation in both current and maintained HEPA were self-efficacy, social support, and outcome expectations related to physical activity.

Conclusion

To our knowledge, this is the first study exploring maintained physical activity in a large well-defined sample of persons with RA. Our results indicate that a minority perform maintained HEPA, including both aerobic physical activity and muscle strength training, and that psychosocial factors are the most salient and consistent in the explanation of HEPA variation.

INTRODUCTION

There is solid evidence for the safety and benefit of aerobic exercise and muscle strength training in persons with rheumatoid arthritis (RA) ([1-4]). Moreover, epidemiologic research on numerous populations has consistently demonstrated that cardiovascular disease, a prevalent comorbidity in RA ([5]), can be prevented with sufficient physical activity ([6]). Health-enhancing physical activity (HEPA), i.e., a minimum of 150 weekly minutes of moderately intense aerobic physical activity, has long been recommended to reduce the risks of cardiovascular and other lifestyle-related diseases ([7]). More recently, it has been suggested that this may be replaced with 60–75 minutes of vigorously intense aerobic physical activity and additional twice-weekly muscle strength training to further reduce the risk of cardiovascular disease in the general population ([8-10]), as well as in populations with long-term diseases such as arthritis ([11]).

A substantial proportion of the RA population does not accumulate enough aerobic physical activity ([12-14]), and the proportion that performs muscle strength training is not known. Health behavior usually changes stepwise, with maintained behavior defined as being performed during at least 6 months ([15]) and, while current physical activity has previously been reported, there is no information on maintained physical activity in large, well-defined samples of persons with RA.

Increased knowledge of the explanation of variation in physical activity in RA may form a more solid base for the design of programs aimed at the promotion of maintained physical activity. Previous research has identified sociodemographic explanatory factors for current physical activity in general populations ([16]) and for physical inactivity in people with RA ([13, 17]). Disease-related variables ([13, 18, 19]) and psychosocial variables ([19-21]) have also previously been identified as contributing to the explanation of current physical activity/inactivity in RA. However, none of these studies have used a comprehensive set of biopsychosocial factors to explain variation in current or maintained physical activity in RA and thus their respective contributions are unknown.

The aim of the present study was to describe and to identify the explanatory factors of variation in current and maintained HEPA in persons with RA.

Box 1. Significance & Innovations

  • This is the first study to explore, in a large well-defined sample of persons with rheumatoid arthritis (RA), maintained health-enhancing physical activity (HEPA) according to present guidelines, including both 150 minutes of weekly moderately intense aerobic physical activity and twice-weekly muscle strength training.
  • While close to 70% reported current HEPA (past week), only 11% reported maintained HEPA (>6 months).
  • The most salient and consistent factors explaining variation in both current and maintained HEPA were self-efficacy, social support, and outcome expectations related to physical activity.
  • Health care professionals need to support not only the adoption, but also the maintenance of HEPA in patients with RA.

SUBJECTS AND METHODS

Design

Our cross-sectional study was performed using the Swedish Rheumatology Quality (SRQ) registers. The registers comprised approximately 27,000 individuals with RA. Six rheumatology clinics (8 sites) were chosen to represent university hospitals (n = 2) and county hospitals (n = 4), rural and urban areas, and different parts of Sweden. The SRQ registers were searched to identify eligible participants in August 2010, and data on sex, age, time since first clinical visit, as well as the Disease Activity Score in 28 joints (DAS28), pain, general health perception, and the Stanford Health Assessment Questionnaire (HAQ) disability index (DI) collected at the most recent medical check-up were retrieved for each individual in the target population for the study. A questionnaire was mailed 1 month later to those fulfilling the study entry criteria, with a reminder yet another month later.

Participants

The 6 participating clinics were represented in the SRQ registers by 9,560 registered patients with RA according to the 1987 American College of Rheumatology criteria ([22]). To identify a population that would be at target for a physical activity intervention, only those up to age 75 years and with a HAQ DI score up to 2 were included (n = 5,593). Some registered patients had died (n = 164), emigrated (n = 24), or had protected identity (n = 14). The questionnaire was thus mailed to 5,391 persons.

The study was carried out in compliance with the Helsinki Declaration. Ethics approval was obtained from the Stockholm Regional Ethical Review Board. Participation was sought in a letter containing information about the study, and the participants consented by returning their questionnaires.

Dependent measures

Current and maintained physical activity among questionnaire responders was assessed by the International Physical Activity Questionnaire (IPAQ) and the Exercise Stage Assessment Instrument (ESAI), respectively, both of which were included in the mailed questionnaire. The short version of the IPAQ ([23]) assesses overall physical activity during the past week without separating aerobic physical activity from muscle strength training. Three intensity levels (vigorous, moderate, and walking) and 4 domains (home, work, transport, and leisure time) are assessed. The short version has acceptable test–retest reliability (ρ = 0.8) and criterion-related validity compared with accelerometers (ρ = 0.3), as shown in a 12-country evaluation study that included Sweden ([24]).

The ESAI ([25]) was originally a single item on regular physical exercise that was modified for the present study to include 2 items; i.e., one item defining aerobic physical activity as that of moderate intensity during minimum 30 minutes at least 5 times per week and one item defining muscle strength training as resistance training at least twice a week, both followed by the question, “are you physically active according to this description?” Five response options based on the stages in the transtheoretical model ([15]) are given after each of the 2 items. Three options represent the precontemplation, contemplation, and preparation stages and, thus, not HEPA. The remaining 2 options represent recently adopted (<6 months) and maintained (>6 months) HEPA, respectively. The original ESAI has demonstrated sufficient test–retest reliability ([26]) and construct validity with other exercise stage of change measures ([27]).

Independent variables

Data on age, sex, and time since first visit to a rheumatology clinic were retrieved from the SRQ registers, while additional sociodemographic data on members of household, education, income, and Swedish language comprehension were collected by the postal questionnaire.

The questionnaire was also used to collect concurrent information on disease-related data regarding comorbidity, pain, fatigue, and general health perception rated on visual analog scales (range 0–100), and the HAQ DI was used to rate activity limitation (range 0–3), where 0 = “without difficulty” and 3 = “unable to do” in 8 domains ([28, 29]).

Psychosocial data were also obtained by the questionnaire. The Exercise Self-Efficacy Scale (ESES) consists of the stem, “how confident are you to exercise when …,” followed by 6 items describing common barriers for exercise. Ratings are made on response scales, where 1 = “not at all confident” and 6 = “very confident.” The ESES has sufficient internal consistency and concurrent validity ([30]). The Social Support for Physical Activity Scale consists of the stem, “during the past 3 months, my family (or members of my household) or friends … ,” followed by 13 items describing supportive activities, e.g., “… offered to be physically active with me.” Ratings are made on scales where 0 = “none” and 5 = “very often” for family and friend support, respectively, resulting in a total score between 0 and 130. The scale has satisfactory internal consistency and test–retest reliability ([31]). The modified Fear-Avoidance Beliefs Questionnaire consists of 4 items on beliefs about physical activity causing pain and injury. Ratings are made on 7-point scales where 0 = “do not agree at all” and 6 = “agree completely” ([32]). Outcome expectations of physical activity were measured with 2 study-specific items; one concerning long-term effects of physical activity on general health and one concerning effects of physical activity on RA symptoms. Ratings are made on a scale where 1 = “not at all certain” and 10 = “completely certain.”

Data management

The IPAQ scoring was performed according to its protocol, version 2.0 ([23]), with 1 exception. All participants who, in 1 or more intensity category, had reported days (frequency) but not time (duration) of physical activity, or vice versa, were recoded as having spent zero time in that intensity category. Otherwise, if any intensity category had missing values, the physical activity could not have been summed, and the entire case would have been excluded from analysis due to these missing values. To reduce the effect of known measurement errors of self-reports ([33, 34]), and to minimize the effect of skew in the data, physical activity was categorized as low, moderate, or high using the IPAQ scoring protocol ([23]). For the purpose of this study, the moderate and high physical activity categories were classified as “current HEPA” and the low physical activity category as “no current HEPA.”

The ESAI data were, for the purpose of the logistic regression models, dichotomized. One category consisted of maintained (>6 months) aerobic physical activity in combination with maintained (>6 months) muscle strength training, i.e., “maintained HEPA,” while the other category consisted of all other combinations of answering options to the 2 ESAI questions, i.e., “no maintained HEPA.”

Independent variables were coded into 2–4 categories (Table 1). Time since first visit at a rheumatology clinic was used as a proxy for disease duration, forming 3 categories based on tertiles of the number of months since first clinical visit. Pain was divided into 3 groups based on the categories suggested by Collins et al ([35]). Fatigue, general health perception, exercise self-efficacy, and social support for physical activity categories were based on tertiles in the present material. Fear-avoidant beliefs and outcome expectations of physical activity demonstrated small variation and were dichotomized according to median values.

Table 1. Descriptive statistics of responders and percentage performing current HEPA and maintained HEPA*
 Responders, no.Current HEPA (IPAQ)Responders, no.Maintained HEPA (ESAI)
%Pa%Pa
  1. HEPA = health-enhancing physical activity; IPAQ = International Physical Activity Questionnaire; ESAI = Exercise Stage Assessment Instrument; VAS = visual analog scale; HAQ = Health Assessment Questionnaire; DI = disability index; ESES = Exercise Self-Efficacy Scale; SSES = Social Support for Physical Activity Scale; mFABQ = modified Fear-Avoidance Beliefs Questionnaire; NRS = numeric rating scale; RA = rheumatoid arthritis.
  2. aBy chi-square test.
Sex      
Men84266.90.10279211.40.351
Women2,29869.3 2,17810.8 
Age, years      
18–3412169.40.07611919.3< 0.001
35–5473972.0 71913.4 
≥552,28067.5 2,1329.7 
Education      
University1,01876.7< 0.00198913.20.01
College78769.8 74711.4 
Other37869.0 3558.7 
Basic92459.3 8538.8 
Income      
Above average1,42675.4< 0.0011,37611.90.112
Below average1,62563.9 1,51910.1 
Adults in household      
3 or 432571.60.00733612.20.713
21,98270.2 1,88711.0 
Living alone76464.4 71410.5 
Children age <18 years      
02,62668.20.0262,46910.60.065
123467.1 23115.6 
≥226076.2 25410.6 
Swedish language comprehension      
No difficulties3,02969.7< 0.0012,87511.00.758
With difficulties6435.9 5712.3 
Disease duration (months since first clinic visit)      
Short (0–24)1,04968.70.48099110.80.148
Moderate (25–55)1,04667.4 9859.6 
Long (≥56)1,04569.9 99412.4 
Comorbidities      
01,91771.7< 0.0011,82512.40.007
188467.4 8348.6 
≥234027.6 3108.7 
VAS pain (range 0–100 mm)      
Low (0–29)1,64775.5< 0.0011,57913.2< 0.001
Moderate (30–54)72965.4 6918.1 
High (55–100)74465.4 6868.7 
VAS fatigue (range 0–100 mm)      
Low (0–22)1,01277.2< 0.00197014.5< 0.001
Moderate (23–52)1,04369.8 99111.3 
High (53–100)1,06359.9 9957.1 
VAS health (range 0–100 mm)      
Good (0–15)96280.4< 0.00192015.5< 0.001
Moderate (16–40)1,06868.4 1,0229.1 
Poor (41–100)1,01959.4 9478.6 
HAQ DI activity limitation score (range 0–3)      
074981.0< 0.00171416.4< 0.001
0.1–1.01,64471.6 1,56610.3 
1.1–3.073050.7 6786.5 
ESES score (range 6–60)      
High (37–60)91282.6< 0.00189020.4< 0.001
Moderate (25–36)85673.5 8349.2 
Low (6–24)91658.5 9004.0 
SSES score (range 0–130)      
High (60–130)65881.9< 0.00164917.4< 0.001
Moderate (43–59)64669.0 6317.0 
Low (0–42)73362.6 7096.6 
mFABQ (range 0–24)      
Low (0–6)1,38677.9< 0.0011,34614.9< 0.001
High (7–24)1,54163.5 1,4708.0 
Outcome expectations on health (NRS, range 1–10)      
Median or higher (10)1,89276.4< 0.0011,83014.6< 0.001
Below median (1–9)1,17358.7 1,1085.1 
Outcome expectations on RA symptoms (NRS, range 1–10)      
Median or higher (8–10)1,75678.4< 0.0011,68415.3< 0.001
Below median (1–7)1,30557.4 1,2515.1 

Statistical analyses

The SRQ register data available for nonresponders on sex, age, disease duration, DAS28 score, pain, general health perception, and HAQ DI score from the most recent medical examination allowed for a dropout analysis.

Univariate associations between sample characteristics and HEPA were analyzed with Pearson's chi-square test of homogeneity. Bivariate relationships between HEPA and sociodemographic, disease-related, and psychosocial variables were analyzed by chi-square tests. Independent effects of each independent variable on HEPA variation were assessed using a logistic regression. Crude and adjusted odds ratios (ORs) with 95% confidence intervals were calculated. Adjustments were performed for all study variables. The ORs were calculated against the reference categories of women, those ages ≥55 years, basic education, below-average income, living alone, having ≥2 children living at home, difficulties understanding the Swedish language, long disease duration, ≥2 comorbidities, highest tertile of pain, fatigue, and general health perception, a HAQ DI score of ≥1, above median for fear of avoidance beliefs, low self-efficacy for exercise, low expectations on physical activity for general health, and RA symptoms, respectively, and having low social support for physical activity. The model fit was estimated using Nagelkerke's R2. Any signs of multicollinearity among the independent variables were examined by entering the model in a linear regression. The variation inflation factors were all below 3, indicating no severe multicollinearity.

A multiple imputation for missing data was run before the analyses due to missing data ([36]). All possible information among the independent variables was used to predict the outcome. A total of 5 imputations were run. The results are shown both for the complete case analyses and the pooled results after the multiple imputations. All statistical analyses were performed using SPSS, version 20.0 (IBM).

RESULTS

In all, 3,152 (58.5%) of the 5,391 persons targeted answered the questionnaire (Figure 1). The nonresponders were younger and had shorter disease duration, more disease activity, pain, fatigue, activity limitation, and poorer general health perception at their most recent medical examination compared to the responders (Table 2).

Figure 1.

Flow chart of the selection procedure of the study sample. SRQ = Swedish Rheumatology Quality.

Table 2. Data from the Swedish Rheumatology Quality registers on nonresponders and responders*
 NonrespondersRespondersP
MedianIQRMedianIQR
  1. Sex distribution was analyzed using chi-square test and all others using Mann-Whitney U test. IQR = interquartile range; DAS28 = Disease Activity Score in 28 joints; VAS = visual analog scale; HAQ = Health Assessment Questionnaire; DI = disability index.
Sex    0.226
Men, no. (%)605 (28) 843 (27)  
Women, no. (%)1,536 (72) 2,309 (73)  
Age, years6048–666254–68< 0.001
Time since first visit to clinic, months3513–653714–680.018
DAS28 score, range 0–102.862.06–4.022.822.00–3.800.010
VAS pain, range 0–10 mm2710–502510–480.016
VAS general health, range 0–100 mm2811–522510–490.001
HAQ DI activity limitation score, range 0–30.630.13–1.130.500.03–1.000.011

Among responders, missing data in the questionnaire were <7% in all variables, except for self-efficacy for exercise (15% missing) and social support for physical activity (37% missing). Those without data on self-efficacy for exercise reported less current HEPA with the IPAQ (P < 0.001 by Pearson's chi-square test), but no difference in maintained HEPA with the ESAI (P < 0.015 by Pearson's chi-square test) compared to those with self-efficacy data. The same pattern was seen for those without data on social support with less HEPA according to the IPAQ (P < 0.001 by Pearson's chi-square test), but not according to the ESAI (P < 0.051 by Pearson's chi-square test). Those without data on social support were more likely to be men, older, live alone, have lower education and income below average, report Swedish language difficulties, have more comorbidities, longer disease duration, more disease activity, more pain, worse health perception, more disability, more fear-avoidance beliefs, and lower outcome expectations of physical activity.

A total of 69% of the responders reported current HEPA according to the IPAQ, while 21% and 14% reported maintained aerobic physical activity and muscle strength training, respectively, according to the ESAI. Eleven percent reported maintained HEPA, including both aerobic physical activity and strength training with the ESAI. Descriptions of current and maintained HEPA are presented in Table 1. Univariate analyses demonstrated 16 independent variables to be significantly correlated to current HEPA (IPAQ), whereas 12 variables were significantly correlated to maintained HEPA (ESAI) (Table 3).

Table 3. Crude ORs for current HEPA and maintained HEPA based on multiple imputations (n = 3,152)*
 Current HEPA (IPAQ)Maintained HEPA (ESAI)
OR95% CIOR95% CI
  1. OR = odds ratio; HEPA = health-enhancing physical activity; IPAQ = International Physical Activity Questionnaire; ESAI = Exercise Stage Assessment Instrument; 95% CI = 95% confidence interval; VAS = visual analog scale; HAQ = Health Assessment Questionnaire; DI = disability index; ESES = Exercise Self-Efficacy Scale; SSES = Social Support for Physical Activity Scale; mFABQ = modified Fear-Avoidance Beliefs Questionnaire; NRS = numeric rating scale; RA = rheumatoid arthritis.
Sex    
Men0.890.75–1.061.060.82–1.37
Women1.00   
Age, years    
18–341.090.89–1.332.241.39–3.61
35–541.231.03–1.481.441.11–1.87
≥551.00 1.00 
Education    
University2.261.86–2.751.601.19–2.16
College1.591.30–1.941.340.96–1.85
Other1.541.20–1.991.000.64–1.55
Basic1.00 1.00 
Income    
Above average1.741.49–2.041.230.98–1.55
Below average1.00 1.00 
Adults in household    
3 or 41.401.06–1.841.180.79–1.78
21.301.09–1.551.050.79–1.38
Living alone1.00 1.00 
Children age <18 years    
00.670.49–0.891.000.66–1.51
10.630.42–0.931.550.91–2.65
≥21.00 1.00 
Language comprehension    
No difficulties3.962.37–6.620.900.41–2.00
With difficulties1.00 1.00 
Disease duration (months since first clinic visit)    
Short (0–24)1.060.89–1.280.860.65–1.13
Moderate (25–55)1.120.93–1.350.760.57–1.00
Long (≥56)1.00 1.00 
Comorbidities    
02.101.66–2.661.481.20–1.83
11.711.33–2.210.990.78–1.25
≥21.00 1.00 
VAS pain (range 0–100 mm)    
Low (0–29)2.291.91–2.761.591.18–2.14
Moderate (30–54)1.401.13–1.730.920.63–1.34
High (55–100)1.00 1.00 
VAS fatigue (range 0–100 mm)    
Low (0–22)2.261.87–2.742.231.65–3.01
Moderate (23–52)1.531.28–1.841.661.21–2.26
High (53–100)1.00 1.00 
VAS health (range 0–100 mm)    
Good (0–15)2.782.28–3.391.931.45–2.58
Moderate (16–40)1.491.25–1.781.060.78–1.44
Poor (41–100)1.00 1.00 
HAQ DI activity limitation score (range 0–3)    
04.093.24–5.162.821.96–4.06
0.1–1.02.442.03–2.921.661.18–2.35
1.1–3.01.00 1.00 
ESES (range 6–60)    
High (37–60)1.761.45–2.142.131.47–3.09
Moderate (25–36)2.952.36–3.705.103.63–7.16
Low (6–24)1.00 1.00 
SSES (range 0–130)    
High (60–130)1.331.08–1.631.080.71–1.65
Moderate (43–59)2.331.72–3.162.341.71–3.20
Low (0–42)1.00 1.00 
mFABQ (range 0–24)    
Low (0–6)2.041.74–2.392.001.58–2.54
High (7–24)1.00 1.00 
Outcome expectations on health (NRS, range 1–10)    
Median or higher (10)2.301.96–2.693.192.37–4.29
Below median (1–9)1.00 1.00 
Outcome expectations on RA symptoms (NRS, range 1–10)    
Median or higher (8–10)2.622.24–3.053.272.47–4.33
Below median (1–7)1.00 1.00 

The contributions of individual independent variables in the logistic regression models are presented in Tables 4 and 5. After adjustments for all other independent variables, low age, higher education, good Swedish language comprehension, low activity limitation, moderate or high self-efficacy, and high outcome expectations of physical activity contributed significantly to the explanation of variation in current HEPA in the IPAQ model with imputed variables (Table 4). Variation in maintained HEPA according to the ESAI model with imputed variables was explained by male sex, low age, low income, low or moderate fatigue, moderate or high self-efficacy, moderate social support, and high outcome expectations of physical activity on both general health and RA symptoms (Table 5).

Table 4. Outcome of logistic regression for current HEPA (IPAQ)*
 Complete casesMultiple imputation
OR95% CIOR95% CI
  1. Odds ratios (ORs) are adjusted for variables in the table and based on complete cases (n = 1,733) and multiple imputation analysis (K = 5) (n = 3,141). Nagelkerke's pseudo R2 is 18.6% for complete cases and the average for the imputed data 17.2%. HEPA = health-enhancing physical activity; IPAQ = International Physical Activity Questionnaire; 95% CI = 95% confidence interval; VAS = visual analog scale; HAQ = Health Assessment Questionnaire; DI = disability index; ESES = Exercise Self-Efficacy Scale; SSES = Social Support for Physical Activity Scale; mFABQ = modified Fear-Avoidance Beliefs Questionnaire; NRS = numeric rating scale; RA = rheumatoid arthritis.
Sex    
Men0.780.59–1.030.900.74–1.10
Women1.00 1.00 
Age, years    
18–340.470.28–0.780.570.37–0.90
35–540.770.55–1.080.800.64–1.04
≥551.00 1.00 
Education    
University1.220.87–1.711.571.24–1.99
College1.070.77–1.491.331.06–1.68
Other0.900.59–1.351.321.00–1.73
Basic1.00 1.00 
Income    
Above average1.070.82–1.391.000.83–1.22
Below average1.00 1.00 
Adults in household    
3 or 41.130.74–1.711.130.83–1.53
21.010.76–1.351.050.86–1.28
Living alone1.00 1.00 
Children age <18 years    
00.940.61–1.470.930.65–1.35
10.650.40–1.070.700.46–1.07
≥21.00 1.00 
Language comprehension    
No difficulties4.831.60–14.632.401.38–4.19
With difficulties1.00 1.00 
Disease duration (months since first clinic visit)    
Short (0–24)1.140.86–1.510.990.81–1.22
Moderate (25–55)1.210.92–1.611.110.91–1.36
Long (≥56)1.00 1.00 
Comorbidities    
00.990.65–1.501.270.97–1.65
10.980.64–1.511.270.96–1.68
≥21.00 1.00 
VAS pain (range 0–100 mm)    
Low (0–29)0.890.57–1.371.040.76–1.42
Moderate (30–54)0.930.65–1.351.110.86–1.44
High (55–100)1.00 1.00 
VAS fatigue (range 0–100 mm)    
Low (0–22)1.250.85–1.821.130.86–1.49
Moderate (23–52)1.200.88–1.631.100.88–1.37
High (53–100)1.00 1.00 
VAS health (range 0–100 mm)    
Good (0–15)1.280.79–2.051.210.85–1.71
Moderate (16–40)0.970.68–1.390.960.75–1.24
Poor (41–100)1.00 1.00 
HAQ DI activity limitation score (range 0–3)    
02.111.37–3.262.321.70–3.17
0.1–1.01.661.23–2.251.791.45–2.22
1.1–3.01.00 1.00 
ESES score (range 6–60)    
High (37–60)1.581.21–2.061.471.20–1.81
Moderate (25–36)2.531.88–3.411.911.50–2.45
Low (6–24)1.00 1.00 
SSES score (range 0–130)    
High (60–130)1.280.98–1.691.120.91–1.39
Moderate (43–59)2.081.52–2.831.551.13–2.13
Low (0–42)1.00 1.00 
mFABQ (range 0–24)    
Low (0–6)1.230.95–1.591.150.95–1.39
High (7–24)1.00 1.00 
Outcome expectations on health (NRS, range 1–10)    
Median or higher (10)0.910.68–1.221.120.90–1.38
Below median (1–9)1.00 1.00 
Outcome expectations on RA symptoms (NRS, range 1–10)    
Median or higher (8–10)1.671.27–2.201.581.30–1.93
Below median (1–7)1.00 1.00 
Table 5. Outcome of a logistic regression for maintained HEPA (ESAI)*
 Complete casesMultiple imputation
OR95% CIOR95% CI
  1. Odds ratios (ORs) are adjusted for variables in the table and based on complete cases (n = 1,719) and multiple imputation analysis (K = 5) (n = 2,970). Nagelkerke's pseudo R2 is 21.8% for complete cases and the average for the imputed data 15.2%. HEPA = health-enhancing physical activity; ESAI = Exercise Stage Assessment Instrument; 95% CI = 95% confidence interval; VAS = visual analog scale; HAQ = Health Assessment Questionnaire; DI = disability index; ESES = Exercise Self-Efficacy Scale; SSES = Social Support for Physical Activity Scale; mFABQ = modified Fear-Avoidance Beliefs Questionnaire; NRS = numeric rating scale; RA = rheumatoid arthritis.
Sex    
Men1.110.71–1.711.381.03–1.85
Women1.00 1.00 
Age, years    
18–342.801.47–5.351.911.10–3.32
35–541.570.97–2.511.280.89–1.85
≥551.00 1.00 
Education    
University0.910.54–1.531.120.78–1.60
College0.930.54–1.601.080.75–1.56
Other0.730.36–1.490.920.58–1.46
Basic1.00 1.00 
Income    
Above average0.700.47–1.040.680.51–0.91
Below average1.00 1.00 
Adults in household    
3 or 40.790.43–1.460.850.54–1.32
20.850.54–1.330.890.65–1.21
Living alone1.00 1.00 
Children age <18 years    
01.720.92–3.191.490.90–2.47
11.680.84–3.371.690.96–2.99
≥21.00 1.00 
Language comprehension    
No difficulties0.440.08–2.280.530.22–1.27
With difficulties1.00 1.00 
Disease duration (months since first clinic visit)    
Short (0–24)0.730.48–1.100.760.56–1.02
Moderate (25–55)0.780.52–1.180.740.55–1.00
Long (≥56)1.00 1.00 
Comorbidities    
00.960.45–2.020.910.57–1.44
10.840.38–1.850.750.46–1.23
≥21.00 1.00 
VAS pain (range 0–100 mm)    
Low (0–29)0.940.46–1.900.900.54–1.51
Moderate (30–54)0.770.41–1.450.800.52–1.24
High (55–100)1.00 1.00 
VAS fatigue (range 0–100 mm)    
Low (0–22)1.540.86–2.751.731.12–2.67
Moderate (23–52)1.470.88–2.461.711.19–2.47
High (53–100)1.00 1.00 
VAS health (range 0–100 mm)    
Good (0–15)0.950.46–1.980.780.46–1.33
Moderate (16–40)0.840.46–1.540.700.45–1.07
Poor (41–100)1.00 1.00 
HAQ DI activity limitation score (range 0–3)    
00.930.48–1.821.530.94–2.48
0.1–1.00.810.47–1.421.280.86–1.90
1.1–3.01.00 1.00 
ESES score (range 6–60)    
High (37–60)2.051.15–3.621.811.24–2.66
Moderate (25–36)4.702.77–7.963.422.37–4.92
Low (6–24)1.00 1.00 
SSES score (range 0–130)    
High (60–130)0.760.46–1.260.860.55–1.35
Moderate (43–59)1.761.12–2.771.461.03–2.07
Low (0–42)1.00 1.00 
mFABQ (range 0–24)    
Low (0–6)1.731.15–2.621.210.91–1.61
High (7–24)1.00 1.00 
Outcome expectations on health (NRS, range 1–10)    
Median or higher (10)2.041.14–3.661.701.19–2.44
Below median (1–9)1.00 1.00 
Outcome expectations on RA symptoms (NRS, range 1–10)    
Median or higher (8–10)1.701.03–2.821.661.19–2.34
Below median (1–7)1.00 1.00 

The logistic regression models, including imputed variables, demonstrated significant overall correlations between the independent variables and current HEPA (Nagelkerke's R2 = 0.18) and to maintained HEPA (Nagelkerke's R2 = 0.22). The model of current HEPA overall correctly classified 68.7% of cases as adherent or nonadherent (71.8% with complete cases), while the model of maintained HEPA overall correctly classified 89.1% of cases (89.6% with complete cases).

DISCUSSION

The need for studies using valid and reliable outcome measures to map free-living physical activity in large samples of persons with RA has been highlighted in a recent review ([14]). To our knowledge, our study is the first to explore not only current, but also maintained HEPA in a large well-defined sample of persons with RA.

The small proportion (21%) of our respondents reporting maintained aerobic physical activity might indicate that previous reports, although indicating that people with RA are less physically active compared to healthy controls or normative data ([14]), have overestimated the proportion of adherent persons since their data collection focused entirely on current performance. Further, only approximately one-tenth of our respondents reported aerobic physical activity and strength training in line with present guidelines ([8-10]), which may not be surprising but should nevertheless be of concern considering their potential role in improving functioning and reducing cardiovascular disease in this particular population ([1, 37]).

Crude ORs for current HEPA in our study confirmed previous findings of correlations with demographic factors ([13]), while they seemed less important for maintained HEPA. Our multivariate models identified self-efficacy for physical activity as the most salient and consistent factor explaining HEPA variation, while social support and outcome expectations of physical activity contributed consistently, but to a lesser extent, in all models. Activity limitation according to the HAQ DI contributed significantly in the models of current HEPA, but not in the models of maintained HEPA. The roles of age and language comprehension were unclear, probably due to the small numbers of individuals in one of their respective categories.

Self-efficacy has previously been identified as a significant explanatory factor of physical activity variation in arthritis ([20]) and as a mediator in general populations ([38]), as have positive beliefs and high motivation in people with arthritis ([21]). While this is in line with our findings, the overall absence of contribution of pain, fatigue, and general health perception to the explanation of HEPA variation is in contrast to previous findings ([13, 39]). This may either be due to a relatively smaller variation in disease impact, a possible effect of modern drug treatment ([40]), or to our inclusion of a more comprehensive set of potential explanatory factors, some of which were identified as more salient in our multivariate logistic regression models.

The major strengths of our study include its large, well-defined sample, the inclusion of many potential explanatory factors, the assessment of both current and maintained HEPA, and the transparent management of missing data. Some threats to external and internal validity may, however, deserve attention. Differences between questionnaire responders and nonresponders were small, but systematically skewed towards a poorer health of the latter. This may represent a threat to external validity, indicating that our results are valid for a slightly less affected RA population. However, similar differences have previously been found in register-based samples ([41]) and may be thus difficult to avoid. The use of self-reported physical activity data is subject to several types of bias and the IPAQ and the ESAI represent no exceptions to this. Concerns as to recall bias and individual differences in understanding the various intensity levels of physical activity have been raised ([42, 43]). The rationale for using 2 HEPA assessment methods was to gain knowledge, not only on current HEPA, but also on maintained HEPA, and to contrast the 2 methods to each other. Although the validity of the IPAQ has been questioned ([23, 44, 45]), it seems to discriminate fairly accurately between those performing current aerobic HEPA and those who are not ([44]). Although the modified ESAI used in our study needs further evaluation regarding measurement properties in an RA population, it is a brief, straightforward assessment that leaves little room for misunderstandings. Another limitation of our study is the lack of disease activity data, which were available from the SRQ registers for less than half of our study sample within ±3 months compared to the point in time for questionnaire answers. On the other hand, data on general health perception, pain, fatigue, and activity limitation may represent disease impact fairly well. It should also be recognized that our study hypothesis was developed and tested in the same data set and thus needs to be validated in other data sets.

The low levels of maintained HEPA call for certain measures. First, education of health care providers at basic and advanced levels has to improve and include skills training in supporting behavior change strategies in patients with RA. Methods to increase self-efficacy for HEPA should be used systematically, e.g., by using individual progressive goal setting. Second, new arenas for performance of HEPA within and outside the health care system need to be developed, e.g., by further cooperation with patient organizations ([46]) and with those promoting health and wellness. Third, patient-driven innovations may open up new forms of interactions between health care and patients ([47]), such as web-based tools to support health behavior change towards HEPA.

Despite the comprehensive set of potential explanatory factors and an overall high ability to correctly classify individuals performing current and maintained HEPA, most of the variation remained unexplained in our models. This indicates the complexity of HEPA behavior and the need for an even wider perspective on explanation of its variation in future studies by inclusion of, for example, depression ([48]) and neighborhood environment ([49]). On the other hand, similar levels of explanation have been reported previously ([20]), indicating that merely adding more variables may not solve the problem. Rather, recognizing physical activity behavior as determined by a number of individual and situation-specific variables ([50]) suggests that standardized questionnaires and quantitative analyses should be complemented with qualitative approaches to sufficiently capture this complexity. Future research, including prospective epidemiologic studies focusing HEPA outcome on cardiovascular risk in RA, studies that address the risks of sedentary behavior ([17, 51]) (which cannot automatically be expected to have the same correlates as HEPA behavior), and studies on the implementation of HEPA intervention programs, would be of great value for our understanding and promotion of HEPA in RA populations.

In conclusion, our results indicate that a minority of persons with RA perform maintained HEPA and that psychosocial factors are the most salient and consistent in the explanation of HEPA variation. This calls for new aims and strategies within and outside the health care system, as well as a need for future research applying a more holistic approach on HEPA understanding, promotion, and performance.

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 submitted for publication. Dr. Demmelmaier 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. Demmelmaier, Opava.

Acquisition of data. Nordgren, Opava.

Analysis and interpretation of data. Demmelmaier, Bergman, Nordgren, Jensen, Opava.

Acknowledgments

We would like to thank the rheumatology clinics at Danderyd Hospital, Stockholm, Karolinska University Hospital, Solna and Huddinge, Linköping University Hospital, Linköping and Norrköping, Mälarsjukhuset, Eskilstuna, Östersund Hospital, Sunderby Hospital, Luleå, and the SRQ registers for generously providing data for our study.

Ancillary