Self-determined motivation and physical activity in patients with rheumatoid arthritis: a cross-sectional study

Abstract

Objective

To identify factors associated with two self-reported measures of physical activity (PA) in patients with rheumatoid arthritis (RA).

Method

Hospital outpatients with RA from central Norway filled in questionnaires about symptoms, psychological factors, and PA. Outcomes were two alternative self-reported measures of PA: (i) fulfilling the aerobic PA recommendations of ≥ 150 min/week at moderate intensity or ≥ 75 min/week at vigorous intensity; or (ii) being in the PA maintenance stage of the Stages of Exercise Behaviour Change framework. Logistic regression was applied to identify factors associated with PA. Step 1 included the independent variables sex, age, and smoking habits. Step 2a added self-reported function, joint pain during the past 6 months, and fatigue to Step 1. Step 2b added Exercise Self-Efficacy and the Relative Autonomy Index (RAI), calculated from the Behavioural Regulation in Exercise Questionnaire-2, to Step 1. Step 3 included all the mentioned independent variables. Steps 1–3 were analysed for each PA measure.

Results

In total, 227 patients participated. The RAI had a statistically significant positive association with being physically active according to both PA definitions. Joint pain had a significant negative association with meeting the aerobic PA recommendations but was not associated with being in the PA maintenance stage.

Conclusion

The degree of self-determined motivation was the most consistent variable associated with self-reported PA behaviour. Joint pain was associated with one of the two PA measures. Motivation and joint pain may be useful targets for intervention in clinical practice to improve PA engagement among patients with RA.

Regular physical activity (PA) is important for health both in the general population and among people with rheumatoid arthritis (RA) (1). PA is an important part of self-management in patients with RA and is critical to manage their increased risk of cardiovascular disease (1–4). The PA recommendations from the American College of Sports Medicine (ACSM) and American Heart Association (AHA) for the general population, to perform ≥ 150 min of moderate-intensity PA per week or ≥ 75 min of vigorous-intensity PA per week, or a combination, apply to patients with RA as well (1, 5). However, patients with RA engage in less PA than the general population, especially at high intensity, and less often meet the PA recommendations (6–8). There are many available PA measures and different definitions of being physically active. These definitions may capture an overlapping group of individuals, but are likely to give different estimates of the proportion of active individuals and factors associated with being active.

Factors that have been associated with low PA engagement among patients with RA include functional disability, pain, fatigue, and low self-efficacy (9–12). Self-efficacy is the belief in one’s ability to perform a specific behaviour or task under different circumstances (13). Motivational regulation style has also been associated with PA in patients with RA (14). According to Self-Determination Theory, there are several types of motivational regulation styles, which all lie along a self-determination continuum (15). At one end of the continuum is amotivation, when one has no intention to act. Next are four forms of extrinsic motivation – external regulation, introjected regulation, identified regulation, and integrated regulation – over which there is an increasing degree of self-determination and gradually lower importance of external factors. At the other end of the continuum is intrinsic regulation. When intrinsically motivated, one engages in the behaviour because one values it and does not rely on external factors.

The Transtheoretical model is another framework that suggests that individuals are in different stages related to a specific behaviour based on a combination of intention and action (16). Stages of Exercise Behaviour Change consist of precontemplation (not participating in regular PA and no intention to start in the next 6 months), contemplation (not participating in regular PA but intention to start in the next 6 months), preparation (started to engage in some PA, but not regularly), action (started to participate in regular PA in the last 6 months), and maintenance (participated in regular PA for more than 6 months) (17).

The aim of the present study was to identify disease-related and psychological factors associated with PA in people with RA using two different self-reported PA measures: (i) meeting the aerobic PA recommendations or (ii) being in the maintenance stage according to the Stages of Exercise Behaviour Change. We hypothesized that psychological factors were of greater importance for PA than disease-related factors, and that both PA measures would be associated with motivational regulation style and self-efficacy for exercise.

Method

The present study is part of FysKond2, a cross-sectional study investigating the relationship between patient-reported outcome measures, physical fitness, and PA in people with inflammatory arthritis. The data collection process has been described previously (18). In brief, patients fulfilling the American College of Rheumatology (ACR)/European Alliance of Associations for Rheumatology (EULAR) 2010 criteria for RA were recruited from two Norwegian hospitals (19). We contacted patients with appointments at the Rheumatology Outpatient Clinic at Levanger Hospital in 2019 or at St. Olavs Hospital in 2021. We also contacted a random selection of people following the patient initiative follow-up programme at St. Olavs Hospital. These are patients with well-controlled disease who are followed by their general practitioner and can contact the Rheumatology Department if there is an increase in disease activity. Patients with a physical visit received the questionnaire before or after their appointment. Patients with a digital appointment and those in the patient initiative follow-up programme received the questionnaire by mail together with a return envelope. Those who agreed to participate filled in questionnaires about PA habits, psychological factors, and physical symptoms.

The PA questionnaire used in the present study was first introduced in the Trøndelag Health Study and consists of three questions about frequency, intensity, and duration of habitual PA (20). The question ‘How often do you exercise?’ has the options ‘never’, ‘less than once a week’, ‘once a week’, ‘two to three times per week’, and ‘almost daily’. The question ‘How hard do you push yourself (on average)?’ has the options ‘do not get out of breath or break a sweat’, ‘get out of breath and sweaty’, or ‘to exhaustion’. The question ‘How long does each session last (on average)?’ has the options that the average PA session lasts ‘less than 15 minutes’, ‘15 to 29 minutes’, ‘30 to 59 minutes’, and ‘more than 60 minutes’. These three questions were used to calculate whether an individual meets the ACSM/AHA aerobic PA recommendations (1, 5).

The participants also filled in the Stages of Exercise Behaviour Change questionnaire (17). This questionnaire classifies participants into different stages of exercise behaviour change by self-rating their level of agreement with different statements, each of which corresponds to a stage, using a 1–5 Likert scale. To investigate factors associated with maintained PA, the participants were further categorized as either being in the maintenance stage (yes) or not being in the maintenance stage (no, combining all the other stages into one group). Maintained PA is associated with many health benefits and requires consistent effort from the individual for a prolonged period (21). Different factors may be important to maintaining regular PA compared to starting to engage in PA.

The Exercise Self-Efficacy scale, published by Marcus et al, was applied to assess the participants’ confidence in their ability to engage in regular PA under different circumstances (when tired, in a bad mood, pressed for time, on vacation, or when it rains or snows) (17). The five items in the questionnaire are rated on a 1–7 Likert scale and are added together to produce an overall score with the range 5–35. Higher scores represent a higher degree of confidence in one’s ability to engage in exercise.

The degree of self-determination for exercise was assessed with the Behaviour Regulation in Exercise Questionnaire-2 (BREQ-2) (22). This questionnaire consists of 19 items related to amotivation (four items, e.g. ‘I don’t see the point in exercising’), external regulation (four items, e.g. ‘I exercise because other people say I should’), introjected regulation (three items, e.g. ‘I feel ashamed when I miss an exercise session’), identified regulation (four items, e.g. ‘It’s important to me to exercise regularly’), and intrinsic regulations (four items, e.g. ‘I enjoy my exercise sessions’). Each item is rated on a Likert scale from 0 (not true for me) to 4 (very true for me). The average rating for the items corresponding to each regulatory style is the subscore for this style. The Relative Autonomy Index (RAI) is calculated from the subscores and is a summary score that can be interpreted as the degree of self-determination for exercise. The RAI is calculated by multiplying the subscores with the following weighing: 3 × intrinsic regulation, 2 × identified regulation, −1 × introjected regulation, −2 × external regulation, and −3 × amotivation, and then adding these five numbers.

The participants rated their level of joint pain during the past 6 months and level of fatigue on 0–10 Likert scales. They answered the Patient Global Assessment by rating the level of disease activity when taking all symptoms into account on a 0–100 visual analogue scale. Higher scores indicate more symptoms. Self-reported function was assessed with the modified Health Assessment Questionnaire (mHAQ), giving an overall score of 0.00–3.00, where a higher score indicates more disability (23).

Data related to the rheumatic disease were collected from hospital medical records. Cigarette smoking habits were self-reported and categorized as ever-smoker or never-smoker.

Statistics

Normality was assessed with histograms and the Shapiro–Wilk test. As most variables were not normally distributed, continuous data are presented as the median with 25th and 75th percentiles and compared with the Mann–Whitney U test. Categorical data are presented as number with percentage and compared with Pearson’s chi-squared test. Statistical analyses were performed in Stata version 16 (StataCorp, College Station, TX, USA). A p-value < 0.05 was considered statistically significant.

Multivariable logistic regression was used to evaluate factors associated with PA engagement. There were two dependent variables, and we performed a similar analysis for each of them. The first dependent variable was meeting the ACSM/AHA aerobic PA recommendations. The second dependent variable was being in the maintenance stage according to the Stages of Exercise Behaviour Change.

The analysis was performed in steps following a prespecified plan. Step 1 included the background variables age, sex, and smoking status (ever-smoker vs never-smoker). Steps 2a and 2b built on Step 1. Step 2a added the disease-related variables mHAQ, joint pain during the past 6 months, and fatigue to Step 1. Step 2b added the psychological variables Exercise Self-Efficacy and the RAI to Step 1. Finally, in Step 3, all of the variables were included, namely age, sex, smoking status, mHAQ, joint pain past 6 months, fatigue, Exercise Self-Efficacy, and the RAI.

Model fit was evaluated with the Hosmer–Lemeshow test and the overall chi-squared test. Different models were compared with the Akaike Information Criterion and Bayesian Information Criterion. The assumption of linear logits was tested graphically and with the Box–Tidwell test.

For the main analysis, missing data for the independent variables were imputed using multiple imputation with chained equations assuming data missing at random (N = 50 imputed data sets). Complete case analysis, i.e. excluding participants with incomplete data, was performed as a sensitivity analysis.

To visualize the distributions of the continuous variables by PA status, box plots were created for the continuous variables associated with either of the PA measures in the Step 3 analyses.

Ethics

The study was approved by the Regional Committee for Medical and Health Research Ethics (project number 23420) and followed the principles of the Declaration of Helsinki. All participants provided written informed consent.

Results

As shown in Figure 1, 227 people with RA were included in the study. The rate of acceptance to participate among those invited was 61%. The participants were somewhat younger (median age 64 years vs 66 years) and less often female (67% vs 78%) than the non-participants. [A more detailed comparison is presented elsewhere (18).]

Figure 1. Flowchart of inclusion in the study.

Data regarding the fulfilment of ACSM/AHA aerobic PA recommendations and PA maintenance stage status were missing for three (1%) and 40 (18%) participants, respectively. The relatively high percentage of missingness for the maintenance status was caused by several participants endorsing several stages, for example action and maintenance, who therefore could not be classified further.

Among the participants with available data, 28% met the ACSM/AHA aerobic PA recommendations and 58% were classified in the maintenance stage after the Stages of Exercise Behaviour Change. Table 1 presents the relationship between the two PA measures for those with data for both. Most of the participants who fulfilled the aerobic PA recommendation were in the PA maintenance stage (91%). However, more than half of those in the PA maintenance stage did not fulfil the PA recommendation (55%).

Table 1. Cross-table of the physical activity measures.

In the PA maintenance stage^‡	Meeting the aerobic PA recommendations^†
	Yes	No	Total
Yes	49 (26)	61 (33)	110 (59)
No	5 (3)	71 (38)	76 (41)
Total	54 (29)	132 (71)	186 (100)

Data are shown as n (%) for the participants with data for both measures.

^†According to the physical activity (PA) recommendations published by the American College of Sports Medicine and the American Heart Association (5).

^‡In the maintenance stage according to Stages of Exercise Behaviour Change (17).

Participant characteristics and the PA-related psychological variables are presented in Tables 2 and 3, respectively. The participants who met the aerobic PA recommendations and/or were in the PA maintenance stage had lower body mass index, less fatigue, higher Exercise Self-Efficacy, and higher score for the RAI. Furthermore, those who met the PA recommendations had better self-reported function, less joint pain, and a higher frequency of seronegative disease, and those in the PA maintenance stage were more often never-smokers.

Table 2. Participant characteristics.

	Total	Meeting the PA recommendations^†			In the PA maintenance stage^‡
	(N = 227)	Yes (N = 63)	No (N = 161)	p	Yes (N = 110)	No (N = 77)	p
Female sex	153 (67)	38 (60)	25 (70)	0.16	71 (65)	56 (73)	0.24
Age (years)	64 (53, 71)	61 (49, 68)	64 (55, 72)	0.18	61 (53, 69)	65 (55, 72)	0.26
Ever-smoker	150 (66)	37 (59)	110 (68)	0.17	64 (58)	60 (78)	0.005
BMI (kg/m^2)	26.0 (23.3, 28.9)	24.9 (22.3, 27.7)	26.5 (23.9, 29.4)	0.007	25.0 (22.5, 27.8)	26.7 (24.3, 30.0)	0.004
Seropositive (RF and/or ACPA)	187 (82)	46 (73)	138 (86)	0.03	87 (79)	64 (83)	0.49
Duration of RA (years)	11 (5, 20)	11 (6, 24)	11 (5, 18)	0.28	11 (5, 20)	11 (5, 18)	0.74
Using conventional DMARDs	186 (82)	54 (86)	129 (80)	0.33	89 (81)	65 (84)	0.54
Using biological DMARDs	116 (51)	37 (59)	78 (48)	0.17	60 (55)	35 (45)	0.22
mHAQ (scale 0.00–3.00)	0.25 (0.00, 0.63)	0.25 (0.00, 0.50)	0.38 (0.06, 0.63)	0.04	0.25 (0.00, 0.50)	0.38 (0.00, 0.75)	0.09
Joint pain (scale 0–10)	4 (2, 6)	3 (2, 5)	5 (2, 7)	0.002	4 (2, 5)	4 (2, 7)	0.05
Fatigue (scale 0–10)	5 (2, 7)	5 (2, 7)	5 (2, 8)	0.02	5 (2, 7)	6 (3, 8)	0.04

Data are shown as n (%) or median (25th and 75th percentiles). Compared with Pearson’s chi-squared test or Mann–Whitney U test.

Missing data < 3%, except for physical activity maintenance stage status, 18% (n = 40). The percentage of those with available data for each variable is presented.

^†According to the aerobic physical activity recommendations published by the American College of Sports Medicine and the American Heart Association (5).

^‡In the maintenance stage according to Stages of Exercise Behaviour Change (17).

ACPA, anti-citrullinated protein antibody; BMI, body mass index; DMARD, disease-modifying anti-rheumatic drug; mHAQ, modified Health Assessment Questionnaire; PA, physical activity; RA, rheumatoid arthritis; RF, rheumatoid factor.

Table 3. Physical activity-related variables.

	Total	Meeting the PA recommendations^†			In the PA maintenance stage^‡
	(N = 227)	Yes (N = 63)	No (N = 161)	p	Yes (N = 110)	No (N = 77)	p
Exercise Self-Efficacy (scale 5–35)	24 (17, 29)	29 (20, 33)	22 (17, 27)	< 0.0001	26 (20, 31)	22 (17, 27)	0.002
RAI (scale −24, 20)	12 (7, 15)	15 (12, 17)	10 (6, 14)	< 0.0001	13 (10, 16)	8 (5, 11)	< 0.0001
Subscores BREQ-2 (scale 0.0–4.0)
Amotivation	0.0 (0.0, 0.3)	0.00 (0.0, 0.0)	0.00 (0.0, 0.5)	< 0.0001	0.00 (0.0, 0.0)	0.00 (0.0, 0.5)	0.003
External regulation	0.0 (0.0, 0.5)	0.0 (0.0, 0.3)	0.00 (0.0, 0.8)	0.02	0.00 (0.0, 0.5)	0.3 (0.0, 0.8)	0.02
Introjected regulation	1.3 (0.7, 1.7)	1.3 (0.3, 2.0)	1.3 (0.3, 1.7)	0.09	1.3 (0.7, 2.0)	1.3 (0.7, 1.3)	0.19
Identified regulation	2.8 (2.0, 3.3)	3.5 (2.8, 3.8)	2.5 (1.8, 3.0)	< 0.0001	3.0 (2.5, 3.5)	2.0 (1.5, 2.8)	< 0.0001
Intrinsic regulation	2.8 (2.0, 3.5)	3.5 (3.0, 4.0)	2.5 (1.8, 3.0)	< 0.0001	3.0 (2.5, 3.8)	2.0 (1.5, 3.0)	< 0.0001

Data are shown as n (%) or median (25th and 75th percentile). Compared with Pearson’s chi-squared test or Mann–Whitney U test.

Missing data < 3%, except for physical activity maintenance stage status, 18% (n = 40), and the RAI, 9% (n = 20). The percentage of those with available data for each variable is presented.

^†According to the aerobic physical activity recommendations published by the American College of Sports Medicine and the American Heart Association (5).

^‡In the maintenance stage according to Stages of Exercise Behaviour Change (17).

BREQ-2, Behavioural Regulation in Exercise Questionnaire-2; PA, physical activity; RAI, Relative Autonomy Index.

There were no violations of the logistic regression assumptions. The results from the logistic regression analyses are presented in Table 4. The two variables associated with fulfilling the ACSM/AHA aerobic PA recommendations in Step 3 were the RAI and joint pain. Moreover, the only variable that was associated with being in the PA maintenance stage in Step 3 was the RAI. Figure 2 illustrates the distribution of the RAI and joint pain according to the two PA measures.

Figure 2. Box plots of the Relative Autonomy Index and joint pain during the past 6 months. Groups were compared with the Mann–Whitney U test. Physical activity (PA) recommendations follow the American College of Sports Medicine/American Heart Association, and physical activity maintenance stage follows the Stages of Exercise Behaviour Change. (A) Relative Autonomy Index by fulfilment of the aerobic physical activity recommendations. (B) Relative Autonomy Index by physical activity maintenance stage status. (C) Joint paint past 6 months by fulfilment of the aerobic physical activity recommendations. (D) Joint pain past 6 months by physical activity maintenance stage status. The horizontal line within the boxes represents the median. The lower and upper boxes represent the 25th and 75th percentiles, respectively. The lower and upper whiskers are the minimum and maximum values, and the dots outside the whiskers are the outliers, which lie more than 1.5 × the interquartile range away from the 25th or 75th percentile.

Table 4. Results from logistic regression analysis.

	Step 1	Step 2a	Step 2b	Step 3
	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Dependent variable: Fulfilling the ACSM/AHA aerobic physical activity recommendations
Age	0.99 (0.97, 1.01)	0.98 (0.96, 1.01)	0.98 (0.96, 1.01)	0.98 (0.95, 1.01)
Female sex	0.60 (0.32, 1.11)	0.70 (0.36, 1.35)	0.43 (0.21, 0.90)*	0.50 (0.23, 1.10)
Ever-smoker	0.65 (0.35, 1.22)	0.71 (0.37, 1.36)	1.04 (0.50, 2.15)	1.05 (0.49, 2.23)
mHAQ		1.30 (0.47, 3.60)		1.46 (0.43, 4.93)
Joint pain past 6 months		0.84 (0.71, 0.99)*		0.80 (0.65, 0.98)*
Fatigue		0.93 (0.81, 1.07)		1.02 (0.86, 1.21)
Exercise Self-Efficacy			1.05 (1.00, 1.10)	1.05 (0.99, 1.10)
RAI			1.30 (1.17, 1.44)**	1.30 (1.17, 1.44)*
Dependent variable: Being in the physical activity maintenance stage according to Stages of Exercise Behaviour Change
Age	1.00 (0.97, 1.02)	0.99 (0.97, 1.02)	1.00 (0.97, 1.02)	0.99 (0.97, 1.02)
Female sex	0.57 (0.29, 1.11)	0.64 (0.32, 1.28)	0.54 (0.26, 1.13)	0.55 (0.25, 1.20)
Ever-smoker	0.37 (0.19, 0.73)*	0.39 (0.19, 0.80)*	0.52 (0.24, 1.12)	0.53 (0.24, 1.14)
mHAQ		1.09 (0.44, 2.71)		1.31 (0.47, 3.60)
Joint pain past 6 months		0.91 (0.79, 1.06)		0.87 (0.73, 1.02)
Fatigue		0.95 (0.83, 1.08)		1.01 (0.88, 1.17)
Exercise Self-Efficacy			1.01 (0.96, 1.06)	1.00 (0.96, 1.05)
RAI			1.19 (1.10, 1.28)**	1.20 (1.11, 1.29)**

Number of participants included: 224 in analysis with meeting the aerobic physical activity recommendations as the dependent variable, 187 in analysis with being in the physical activity maintenance stage as the dependent variable.

ACSM, American College of Sports Medicine; AHA, American Heart Association; CI, confidence interval; mHAQ; modified Health Assessment Questionnaire; OR, odds ratio; RAI, Relative Autonomy Index.

*p < 0.05; **p < 0.001.

The results of the sensitivity analysis that excluded participants with missing data for the independent variables also showed that the RAI was significantly associated with fulfilling the ACSM/AHA aerobic PA recommendations and with being in the PA maintenance stage. In contrast to the main analyses, joint pain was not significantly associated with fulfilling the aerobic PA recommendations, whereas Exercise Self-Efficacy was significant in the complete case analyses.

Discussion

The main finding of this study was that the degree of self-determined motivation for exercise measured with the RAI had the most robust association with PA. Pain was associated with fulfilling the ACSM/AHA aerobic PA recommendations, but not with being in the maintenance stage for PA. These findings confirmed our hypothesis that psychological factors had a stronger association with PA than disease-related factors, although only one of the two psychological variables was significantly associated with PA in the main analyses.

Psychology and physical activity

The RAI had a positive association with both measures of PA. This finding adds to the literature on the importance of self-determined motivation for PA engagement among patients with RA (9, 14, 24). The degree of self-determined motivation for exercise has also been associated with cardiorespiratory fitness in this patient group (25). Motivation for exercise can be targeted in exercise intervention and in clinical practice. Clinical trials involving a combination of an exercise programme and behavioural change techniques have demonstrated that it is possible to increase the degree of autonomous motivation for exercise among patients with RA (26, 27). These trials applied techniques such as barrier identification, problem solving, identifying personal reasons for engaging in PA, goal-setting, and feedback on progress, all of which may be applied in clinical practice as well.

Exercise Self-Efficacy was not statistically significantly associated with the two PA measures in the main analysis. However, Exercise Self-Efficacy was associated with meeting the ACSM/AHA aerobic PA recommendations in the sensitivity analysis, with very similar confidence intervals to those in the main analyses (95% confidence intervals in Step 3: 1.01–1.13 vs 0.99–1.10). Exercise Self-Efficacy was also associated with meeting the aerobic PA recommendations when the RAI was removed from the main analysis (data not presented). Thus, self-efficacy for exercise may be an important and relevant factor for PA engagement, although this association was less robust than that with the RAI.

Physical symptoms and physical activity

Joint pain during the past 6 months was negatively associated with meeting the ACSM/AHA aerobic recommendations but was not associated with being in the PA maintenance stage. Pain has been negatively associated with PA among patients with RA in previous investigations as well. In a large international study, pain was associated with physical inactivity (10). Moreover, among Danish patients who fulfilled the PA recommendations in 2010, having more pain at baseline was associated with no longer fulfilling the PA recommendations at follow-up in 2017 (28). Nevertheless, a systematic review did not find a consistent association between pain and PA in people with RA (9). Symptoms such as pain may influence PA intensity, rather than whether one engages in some PA or not, perhaps explaining why patients with RA engage in less PA at vigorous intensity compared to controls (6, 29). Furthermore, some patients believe that pain can be a sign that exercise is harmful for their joints (30). Proper patient education around PA and exercise, as well as adequate pain management, is therefore important (1, 2).

Fatigue and function were not associated with PA in the present study. Both fatigue and function have been associated with PA in some previous studies, although not in all (9–11, 31). Our finding of non-significant associations could be related to the statistical models. We included the variables related to symptoms of RA in the same steps. The pain, fatigue, and function variables may capture overlapping aspects of the impact of RA on PA. Nevertheless, excluding pain, fatigue, or mHAQ from the analyses did not substantially change the association between the other variables and PA (data not presented). The patients in the present study had relatively low fatigue levels and high function. Perhaps the associations would had been significant in a sample of patients with more fatigue or a higher degree of functional disability. However, the mHAQ scores in our participants were similar to those reported in Norwegian patients with RA overall (32). Past studies have also used many different measures of fatigue, function, and PA in their analyses, which could explain the differences in their findings compared to the results of the present study (9, 31).

Physical activity measures

There was a substantial difference in how many participants were classified as active in terms of fulfilling the aerobic PA recommendations or being in the PA maintenance stage (28% vs 58%). This could be due the more open definition of being physically active in the Stages of Exercise Behaviour Change questionnaire. Some individuals may go for casual walks every week and therefore consider themselves to be engaging in regular PA without meeting the recommended PA level. Including a detailed description of what it means to be regularly active in the Stages of Exercise Behaviour Change Questionnaire generally leads to fewer participants being classified in the maintenance category (33). More detailed answer options for the frequency question may also have given slightly different estimates of the percentage meeting the PA recommendations.

Previous estimates of the percentage of patients with RA who meet the aerobic PA recommendations have varied greatly. A study of two Scandinavian cohorts reported that only 16% of Danish patients with RA and 48% of Swedish patients with RA met the aerobic PA recommendations (7). Our estimate of the percentage of patients meeting the aerobic PA recommendations were within the expected range and similar to what was reported in a Norwegian study which used the same PA questionnaire (34). Time trends, cultural and geographic factors, recruitment strategies, and the instrument applied to measure PA may explain the wide range of reported estimates of the percentage of patients with RA who engage in regular PA.

The two PA measures were associated with slightly different factors. For example, in the descriptive analysis, seropositivity, poor function, and joint pain were negatively associated with meeting the PA recommendations but not associated with being in the PA maintenance stage. Similarly, pain was only associated with meeting the PA recommendations in the logistic regressions analysis. On the other hand, smoking was only associated with being in the PA maintenance stage in the descriptive analysis and in Steps 1 and 2a. This illustrates that the choice of instrument used to measure PA is important when interpreting the results and comparing them with the literature. The results for each PA measure would probably become more similar if the Stages of Exercise Behaviour Change questionnaire explained that regular PA meant meeting the PA recommendation.

Unfortunately, 40% of the participants could not be classified as being or not being in the PA maintenance stage, which could have influenced the percentage being classified in the maintenance stage. Furthermore, missingness may have influenced the results from the analysis owing to lower statistical power. Missingness may not have been random and the sample with data for maintenance stage status may be somewhat different from the complete sample. We used the original version of the Stages of Exercise Behaviour Change questionnaire, published in 1992 (17). In 2017, when the present study was planned, we were not aware that modifications to the Stages of Exercise Behaviour Change questionnaire had been published, in which the participants have to choose one of the stages themselves or answer ‘true/false’ for each statement. A review found that using these alternative staging algorithms enabled the researchers to classify a larger percentage of participants into specific stages (33). Future studies should use these modified staging algorithms. We chose to use multiple imputation to include all participants with data for each of the dependent variables in the main analysis. This choice was made so that we could include as many of participants as possible in the main analysis.

Both PA measures used in the present study were based on questionnaires and not on objective measurements. This is an important limitation as self-reported PA is prone to more uncertainty than objective measures of PA. An argument for using self-reported PA measures is that it requires less effort of the participants and may result in less selection bias.

Other strengths and limitations

A strength of the study is that the recruitment strategy ensured that patients with both short and long disease duration were invited to participate. It is also a strength that the choice of independent variables was prespecified and not based on significance testing in the descriptive analysis, as choosing independent variables based on theory and past research reduces the risk of overfitting.

Step 3 of the analysis for meeting the ACSM/AHA aerobic PA recommendations included more independent variables than recommended by the rule of thumb of at least 10 observations per event. Including too many independent variables may also lead to overfitting. Although some argue that the rule of 10 may be too conservative, the results from this particular analysis should be interpreted with some caution (35). As the study was cross-sectional, we cannot conclude on causation and the direction of the associations. We did not have a recent measure of disease activity as a large proportion of the patients participated by mail, which made it impossible to investigate the association between a measure of disease activity and PA. Finally, given the characteristics of our sample, the results may not be generalizable to patients with low function or a high degree of fatigue or pain, or those without access to adequate medical treatment.

Conclusion

The degree of self-determined motivation for PA was the most consistent factor associated with PA among people with RA in the present study. Degree of joint pain was associated with one of the two PA measures. In other words, motivation and pain matter for PA engagement in this patient group and may be useful targets in exercise interventions. Clinicians could address these factors by using behavioural change techniques, discussing PA and pain with their patients, adjusting the plan for pain management, or referring to exercise interventions if needed.

Authors contributions

Study conception and design: VV, MH, ISH. Data collection: ISH, MH, VV. Data analysis: ISH, VV. Data interpretation: ISH, VV, MH. Drafting the manuscript: ISH. Critically revising the manuscript: VV, MH. All authors approved the final version of the manuscript.

Acknowledgements

We are very grateful to the patients who participated in this study and to our patient representatives for their valuable insights. We also thank the Rheumatology Departments at Levanger Hospital and St. Olavs Hospital for facilitating data collection.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The work was funded by the Faculty of Medicine and Health Sciences at NTNU – Norwegian University of Science and Technology.