Advertisement

Longitudinal Change in Physical Activity and Its Correlates in Relapsing-Remitting Multiple Sclerosis

Robert W. Motl, Edward McAuley, Brian M. Sandroff

Abstract

Background Physical activity is beneficial for people with multiple sclerosis (MS), but this population is largely inactive. There is minimal information on change in physical activity and its correlates for informing the development of behavioral interventions.

Objective This study examined change in physical activity and its symptomatic, social-cognitive, and ambulatory or disability correlates over a 2.5-year period of time in people with relapsing-remitting multiple sclerosis.

Methods On 6 occasions, each separated by 6 months, people (N=269) with relapsing-remitting multiple sclerosis completed assessments of symptoms, self-efficacy, walking impairment, disability, and physical activity. The participants wore an accelerometer for 7 days. The change in study variables over 6 time points was examined with unconditional latent growth curve modeling. The association among changes in study variables over time was examined using conditional latent growth curve modeling, and the associations were expressed as standardized path coefficients (β).

Results There were significant linear changes in self-reported and objectively measured physical activity, self-efficacy, walking impairment, and disability over the 2.5-year period; there were no changes in fatigue, depression, and pain. The changes in self-reported and objective physical activity were associated with change in self-efficacy (β=.49 and β=.61, respectively), after controlling for other variables and confounders.

Limitations The primary limitations of the study were the generalizability of results among those with progressive multiple sclerosis and inclusion of a single variable from social-cognitive theory.

Conclusions Researchers should consider designing interventions that target self-efficacy for the promotion and maintenance of physical activity in this population.

Multiple sclerosis (MS) is a common, nontraumatic, and chronic disabling disease of the central nervous system (CNS).1 This disease is typically characterized by intermittent and recurrent periods of inflammatory demyelination and transection of axons in the CNS.2,3 Such a presentation is consistent with the relapsing-remitting clinical course of MS (RRMS) that is initially diagnosed in the majority of cases and described by episodes of symptom worsening followed by varying degrees of recovery and stability.4 The axonal damage results in conduction delay and conduction block of action potentials along CNS pathways and manifests as symptoms (eg, fatigue, depression, pain), mobility impairment, and disability.5 Such manifestations present significant barriers for physical activity6 and appear to be associated with prevalent inactivity in this population.7,8 Importantly, there is increasing evidence for the importance of physical activity in MS,9 but beneficial outcomes are contingent on participating in this behavior.

The design and success of programs for increasing physical activity in people with MS depend, in part, on the identification of correlates of physical activity that can become targets of behavioral and self-management interventions. Such correlates ideally should be variables that are modifiable, on the basis of theory, and consistently associated with physical activity. Researchers often have focused on identifying symptoms as correlates of physical activity among people with MS. This focus is because symptoms are a hallmark manifestation of MS5 that can have a profound influence on performance and behavioral outcomes, including physical activity, on the basis of reality and the Theory of Unpleasant Symptoms.10 Symptoms can further provide a barrier for physical activity by influencing social-cognitive variables such as self-efficacy.11 To date, research has indicated that the symptoms of fatigue, depression, and pain are associated with physical activity in people with MS.12 Research has further indicated that such associations might operate through self-efficacy13,14 and walking impairment15 and are independent of a person's disability status.16

The primary limitation of that previous research has been the reliance on a cross-sectional research design. Such a design only permits an inference regarding associations among changes in focal variables over time.17 To date, there is limited research on longitudinal changes in physical activity and associated correlates in MS, yet adopting a longitudinal design is important for several reasons and represents the novel focus of the current research. First, longitudinal designs are necessary for conclusions about relationships involving actual changes in correlates and physical activity over time.18 Second, correlates in cross-sectional data often are more strongly related to physical activity than in longitudinal designs, and frequently cross-sectional correlations do not replicate in longitudinal applications.19 Third, longitudinal designs allow for developing better informed and targeted interventions for changing physical activity.20 Accordingly, studies that examine changes in physical activity and associations with changes in correlates over time are warranted in people with MS.

The current research project adopted a longitudinal research design and examined changes in symptoms, self-efficacy, walking impairment, disability status, and physical activity over a 2.5-year period of time in people with RRMS. We initially examined the trajectory of change in each of the variables, particularly self-reported and objectively measured physical activity, and then examined associations among the changes in correlates and physical activity, controlling for possible confounding variables. We expected a linear reduction in physical activity over time and that worsening of symptomatic fatigue, depression, pain, self-efficacy, or walking impairment would predict the reduction of physical activity over time. We further controlled for possible changes in disability status over time as well as other confounding variables of age, sex, and disease duration. If successful, this research would inform the subsequent development of an intervention that targets the identified correlates for possibly promoting change in physical activity among people with RRMS.

Method

Sample

The data are the primary outcome variables from a recently completed, longitudinal investigation of symptoms and physical activity over 2.5 years in people with RRMS. The sample was recruited through a research advertisement posted on the National MS Society (NMSS) website and distributed through 12 midwestern chapters of the NMSS. Those who were interested in the study contacted the research team by either e-mail or a toll-free telephone call. This contact was followed by a scripted conversation with the project coordinator, who described the study procedures and undertook screening for inclusion criteria. The inclusion criteria were: (1) diagnosis of RRMS confirmed by a physician, (2) relapse-free in the previous 30 days, (3) ambulatory with or without assistance (ie, walk independently or walk with a cane or crutch or walker or rollator), and (4) willingness to complete the study materials every 6 months over 2.5 years. Those who did not satisfy the inclusion criteria were excluded from participation. We successfully contacted 375 of the 463 people who expressed interest in the study, and 6 were uninterested in participation after the description of the study procedures. The remaining 369 people underwent screening, 44 did not satisfy the inclusion criteria, and 5 declined voluntary participation. We sent an informed consent document (completed by the participant) and RRMS verification form (completed by the participant's treating physician) to the remaining 320 people, and 41 did not return the documents despite 3 attempts for follow-up contact. We sent study materials to the remaining 279 people, and 10 subsequently declined further participation; this distribution of materials occurred in 12 waves of about 25 participants per wave beginning in March of 2008 (wave 1) and ending in February of 2009 (wave 12). There were 269 people with RRMS who provided baseline data. Of the initial 269 people, there were 258, 253, 245, 244, and 238 who provided follow-up data 6, 12, 18, 24, and 30 months later (ie, 88%–96% of the initial sample). This attrition involved either a change in the participant's residential address or loss of materials through the US Postal Service.

The baseline sample consisted of 223 women and 46 men. The participants were mostly Caucasian (91%), well educated (83% had some college education or were college graduates), and reported a median household income that exceeded $40,000/year (68%). The mean age was 45.9 years (standard deviation [SD]=9.6), and the mean MS disease duration was 8.8 years (SD=7.0). The median Patient Determined Disease Steps (PDDS) Scale score was 2 (interquartile range=3.0), and the mean 12-item Multiple Sclerosis Walking Scale (MSWS-12) score was 36.0 (SD=28.2). Those scores indicated that the sample, on average, had minimal walking impairment.21,22 There were 223 people who reported being treated with a disease-modifying therapy; interferon β-1a (50%), glatiramer acetate (31%), and interferon β-1b (13%) represented the most common types of therapy. All 269 participants had a diagnosis of RRMS.

Power Analyses

The target sample size of 250 was based on a series of power analyses undertaken with the use of the Monte Carlo study feature in Mplus.23 We specified a latent growth curve model (LGM) with 6 time points, set the reliability of the 6 indicators to 0.90, used values from pilot data for the mean and standard deviation of the initial status factor, specified the correlation between the growth factors to be 0.1, and selected the standard deviation of the slope factor such that 95% of the units would change within ±20% of average initial status (ie, 2 standard deviations). The mean parameter of the slope factor was set to 5% of the standard deviation of the indicator of the first time point, which represents an average 5% of a standard deviation change within 1 time interval. We used sample sizes of 100, 150, 200, and 250 individuals with 500 replications, and the percentage of replications was recorded where the mean parameter of the slope factor was statistically significant. The power for detecting a small, linear decline in physical activity over time was 62.2%, 83.0%, 92.4%, and 95.6% for sample sizes of 100, 150, 200, and 250, respectively.

We then conducted a power analysis for detecting a small, linear increase in symptoms across time, and the power for the mean parameter of the slope factor was 53.0%, 72.8%, 84.4%, and 89.0% for sample sizes of 100, 150, 200, and 250, respectively. We finally conducted a third power study for a model with 2 parallel growth processes23 representing the relationship between changes in symptoms and physical activity over time. This model had 2 initial status factors (1 for symptoms and 1 for physical activity), 2 slope factors (1 for symptoms and 1 for physical activity), and 2 path coefficients (1 between initial status factors and 1 between change factors). The path coefficients explained the correlations among initial status and growth factors. The parameters for each growth process were established identically as in the first 2 power studies, and the values for the 2 standardized path coefficients were 0.3. The minimal power for the path coefficients was 61.4%, 78.0%, 88.8%, and 92.6% for sample sizes of 100, 150, 200, and 250, respectively.

Measures

Physical activity.

Physical activity was measured using ActiGraph model 7164 accelerometers (ActiGraph, Pensacola, Florida), and the short form of the International Physical Activity Questionnaire (IPAQ).24 Researchers have provided evidence for the validity of scores from these measures in people with MS,8,25 and the inclusion of 2 different measures allowed for examining the possible differential correlates of change in self-reported and objectively measured physical activity. The ActiGraph model 7164 accelerometers were worn on an elastic belt around the waist above the nondominant hip during the waking hours, except while showering, bathing, and swimming, for a 7-day period. Waking hours was defined as the moment on getting out of bed in the morning through the moment of getting into bed in the evening. The participants recorded the time that the accelerometer was worn on a log, and this time was verified by inspection of the minute-by-minute accelerometer data. Regarding data processing, we checked the validity of each day's data (10 or more hours of wear time without periods of 60 minutes of continuous zeros) and then summed the minute-by-minute movement counts across each of the valid days and averaged the total daily movement counts across the valid days. This process yielded accelerometer data in total movement counts per day, with higher scores representing more physical activity. The lower bound of scores was 0, and the upper bound of scores was undefined. Importantly, movement counts are different from step counts. Step counts reflect a binary event recorded for each footstep or occurrence of a foot strike during ambulation, whereas movement counts reflect the magnitude or intensity of the binary event recorded for each footstep (ie, the amount of acceleration of the body's center of mass per foot strike during ambulatory physical activity). By extension, movement counts as recorded and expressed in this study reflect the amount of ambulatory physical activity accumulated over the course of the day.

The short-form of the IPAQ was designed for population surveillance of physical activity among adults and has 6 items that measure the frequency and duration of vigorous-intensity activities, moderate-intensity activities, and walking during a 7-day period. We did not include the duration component in this study on the basis of previous research that identified problems with accurate recall of activity duration in people with MS.25 The respective frequency values for vigorous, moderate, and walking activities were multiplied by 8, 4, and 3.3 metabolic equivalents and then summed to form a continuous measure of physical activity. The scores ranged between 0 and 107.

Symptoms.

Fatigue was measured with the Fatigue Severity Scale (FSS).26 The FSS has 9 items that were rated on a 7-point scale that ranged between 1 (strongly disagree) and 7 (strongly agree). The item scores were averaged to form an overall measure of a participant's severity of fatigue symptoms during the past 4 weeks, and FSS scores ranged between 1 and 7. Higher scores reflect more severe symptoms of fatigue. The FSS has good evidence of internal consistency, test-retest reliability, and score validity.26

Pain was measured with the short-form McGill Pain Questionnaire (SF-MPQ).27 This scale has a 15-item adjective checklist that captures sensory and affective dimensions of pain experienced during the past 4 weeks. The items were rated on a 4-point scale that ranged between 0 (none) and 3 (severe). The items were summed to form a composite that ranged between 0 and 45. Higher scores reflect more severe pain. The SF-MPQ is internally consistent, reliable across time, and has evidence of score validity.27

Depressive symptoms were measured by the Hospital Anxiety and Depression Scale (HADS).28 The HADS has 14 items: 7 items measure anxiety and 7 items measure depression. The items were rated on a 4-point scale that ranged between 0 (most of the time) and 3 (not at all). We did not include the 7 items for anxiety because we were only focusing on depressive symptoms as a specific correlate of physical activity. The negatively worded items were reverse-scored, and scores from the 7 items were summed for a composite score of the frequency of depressive symptoms during the previous 4 weeks. The scores ranged between 0 and 21, and higher scores reflect a greater frequency of depressive symptoms. This scale has good evidence of score reliability and validity.28

Self-efficacy.

Self-efficacy was assessed by the Exercise Self-Efficacy Scale (EXSE).29 The EXSE scale has 6 items that assess a person's beliefs relative to engaging in 20+ minutes of moderate physical activity 3 times per week, in 1-month increments, across the next 6 months. The items were rated on a scale from 0 (not at all confident) to 100 (completely confident) and averaged into a composite score that ranges between 0 and 100. Higher scores reflect greater confidence in a person's ability to engage in regular physical activity over time. This scale is internally consistent and has evidence of score validity,29 and it has been included in previous research on physical activity in MS.30

Walking impairment.

The MSWS-12 is a 12-item patient-rated measure of the impact of MS on walking.21 The items are rated on a 5-point scale from 1 (not at all) to 5 (extremely), and the items represent limitations of walking during the previous 2 weeks. The MSWS-12 is scored by summing the 12 item scores, subtracting 12, dividing the difference by 48, and then multiplying by 100. This method of scoring scales the MSWS-12 score between 0 and 100. The MSWS-12 has good evidence for its internal consistency, test-retest reliability, and validity of scores as a measure of walking impairment in MS.21

Disability status.

Disability status was measured with the use of the PDDS Scale.22 The PDDS Scale is a self-report questionnaire for measuring neurological impairment with the use of an ordinal scale of 0 (normal) through 8 (bedridden). This scale was developed as an inexpensive surrogate for the Expanded Disability Status Scale (EDSS), and scores from the PDDS Scale have been reported to be linearly and strongly related to physician-administered EDSS scores (r=.93).22 This scale was included rather than the EDSS because the data were collected entirely through the US Postal Service.

Procedure

After initial telephone contact, screening for inclusion, and return of informed consent and MS verification documentation, participants were sent an accelerometer and battery of questionnaires through the US Postal Service. We further provided prestamped and preaddressed envelopes for return postal service. The project coordinator called to make sure the participants received the materials and understood the instructions. The participants then completed the battery of questionnaires that included measures of symptoms, self-efficacy, walking impairment, disability status, and physical activity and wore the accelerometer for 7 days. After completing the measures and wearing the accelerometer, participants returned the study materials through the US Postal Service. We contacted participants by telephone and e-mail up to 3 times as a reminder to return the study materials. We further collected any missing questionnaire data on the basis of follow-up telephone calls. This same procedure was completed every 6 months over a 2.5-year period of time. All participants received $120 remuneration, which was prorated to be $20 per completion and return of the study materials.

Data Analysis

The data were analyzed with LGM with the use of the full-information maximum likelihood (FIML) estimator and the Mplus software package.23 The LGM is a powerful approach for studying the pattern, predictors, and consequences of longitudinal change processes.18 This approach has a number of advantages over other more commonly adopted approaches used to study change among continuous variables (eg, analysis of variance, multivariate analysis of variance, lagged regression, use of change scores), including the ability to: (1) model change at the individual as well as the group-level of analysis, (2) model individual differences in change trajectories (initial status and slope factors), (3) model change in several focal variables concomitantly, and (4) directly model important predictors and outcomes of longitudinal change.18 Good model-data fit in the LGM analyses was established on the basis of a comparative fit index (CFI) of ≥.95 and standardized root mean residual (SRMR) of ≤.08.31

We initially examined linear changes in the study variables, particularly a reduction in physical activity, through the use of standard linear LGM. This examination involved testing a fixed, linear time series (0, 1, 2, 3, 4, 5, 6) for establishing initial status and rate of change in all variables measured over the 30-month time period. When the rate of change was statistically significant, we estimated the magnitude of change over the entire 30-month period on the basis of Cohen d (absolute difference in 0- and 30-month mean scores divided by baseline standard deviation) and guidelines of 0.2, 0.5, and 0.8 for small, moderate, and large effects, respectively.32

We then examined associations among changes in symptoms, self-efficacy, and walking impairment with changes in physical activity through the use of LGM with parallel growth processes.18,23 The LGM with parallel growth processes included the standard linear LGM (eg, establishing the pattern or trajectory of change in symptoms and physical activity over time) and the addition of path coefficients between initial status and rate of change. The path coefficients are interpreted, for example, as a cross-sectional relationship between symptoms and physical activity (initial status) as well as a longitudinal relationship between changes in symptoms and physical activity over time (rate of change). One final set of LGM analyses involved examining the association between changes in variables after accounting for possible confounding variables of disability status and age, sex, and disease duration. We interpreted the magnitude of the path coefficients as standardized estimates (ie, standardized on a scale of ±1.0) through the use of the guidelines of .1, .3, and .5 for small, moderate, and large coefficients, respectively.32

Role of the Funding Source

This investigation was supported by a grant from the National Multiple Sclerosis Society (RG 3926A2/1).

Results

Standard Latent Growth Curve Modeling: Establishing Changes in Variables

The model-data fit indexes and parameter estimates from the standard linear LGM analyses on all 8 variables are provided in Table 1. The mean scores and standard errors for all of the variables are provided in Table 2, and, importantly, the mean value for accelerometer counts is consistent with previous samples of MS.16,30,33 The 64 × 64 matrix of correlations among scores from the 8 variables over 6 time points can be obtained by contacting the corresponding author.

Table 1.

Model Fit Indexes and Parameter Estimates From the Latent Growth Curve Modeling Analysis of Linear Change in the Study Variables Across 6 Time Points in 269 People With Multiple Sclerosisa

Table 2.

Descriptive Statistics for Each Variable Included in the Latent Growth Curve Modeling Analysis Across 6 Time Points in 269 People With Multiple Sclerosisa

Physical activity.

The linear model had a good fit for the accelerometer data. Initial status was significantly different from zero (P<.0001), and there was a significant, linear decrease in accelerometer counts over time (P<.001). The effect size for the 30-month change (d=0.17) was small in magnitude. There was significant variance around initial status (P<.0001) and group mean change (P<.001). This finding indicates the presence of variability in the trajectory of change in accelerometer counts over time that can be explained by other variables. The slope and intercept were significantly (P=.005) and negatively correlated. This finding indicates that people with higher initial accelerometer counts had less of a reduction in accelerometer counts over the 30-month period.

The linear model similarly had a good fit for the IPAQ data. Initial status was significantly different from zero (P<.0001), and there was a significant, linear decrease in IPAQ scores over time (P<.05). The effect size for the 30-month change (d=0.16) was small in magnitude. There was significant variance around initial status (P<.0001) and group mean change (P<.001). This finding indicates that there was variability in the trajectory of change in IPAQ scores over time that can be explained by other variables. The slope and intercept were significantly (P=.001) and negatively correlated. This finding indicates that people with higher initial IPAQ scores had less of a reduction in IPAQ scores over the 30-month period.

Symptoms.

The linear model had a good fit for the FSS data. Initial status was significantly different from zero (P<.0001), but there was not a significant, linear change over time (P=.70). There was significant variance around initial status (P<.0001) and group mean change (P<.001). The slope and intercept were not significantly correlated (P=.10).

The linear model had a good fit for the SF-MPQ data. Initial status was significantly different from zero (P<.0001), but there was not a significant, linear change over time (P=.06). There was significant variance around initial status (P<.0001) and group mean change (P<.001). The slope and intercept were not significantly correlated (P=.76).

The linear model had a good fit for the HADS data. Initial status was significantly different from zero (P<.0001), but there was not a significant, linear change over time (P=.80). There was again significant variance around initial status (P<.0001) and group mean change (P<.001). The slope and intercept were not significantly correlated (P=.27).

Other variables.

The linear model had a good fit for the EXSE data. Initial status was significantly different from zero (P<.0001), and there was a significant, linear decrease over time (P<.05). The effect size for the 30-month change (d=0.16) was small in magnitude. There was significant variance around initial status (P<.0001) and group mean change (P<.001). The slope and intercept were significantly (P<.01) and negatively correlated, indicating that people with higher initial EXSE scores had less of a reduction in EXSE scores over the 30-month period.

The linear model had a good fit for the MSWS-12 data. Initial status was significantly different from zero (P<.0001), and there was a significant, linear increase over time (P<.05). The effect size for the 30-month change (d=0.08) was very small in magnitude. There was significant variance around initial status (P<.0001) and group mean change (P<.0001). The slope and intercept were not significantly correlated (P=.49).

The linear model had a good fit for the PDDS data. Initial status was significantly different from zero (P<.0001), and there was a significant, linear increase over time (P<.001). The effect size for the 30-month change (d=0.15) was small in magnitude. There was significant variance around initial status (P<.0001) and group mean change (P<.001). The slope and intercept were not significantly correlated (P=.25).

Summary.

There were linear changes in physical activity (accelerometer and IPAQ), self-efficacy (EXSE), walking impairment (MSWS-12), and disability (PDDS Scale) over time. There were not significant linear changes in symptoms (FSS, SF-MPQ, or HADS) over time. The next set of analyses, therefore, examined changes in self-efficacy, walking impairment, and disability (EXSE, MSWS-12, and PDDS Scale) as correlates of change in physical activity (accelerometer and IPAQ) over time; these variables demonstrated change and became the focus for understanding the reduction in physical activity.

Parallel Process Latent Growth Curve Modeling: Correlates of Change in Physical Activity

Accelerometer outcome.

We initially conducted analyses of associations between changes in EXSE, MSWS-12, PDDS Scale, and accelerometer data. The model-data fit indexes and parameter estimates from the parallel process LGM analyses for examining correlates of change in accelerometer data are provided in Table 3. The first model examined the association between changes in EXSE scores and accelerometer data over time and had a good fit to the data. There were significant associations between initial status for EXSE and accelerometer data (P<.0001) and between the linear changes in EXSE and accelerometer data over time (P<.0001). The latter association indicated that a 1-standard deviation unit change in EXSE scores was associated with a 0.50-standard deviation unit change in accelerometer counts over time. The change in EXSE scores explained 25% of variance in accelerometer changes over time. The second model examined the association between changes in MSWS-12 scores and accelerometer data over time. This model had a good fit to the data. There was a significant association between initial status for MSWS-12 and accelerometer data (P<.0001) but not between the linear changes in MSWS-12 and accelerometer data (P=.10). The third model examined the association between changes in PDDS Scale scores and accelerometer data over time and had a good fit to the data. There was a significant association between initial status for PDDS Scale and accelerometer data (P<.0001) but not between the linear changes in PDDS Scale and accelerometer data (P=.07). The last model examined the association between changes in EXSE scores and accelerometer data over time controlling for age, sex, disease duration, and baseline PDDS Scale and MSWS-12 scores. The model had a good fit to the data: χ2 (df=107, N=269)=196.31, P<.0001, CFI=.97, SRMR=.04. The association between the linear changes in EXSE and accelerometer data over time was statistically significant, nearly large in magnitude, and unchanged after accounting for those additional variables (standardized path coefficient=.49, P<.005).

Table 3.

Model Fit Indexes and Parameter Estimates From the Parallel Process Latent Growth Curve Modeling Analysis Examining Correlates of Change in Objectively Measured Physical Activity Among 269 People With Multiple Sclerosisa

IPAQ.

Finally, we conducted analyses of associations between changes in EXSE, MSWS-12, PDDS Scale, and IPAQ scores. The model-data fit indexes and parameter estimates from the parallel process LGM analyses for examining correlates of change in IPAQ data are provided in Table 4. The first model examined the association between changes in EXSE and IPAQ scores over time and had a good fit to the data. There were significant associations between initial status for EXSE and IPAQ scores (P<.0001) and between the linear changes in EXSE and IPAQ scores over time (P<.0001). The later association indicated that a 1-standard deviation unit change in EXSE scores was associated with a 0.60-standard deviation unit change in IPAQ scores over time. The change in EXSE scores explained 36% of variance in IPAQ scores over time. The second model examined the association between changes in MSWS-12 and IPAQ scores over time and had a good fit to the data. There were significant associations between initial status for MSWS-12 and IPAQ scores (P<.0001) and between the linear changes in MSWS-12 and IPAQ scores (P<.005). The latter association indicated that a 1-standard deviation unit change in MSWS-12 scores was associated with a 0.29-standard deviation unit change in IPAQ scores over time. The change in MSWS-12 scores explained 8% of variance in IPAQ scores over time. The third model examined the association between changes in PDDS and IPAQ scores over time. This model had a good fit to the data. There was a significant association between initial status for PDDS and IPAQ scores (P<.0001). There was a significant association between the linear changes in PDDS and IPAQ scores over time (P<.01). The latter association indicated that a 1-standard deviation unit change in PDDS scores was associated with a 0.28-standard deviation unit change in IPAQ scores over time. The change in PDDS scores explained 8% of variance in IPAQ scores over time.

Table 4.

Model Fit Indexes and Parameter Estimates From the Parallel Process Latent Growth Curve Modeling Analysis Examining Correlates of Change in Self-Reported Physical Activity Among 269 People With Multiple Sclerosisa

The last 3 models examined the association between changes in EXSE and IPAQ scores controlling for changes in MSWS-12 scores, PDDS Scale scores, and then age, sex, disease duration, and baseline PDDS Scale and MSWS-12 scores. The model controlling for changes in MSWS-12 scores had a good fit to the data: χ2 (df=148, N=269)=292.57, P<.0001, CFI=.97, SRMR=.04. There still was a significant association between the linear changes in EXSE and IPAQ scores (standardized path coefficient=.53, P<.005); there was not a significant association between the linear changes in MSWS-12 and IPAQ scores (standardized path coefficient=−.07, P=.61) in this model. The model controlling for changes in PDDS scores had a good fit to the data: χ2 (df=148, N=269)=247.72, P<.0001, CFI=.98, SRMR=.04. There still was a significant association between the linear changes in EXSE and IPAQ scores over time (standardized path coefficient=.52, P<.005); there was not a significant association between the linear changes in PDDS and IPAQ scores (standardized path coefficient=−.13, P=.31) in this model. The final model controlling for confounders had a good fit to the data, χ2 (df=148, N=269)=156.65, P<.001, CFI=.98, SRMR=.04. There still was a statistically significant and strong association between the linear changes in EXSE and IPAQ scores (standardized path coefficient=.61, P<.005).

Summary.

The linear change in accelerometer counts was associated with change in self-efficacy (EXSE) but not changes in disability (PDDS Scale) and walking impairment (MSWS-12), and this was unchanged when controlling for confounders. The linear change in IPAQ scores was associated with changes in self-efficacy (EXSE), disability (PDDS Scale), and walking impairment (MSWS-12), but the association between changes in IPAQ and self-efficacy was not accounted for by changes in disability (PDDS Scale) or walking impairment (MSWS-12) or when controlling for confounders.

Discussion

This study documented a significant linear reduction of physical activity over a 2.5-year period of time in people with MS. To our knowledge, this is the first study that documents such a change, as 3 previous studies of people with MS demonstrated no significant change in physical activity over time.3436 For example, one study administered the exercise/physical activity subscale of the Health Promoting Lifestyle Profile II (HPLP-II) annually over a 5-year period in a sample of 611 people with MS and reported no mean change in physical activity.36 Another study included the Physical Activity Scale for Individuals With Physical Disabilities (PASIPD) and a sample that consisted primarily of people with MS and spinal cord injuries but did not document a statistically significant mean change in physical activity over a 12-month period of time.34 The primary difference between the present study and previous research is that we included objective and self-report measures of physical activity with established evidence for the validity of scores in people with MS.8,25 The validity of the physical activity measures included in previous research has not been systematically tested in MS.6 The HPLP-II and PASIDP might not be valid or sensitive for capturing naturally occurring changes in physical activity over time among those with the RRMS. Accordingly, there does appear to be a reduction of physical activity over time in people with RRMS that is captured by validated, objective, and self-report measures. This observation further underscores the importance of identifying correlates of physical activity, when considering that people with MS are typically physically inactive68 and probably becoming more physically inactive over time. When combined, such behavioral patterns probably increase the risk of secondary health conditions such as cardiovascular disease33 and negate the benefits of physical activity for people with MS.9

This study documented significant changes in self-efficacy for exercise, walking impairment, and disability status over a 2.5-year period of time in people with MS; there were not statistically significant changes in symptoms of fatigue, depression, and pain. We further documented that change in self-efficacy for exercise correlated with changes in both objective and self-report measures of physical activity, even when controlling for walking impairment, disability status, and other confounding variables. These findings suggest that self-efficacy is an important correlate of physical activity in people with MS, and such an observation extends existing research.13,14,30,37 For example, cross-sectional research has demonstrated that self-efficacy correlated with physical activity even after controlling for symptoms and disability in people with MS.13,14 One prospective study demonstrated that baseline self-efficacy predicted change in physical activity over a 3-month period of time in a sample of 16 people with MS.37

One unique aspect of this study, which has not previously been reported in the published literature, is the demonstrated change in self-efficacy being associated with change in physical activity over 2.5 years in a large sample of people with MS. Collectively, self-efficacy is emerging as a cross-sectional, prospective, and longitudinal correlate of physical activity in people with MS. We do not specify that self-efficacy is causing physical activity levels as the existing evidence is more consistent with the concept of reciprocal determinism.11 Reciprocal determinism suggests bidirectional associations between personal factors such as self-efficacy and behaviors such as physical activity. Nevertheless, self-efficacy is presumably a modifiable variable, and there are established influences on self-efficacy (eg, mastery, social modeling) that can be targeted on the basis of social-cognitive theory.11 Researchers should consider designing interventions that target self-efficacy for the promotion and long-term maintenance of physical activity behavior in people with MS. This research might not only change physical activity but further inform our understanding of the causal association between self-efficacy and physical activity.

Regarding changes in symptoms and associations with physical activity, the results of the present study were not consistent with our expectations on the basis of previous cross-sectional research.1216 Indeed, cross-sectional research has indicated that symptoms of fatigue, depression, and pain were correlated with physical activity in people with MS.12 This body of research implies that changing such symptoms might be an avenue for changing physical activity in people with MS. Such an inference required verification in a longitudinal analysis and resulted in the current focus on changes in symptoms as correlates of changes in physical activity. To that end, we did not observe statistically significant changes in the measures of fatigue, pain, and depression over the 2.5-year period of time. The lack of changes undermined our capacity for examining changes in symptoms as correlates of changes in physical activity and would not support these variables as targets of an intervention for changing physical activity in MS. The discrepancy between cross-sectional and longitudinal results further highlights the importance of verifying correlates in longitudinal research, as cross-sectional correlates often do not replicate in prospective designs.

There were similarities and differences in the variables that correlated with changes in objective versus self-reported physical activity in this study of people with MS. The change in self-efficacy was associated with changes in both objective and self-reported physical activity. By comparison, changes in walking impairment and disability status were only associated with changes in self-reported physical activity, but such associations were no longer significant when accounting for the change in self-efficacy. This finding might indicate that walking impairment and disability are factors that inform a person's self-efficacy beliefs consistent with previous cross-sectional research.13 The inconsistency in association between walking impairment and disability status with the objective and self-report measures of physical activity might reflect a self-report bias (ie, scores from self-report measures might be correlated simply based on overlapping variance associated with the method of collecting data).

This research might have implications for the future development and testing of approaches for increasing physical activity in people with MS. Such a focus is important because physical activity has many benefits in people with MS,9 but this population has prevalent inactivity68 that seemingly becomes more prominent over time, on the basis of the current study. To that end, we believe that our results point toward the adoption of social-cognitive theory as a backdrop for informing the development and testing of interventions for increasing physical activity in MS; such a recommendation has been made previously.6,7 This recommendation is because self-efficacy is considered an active agent and proximal determinant of behavior change,11 including physical activity,38 and the magnitude of association between changes in self-efficacy and physical activity was quite strong in the present study. There further are factors for targeting a change in self-efficacy levels.11 Such factors include mastery or performance accomplishment, verbal persuasion or social support, vicarious experience or social learning, and interpretation of physiological and affective cues. Importantly, there are several small-scale studies of people with MS that have targeted the enhancement of self-efficacy on the basis of a social-cognitive perspective and reported beneficial changes in exercise adherence39 and physical activity levels40 over short, 12-week periods of time. To date, we are unaware of large-scale studies that have documented the effect of such an approach for changing physical activity over long periods of time (6 or 12 months) with additional beneficial changes in fitness, walking, disability, and quality-of-life outcomes in MS. The current research sets the stage for designing interventions that can result in long-term changes in physical activity and have potential effects on other outcomes in the MS population.

There are several limitations of the current study. We focused only on people with RRMS because this is the most common clinical course. Our results are not generalizable among those with clinically isolated syndrome or progressive disease courses. We further focused on only one social-cognitive variable, namely, self-efficacy. This approach was based on previous research involving symptoms and physical activity41 and practical issues of balancing the length of the survey battery with patient burden and adherence. Consequently, we do not have information regarding changes in outcome expectations, facilitators or impediments, and behavioral processes as correlates of changes in physical activity. We only measured symptoms twice annually with a single scale per symptom because of the number of measures and duration of the research study. The limited sampling of symptoms might not have captured important changes in symptoms that occurred over shorter time intervals when considering correlates of changes in physical activity. There is a lack of published information on the clinical meaningfulness of changes in scores on the scales included in this study. We were unable to characterize if the observed small changes in physical activity, self-efficacy, walking, and disability are clinically meaningful. There is limited published information on the sensitivity of all the measures, and perhaps the symptomatic outcomes were not sensitive for capturing actual changes in fatigue, depression, and pain over time.

Overall, the current research documented linear changes in physical activity, self-efficacy, walking impairment, and disability status over a 2.5-year period of time in people with RRMS. There were not significant linear changes in the symptom scores over time. We identified change in self-efficacy as a correlate of change in physical activity variables over time, even after controlling for walking impairment, disability status, and other confounding variables. Such results support the consideration of developing, delivering, and testing a behavior intervention based on social-cognitive theory for increasing physical activity over a long period of time in a large sample of people with MS. This endeavor will further our efforts in understanding the importance of physical activity in the lives of people with MS.42

Footnotes

  • Professors Motl and McAuley provided concept/idea/research design and fund procurement. All authors provided writing. Professor Motl and Mr Sandroff provided data collection. Professor Motl provided data analysis, study participants, and facilities/equipment. Mr Sandroff provided project management.

  • This investigation was supported by a grant from the National Multiple Sclerosis Society (RG 3926A2/1).

  • Received November 29, 2012.
  • Accepted April 8, 2013.

References

View Abstract