Blanchard, Chris M. PhD; Giacomantonio, Nicholas MD; Lyons, Renee PhD; Cyr, Cleo RN; Rhodes, Ryan E. PhD; Reid, Robert D. PhD; Spence, J. C. PhD; McGannon, Kerry R. PhD
It is recommended that patients with heart disease accumulate 30 minutes of moderate to vigorous physical activity (PA) on at least 5 days of the week.1 Unfortunately, less than 55% of patients are meeting these recommendations during and after cardiac rehabilitation (CR),2–5 suggesting that PA adherence is a problem. However, these studies used self-reported PA measures, whereas the use of objective measures (eg, pedometers) has been encouraged for patients with heart disease to obtain more valid and reliable PA assessments.6,7
In the context of CR, where resources tend to be scarce, an affordable way to obtain an objective PA assessment is to use pedometers.8 To date, 3 studies during CR have shown that patients average 8658 steps per day on CR days and 5364 steps per day on non-CR days,9–11 whereas patients averaged 6749 steps per day 6 months after CR.12 There are several concerns relative to these studies, however. First, although informative, 2 of these studies9,10 did not standardize the timing of recruitment, making it difficult to discern how active patients were at a given time point (eg, at the end of CR). Second, only 1 study12 assessed steps per day after CR only in patients with myocardial infarction. Whether similar steps per day after CR are present in a more heterogeneous sample remains unknown. Third, none of the studies measured PA during and after CR with the same sample. As a result, it occurs that the use of an objective PA assessment with a heterogeneous sample of CR patients might provide important insights regarding the stability of PA for these patients.
The first purpose of this study was to examine the steps-per-day growth trajectories of CR patients from the end of CR to 3, 6, and 9 months after completing CR, using latent class growth analysis (LCGA). This approach was chosen because traditional growth-modeling approaches (eg, repeated-measure analysis of variance) assume patients come from a single population and that a single growth trajectory can adequately approximate an entire population.13 However, a recent study14 in patients with heart disease not attending CR challenged this approach by identifying 2 distinct classes (subgroups) of patients using self-reported PA data. The first class (ie, ∼20% of the sample) averaged 143 minutes per day of PA at baseline, after which the PA levels significantly declined to 76 minutes per day 1 year after hospitalization. The second class (ie, ∼80% of the sample) averaged 11 minutes per day of PA at baseline, which significantly increased to 21 minutes per day at 1 year. Whether similar classes of patients exist in CR patients using an objective PA measure remains unknown.
The second purpose of the present study was to identify key predictors of class membership. This is particularly important to identify patients who may be at risk for inactivity (ie, if an inactive steps-per-day trajectory emerges).
Ten programs received ethical approval, although 2 did not complete participation and informed their respective ethics boards. The remaining 8 programs (ie, 5 in New Brunswick and 3 in Nova Scotia, Canada) combined medically supervised PA with nutrition and behavior modification education to reduce risk factors associated with heart disease. The programs ranged in duration (ie, 6, 10, or 12 weeks), frequency of PA sessions (ie, once versus twice per week), and location (ie, 2 community-based and 6 hospital-based).
Patients were eligible to participate if they were currently participating in CR, were ≥18 years of age, could read and write English, and provide informed consent. Patients were excluded if they were unable to ambulate independently or failed to complete at least 1 pedometer assessment.
Demographic characteristics included self-reported age, gender, education, marital status, race, employment, and income. Clinical characteristics (ie, whether or not it was the patient first cardiac event; height, weight, and body mass index; and the presence or absence of diabetes, stroke, lung disease, arthritis, cancer, circulation problems, back pain, hip pain, knee pain, and foot pain) were obtained via self-report and chart review and were summed to form a comorbidity score. Each site completed a standardized form to provide the diagnosis and baseline stress test scores in metabolic equivalents (METS). Finally, the CR program characteristics (ie, program length, program frequency, patient adherence to PA sessions, and distance to CR from the patient homes) were recorded via a standardized form that was calculated in ArgGIS 10.1 (ESRI Inc, Toronto, Ontario, Canada).
Physical activity was measured via a modified version of the Godin Leisure-Time Exercise Questionnaire.15,16 Patients were asked, (a) “How many days in a typical week out of the past 3 months did you do light (eg, walking) exercises for at least 10 minutes at a time? and (b) “On the days when you did light exercises, how much total time on average did you spend per day doing these light exercise(s)?” The same 2 questions assessed the frequency and duration of moderate (eg, brisk walking and easy cycling) and vigorous (eg, running and jogging) activities. The frequency and duration for each intensity were multiplied to obtain the total minutes of activity per week within each intensity. For the current study, we summed the moderate and vigorous minutes to obtain the total minutes per week of moderate to vigorous PA, which has been validated in patients with heart disease.14
Steps per day
The number of steps was assessed using the Yamax DIGI-WALKER (San Antonio, TX), which has been shown to be a valid and reliable pedometer17 (ie, accurately measuring the number of steps taken during a 400-m walk) and has been used in other trials in patients with heart disease.18 Before wearing, the pedometer was programmed for the patient stride length and weight to ensure accuracy of the step counts obtained. Patients were asked to wear the pedometer on their right hip for all waking hours of the day for 7 consecutive days and record their daily steps in a logbook at the end of each day. The total number of steps was added across the 7 days and divided by 7 to obtain a steps-per-day outcome variable.
On receiving ethical approval at each site, patient charts were reviewed for eligibility on program entry. Eligible patients were approached by a CR staff member to ask for their permission to have a research assistant (RA) approach them about the study. Patients who agreed were approached by the RA in the third week of their program. At that time, patients not interested were thanked for their time and were asked to complete a 1-page nonparticipant questionnaire. Those who agreed were given a consent form and baseline questionnaire that included the self-reported demographic and clinical questions to complete that day (ie, either on site or at home). Patients who took the documents home returned them at their next CR class. At the end of CR (ie, within the last 2 weeks of program completion), patients were asked to complete the same questionnaire in addition to wearing a pedometer for a 7-day period. Patients recorded their daily steps in a logbook and returned their study materials during the following week's class. Finally, patients were asked to complete the same questionnaire and wear the pedometer 3, 6, and 9 months after completing CR. For each assessment, patients were contacted by the RA to provide them with the study materials (ie, in person at the CR site or via the mail), after which an arrangement was made to obtain the completed materials at the CR site or receive them via the mail in a postage-paid envelope.
Recruitment rates were calculated followed by between-subject analyses of variance and χ2 analyses to determine potential differences between participants and nonparticipants, after which the baseline demographic, clinical, and program characteristics were generated. Next, a series of LCGAs19 was conducted in MPLUS 6.1 using full information maximum likelihood estimation to include all patients with at least 1 pedometer assessment. A single-class model was specified first with a latent intercept growth factor (ie, steps per day at the end of CR) and a latent slope growth factor (0, end of CR; 3, 3 months after CR; 6, 6 months after CR; and 9, 9 months after CR) to reflect the monthly change in steps per day. Next, a quadratic term was added to test for its necessity in the model. The same procedure was then used to examine 2-, 3-, and 4-class models. To identify the number of classes (objective #1), the Bayesian Information Criterion, the Bootstrapped Likelihood Ratio Test, and entropy indices were used. When comparing a 2-class versus single-class model, for example, a change in the Bayesian Information Criterion >10, higher entropy value (near 1.0), and a Bootstrapped Likelihood Ratio Test P < .05 would be considered as evidence favoring the 2-class over the single-class model.20 Once the final number of classes was generated, a series of χ2 analyses was conducted in SPSS 20.0 to determine potential demographic, clinical, and program differences across the classes. All significant variables were then entered into a series of logistic regressions to delineate the most important class membership predictors (objective #2).
There were 606 patients deemed eligible, and 235 patients agreed to participate and completed at least 1 pedometer assessment. The nonparticipants were similar in age (F1,432= 1.78, P = .18), gender (χ21= 0.46, P = .49), and self-reported minutes of PA (F1,378= 0.07, P = .79) compared with participants. Of the 235 patients who participated in the study, 149 completed their assessment 3 months after CR, 144 completed 6 months after CR, and 150 completed 9 months after CR. Detailed demographics for the 235 patients are presented in Table 1.
In terms of the LCGAs, results indicated that the 4-class model showed the best model fit (Table 2). However, the first class had only 12 patients, making subsequent analytical comparisons difficult. Therefore, the 3-class model was chosen. The first class (ie, nonadherers) comprised 27.2% of the sample and averaged 3113 steps per day at the end of CR, which remained stable up to 9 months after CR. The second class (ie, optimal adherers) comprised 27.4% of the sample and averaged 10 609 steps per day at the end of CR that remained stable up to 9 months after CR. The third class (ie, 45.4%) averaged 7011 steps per day at the end of CR; however, the patient steps per day significantly declined by 77 per month, and they were termed significant decliners.
Table 1 shows that the distributions for age, education, marital status, income, whether or not it was the patient's first cardiac event, obesity, whether or not the patient had ≥1 comorbidity, and whether or not the patient achieved good exercise capacity (ie, ≥9 METS on their stress test upon entry to CR) varied by class. Therefore, these variables were entered into 3 separate logistic regression models. As can be seen from Table 3, nonadherers were significantly more likely to be obese (OR = 0.23), have at least 1 comorbidity (OR = 0.23), and have an exercise capacity <9 METS (OR = 3.67) compared with optimal adherers. In terms of the nonadherers versus significant decliners comparison, significant decliners were more likely to have an exercise capacity ≥9 METS (OR = 2.36). Finally, optimal adherers were more likely to have experienced their first cardiac-related event than the significant decliners (OR = 0.29), whereas significant decliners were more likely to be obese than optimal adherers (OR = 2.11).
The single-class model showed that patients averaged ∼6911 steps per day during and after CR, which is consistent with previous research. However, an important question to pursue is whether this represents a “complete” picture of CR patient PA levels. On the basis of the LCGA results, the answer is no. The good news is that a class emerged (ie, optimal adherers) that averaged ∼10 609 steps per day during and up to 9 months after CR. This suggests that ∼27% of CR patients are experiencing health benefits resulting from their PA.9 The bad news, however, is that another 27.2% of the sample (ie, nonadherers) was averaging 3113 steps per day during and after CR. Furthermore, although the significant decliners (ie, 45.4% of the sample) averaged 7011 steps per day at the end of CR, they decreased their steps per day by 77 each month after CR, resulting in an average of 6317 steps per day of 9 months after CR. Therefore, the nonadherers and significant decliners (ie, 63% of the sample) need a PA intervention. Before doing so, however, it will be important to identify the key theoretical step-per-day correlates (eg, intentions and self-efficacy) for these target groups to better inform the development of a behavioral intervention designed to increase their steps per day.21
The present study also showed that nonadherers were more likely to be obese, have at least 1 comorbidity, and have a lower exercise capacity (ie, <9 METS) at the beginning of CR compared with the optimal adherers and significant decliners. Furthermore, significant decliners were more likely to be obese than optimal adherers. As such, these clinical characteristics provide potential target groups (eg, obese vs nonobese) that require behavioral interventions tailored to their specific needs to increase their steps per day.
Despite the strengths of the study (ie, repeated-measures design and use of pedometers), some limitations need consideration. First, despite having a heterogenous sample, it was self-selected and potentially biased. Future studies should attempt to recruit randomly if possible. Second, although pedometers provide an objective PA measure, they do not assess intensity or capture all PA modes (eg, swimming). Future studies should use accelerometers to examine PA trajectories by intensity during and after CR and incorporate logbooks to capture all modes of PA. Finally, we were not able to determine whether patients participated in a maintenance program after completing CR and future studies should try to account for this potential confounder.
Results showed that 3 distinct classes of patients emerged with different steps-per-day trajectories. It appears that being obese and having at least 1 comorbidity and a lower exercise capacity upon entry to a CR program may place patients at increased risk of being a nonadherer. Therefore, these patients may be an important high-risk group to target in PA interventions.
The project was funded by the Canadian Institutes of Health Research, the Heart and Stroke Foundation of Nova Scotia, and the Heart and Stroke Foundation of New Brunswick.
1. Smith S, Allen J, Blair S, et al. AHA/ACC guidelines for secondary prevention for patients with coronary and other atherosclerotic vascular disease: 2006 update. Circ Res. 2006;113:2363–2372.
2. Ayabe M, Brubaker P, Dobrosielski D, et al. The physical activity patterns of cardiac rehabilitation participants. J Cardiopulm Rehabil. 2004;24:80–86.
3. Blanchard C, Reid R, Morrin L, et al. Demographic and clinical determinants of moderate to vigorous physical activity during home-based cardiac rehabilitation: the home-based determinants of exercise (HOME) study. J Cardiopulm Rehabil Prev. 2010;30:240–245.
4. Bock BC, Albrecht AE, Traficante RM, et al. Predictors of exercise adherence following participation in a cardiac rehabilitation program. Int J Behav Med. 1997;4:60–75.
5. Moore SM, Ruland CM, Pashkow FJ, Blackburn GG. Women's patterns of exercise following cardiac rehabilitation. Nurs Res. 1998;47:318–324.
6. Le Grande M, Elliott P, Worcester M, Murphy B, Goble A. An evaluation of self-report physical activity instruments used in studies involving cardiac patients. J Cardiopulm Rehabil Prev. 2008;28:358–369.
7. Vanhees L, Lefevre J, Philippaerts R, et al. How to assess physical activity? How to assess physical fitness? Eur J Cardiovasc Prev Rehabil. 2005;12:102–114.
8. Tudor-Locke C, Washington T, Hart T. Expected values for steps/day in special populations. Prev Med. 2009;49:3–11.
9. Ayabe M, Brubaker P, Dobrosielski D, et al. Target step count for the secondary prevention of cardiovascular disease. Circulation J. 2008;72:299–303.
10. Jones N, Schneider P, Kaminsky L, Riggin K, Taylor A. An assessment of the total amount of physical activity of patients participating in a phase III cardiac rehabilitation program. J Cardiopulm Rehabil Prev. 2007;27:81–85.
11. Savage P, Ades P. Pedometer step counts predict cardiac risk factors at entry to cardiac rehabilitation. J Cardiopulm Rehabil Prev. 2008;28:370–377.
12. Izawa K, Yamada S, Oka K, et al. Long-term exercise maintenance, physical activity, and health-related quality of life after cardiac rehabilitation. Am J Phys Med Rehabil. 2004;83:884–892.
13. Nagin D, Tremblay R. Analyzing developmental trajectories of distinct but related behaviors: a group-based method. Psychol Methods. 2001;6:18–34.
14. Blanchard C, McSweeney J, Giacomantonio N, et al. Distinct trajectories of light and moderate to vigorous physical activity in heart disease patients: results from the Activity Correlates afTer cardIac hospitalizatiON (ACTION) trial. J Sci Med Sport. 2013;S1440–2440(13)00050–00059.
15. Godin G, Shepard RJ. A simple method to assess exercise behavior in the community. Can J Appl Sport Sci. 1985;10:141–146.
16. Jacobs DR Jr, Ainsworth BE, Hartman TJ, Leon AS. A simultaneous evaluation of 10 commonly used physical activity questionnaires. Med Sci Sports Exerc. 1993;25:81–91.
17. Schneider P, Crouter S, Lukajic O, Bassett D. Accuracy and reliability of 10 pedometers for measuring steps over a 400-m walk. Med Sci Sports Exerc. 2003;35:1779–1784.
18. Reid R, Morrin L, Beaton L, et al. Randomized trial of an Internet-based computer-tailored expert system for physical activity in patients with heart disease. Eur J Prev Cardiol. 2012;19:1357–1364.
19. Jones B, Nagin D, Roeder K. A SAS procedure based on mixture models for estimating developmental trajectories. Soc Methods Res. 2001;29:374–393.
20. Nylund K, Asparouhov T, Muthen B. Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct Eq Model. 2007;14:535–569.
21. Blanchard C. Heart disease and physical activity: looking beyond individual characteristics. Exerc Sport Sci Rev. 2012;40:30–36.
cardiac rehabilitation; latent class growth analysis; steps per day