Using a prospective cohort study design, 72 participants aged ≥ 18 years and enrolled in the 6-week Australian hospital-based phase II cardiac rehabilitation program were recruited to the 12-month observational study between November 2015 and August 2016. The cardiac rehabilitation program is multidisciplinary, time-limited (12 sessions [2 per week for 6 weeks]), conducted in groups, and has educational and supervised exercise components (one hour education plus one hour exercise). Participants were included if they had stable CHD and were receiving optimal medical treatment ± revascularisation . All participants provided written consent. The study protocol, baseline, 6-week and 6 and 12-month results have been described elsewhere [19,20,21]. The investigation conforms with the principles outlined in the Declaration of Helsinki . Reporting was guided by the STROBE Statement (cohort studies) (Additional file 1).
A triaxial commercial accelerometer (ActiGraph ActiSleep, Fort Walton Beach, FL) was used to objectively assess sedentary behaviour (min/day, bouts, breaks) and physical activity (MVPA, light-intensity (LIPA); min/day). Participants were asked to wear the monitor on their right hip for 24 h/day for 7-consecutive days by cardiac rehabilitation staff. Sleep time was eliminated by using a time filter applied from 0700 to 2230 h, the average wake and sleep time reported by participants. All data was sampled and downloaded as raw data (30 Hz), converted to 1-s epochs (time interval), and then counts per minute (cpm) using the Actilife software. Data was screened, excluding data with < 10 h/day wear time (non-wear defined as > 60 consecutive minutes of zero activity, allowing for 2 min of counts between 0 and 100) and < 4 days of valid data. The Freedson Combination energy expenditure algorithm was used to determine time spent in LIPA (100–1951 cpm), MVPA (≥ 1952 cpm) and sedentary behaviour (< 100 cpm) . Estimating time spent in physical activity and sedentary behaviour was calculated by dividing the total time spent (minutes) in each threshold by the number of valid days. Sedentary behaviour bout data used a minimum length of 10 min, with no drop time . Sedentary bouts are the number of bouts (≥ 10 consecutive minutes) of sedentary time per day. Average sedentary bout length is the total time in sedentary bouts divided by the total number of bouts per day. A break is an interruption in sedentary time (≥ 100 cpm).
Outcomes included BMI (kg/m2); waist circumference; resting systolic blood pressure; fasting blood lipid (total cholesterol, HDL) and glucose levels; exercise capacity (6MWTD ); health-related quality of life (MacNew Heart Disease Health-related Quality of Life Questionnaire Global score (MacNew Global), with scores from 1, low health-related quality of life, to 7, high health-related quality of life ); and anxiety and depression (Hospital Anxiety and Depression Scale total score (HADS total), with maximal score of 42, high scores indicating high anxiety and depression ). The MacNew and HADS have good reliability and validity in adults with cardiovascular disease. Sociodemographic data was collected. All exposures and outcome measures (physical activity, sedentary behaviour, cardiovascular risk factors, health-related quality-of-life and exercise capacity) were assessed at baseline, 6-weeks, and 6 and 12-months.
Data were analyzed using SPSS version 27. Descriptive statistics were reported using means and SDs, medians with IQRs or proportions, where appropriate. Normality was assessed using the Shapiro–Wilk test. Accelerometer missing data was considered and differences in baseline characteristics (age, gender, education, employment, 6MWTD) between participants with missing data and participants without missing data were assessed using independent t-tests or Mann–Whitney U tests, where appropriate. Missing data was handled by bringing the last value forward.
Models using the Generalized Estimating Equations (GEE) approach were used to investigate the association between physical activity and sedentary behaviour and cardiovascular risk factors, health-related quality-of-life and exercise capacity over 12-months after starting cardiac rehabilitation. Physical activity and sedentary behaviour measures (LIPA min/day, MVPA min/day, sedentary behaviour min/day, sedentary behaviour bouts/day, sedentary behaviour breaks/day, average sedentary behaviour bout length min/day: independent variables) were modelled separately with cardiovascular risk factors (waist circumference, BMI, systolic blood pressure, blood lipid and glucose levels, anxiety and depression (HADS total)), health-related quality of life (MacNew Global) and 6MWTD as response outcomes (dependent variables) over the 4-time points. Interaction terms involving Time (moderator) and each independent variable were added in the models to assess if the effect predictor was significantly changing over time. The effects of independent variables were reported as regression coefficients with their associated 95% CIs. To account for multiple tests, a Bonferroni correction was applied (6 × 9 = 54) with p ≤ 0.001 used as the threshold for statistical significance. All models were adjusted for total accelerometer counts/day, age, gender, education and employment. Age, gender, education and employment are known factors to be associated with physical activity and sedentary behaviour levels . In addition, systolic blood pressure models were also adjusted for blood pressure medications, total cholesterol and HDL models were adjusted for cholesterol medications, and blood glucose level models were adjusted for type 2 diabetes. Total accelerometer counts/day is a proxy for total physical activity volume, encompassing frequency, intensity and duration of activity bouts [28, 29], although it is understood that it may be highly correlated with MVPA, LIPA and sedentary behaviour. Thus, sensitivity analyses were performed where GEE models were not adjusted for total accelerometer counts/day to determine if this changed findings. Additional sensitivity analyses were also performed exploring on-protocol and unadjusted models.