Study design
This study is part of a six-month cluster randomized control trial targeting physical activity and sedentary behaviour among office workers to promote healthy brain functions. The intervention consisted of three groups: the physical activity intervention group (iPA), the sedentary behaviour intervention group (iSED) and the control group (C). The intervention was carried out in 2018 and 2019. The detailed study protocol has earlier been published (18). The trial was prospectively registered as ISRCTN92968402 on 27/02/2018, recruitment started 15/03/2018. Ethical approval was granted by The Stockholm Regional Ethical Review Board (2017/2409-31/1). All participants provided written informed consent before first data collection.
Participants and data collection
Office workers from two private Swedish companies with around 2000 employees were invited. Inclusion criteria were to be age between 18-70 years and to have the capability of standing and exercising. Individuals were also excluded from the study at baseline if they showed a very high physical activity level assessed by accelerometer: more than 30 minutes of moderate to vigorous physical activity per day in prolonged bouts (≥10 min). For further information, please see Nooijen et al (18).
The participants were allocated to a cluster group within their company, grouped by human resources personnel at each company. The aim was that a cluster group should share team or line manager, have regular group meetings and limited regular meetings with other groups. The 22 clusters were then randomized into the three different intervention groups: iPA, iSED or control group. The research assistants working with the data collection were blinded for group allocation.
Intervention and motivational counselling
The two interventions were based on ecological framework targeting multiple intervention levels, including individual, organizational and environmental (20,21). The intervention aimed to target both work- and leisure time. The motivational counselling consisted of five counselling sessions based on Cognitive behavioural therapy (CBT) and Motivational interviewing (MI), three individual- and two group sessions. The counselling sessions were performed by health coaches from a health promotion company, who received additional training on CBT techniques and on physical activity and sedentary behaviour by CBT educated psychologists and physical activity experts. The CBT techniques included exploring and setting valued goals, functional analysis to explore what maintains the behaviour and acceptance techniques, all targeting the individual’s behavior change of physical activity pattern (Hofmann, et al, 2012 (22). The participants also registered their physical activity pattern and received feedback on their activity behaviour (6). One team leader per cluster were recruited with the aim to encourage the employees to participate, remain in the study and communicate with the research team. The intervention and the motivational counselling is described more fully in the protocol article (18).
Self-efficacy
Exercise self-efficacy was assessed with a ten item question using the Swedish version of the Exercise Self-efficacy Scale (7). The scale contains questions regarding the confidence the person feels for performing exercise and physical activity. The response alternatives consist of a four-point scale: 1=not at all confident; 2=somewhat confident; 3=confident; 4=totally confident. The possible scores ranged from 10 to 40, with a higher value indicating higher self-efficacy. The scale has shown high reliability and internal consistency among adults with neurological disease (7). However, the reliability and validity has not been tested in a healthy working population. Sedentary Self-Efficacy Scale was an additional scale developed by the authors based on the Exercise Self-efficacy Scale. The reliability and validity of this scale has not been examined, but the scale follows the same structure as the Exercise Self-Efficacy Scale. Median values were imputed for missing values for participants with one or two values missing in their Self-Efficacy scores (n=22 in total). The missing values were evenly distributed between individuals and items.
Motivation to physical activity
The Behavioural Regulation in Exercise Questionnaire (10), BREQ-4 Short version, was used to measure motivation to physical activity. The original response scores range from one to seven, but we used a five-point Likert scale, ranging from 1=”I don’t agree at all” to 5= “I completely agree”. The scale measure behavioural regulations with 7 factors; amotivation, external regulation, introjected avoidance, introjected approach, identified regulation, integrated regulation and intrinsic regulation. A mean score for autonomous motivation was created by combining intrinsic, integrated and identified regulation and for controlled motivation external regulation, introjected avoidance and introjected approach was combined (10).
Perceived barriers to reduce sedentary behaviour
An earlier publication has descriptively examined perceived barriers to decrease sedentary behaviour among Swedish office workers (5). The study examined 13 different reasons why individuals were sedentary during working hours. Three barriers were found particularly common, “sitting is a habit” (67 %), “standing is uncomfortable” (29 %) and “standing is tiring” (24 %). These three barriers were used in present study. The response alternatives are binary, “yes” or “no”.
Covariates
Included covariates were age, education (number of years), gender, and company (2 options).
Statistical analysis
Mean values and standard deviations are presented for both baseline and 6-month follow-up stratified on group allocation for exercise self-efficacy, sedentary self-efficacy, autonomous motivation, controlled motivation and amotivation, together with percentage of individuals responding yes on each of the three perceived barriers. Independent t-tests and chi-square statistics were used to test for differences in baseline characteristics between drop-outs within the three groups.
Comparison between groups were made with multilevel models using Bayesian statistics, similar to what was presented for the main results of the randomized control trial (23,24). In brief, Bayesian regression was used to calculate estimates and 95 % credible intervals. The credible intervals can be interpreted as probabilities. Other benefits with Bayesian statistics is the calculation of exact parameters estimates without reliance on large sample size, and that it can be applied to a large range of models, like multilevel models that are used in present study (23,24).
R statistical program language (3.6.1) were used in Rstudio (1.2.5) for the multilevel analyses using the brms (2.11.1), tidyverse (1.3) and tidybayes (1.1) packages. SPSS version 25 were used for descriptive statistics and the drop-out analyses.
Models
Clustering of data was in all models taken into account using a two-level structure with clusters as the first and participants as the second level. All groups were simultaneously regressed on the post-test value of the outcome and adjusted for the outcome baseline value. Covariates were age, gender, education, and company. We used uninformative prior (student t) for the coefficients. The Gaussian function was used (4 chains, 4 cores and 3000 iterations, 1000 warm-up) for the outcomes self-efficacy and motivation. Parameter estimates and 95% credible intervals were presented. The Bernoulli function was used (4 chains, 4 cores and 3000 iterations, 1000 warm-up) for the analyses on perceived barriers. Parameter estimates and credible intervals were exponentiated to provide posterior odds ratios. For all models, the achieved level of convergence was Rhat values of 1.
For participants with missing or impossible values, education was imputed with the median value. No further imputations were performed. There were no missing values for gender or age.
For each of the different defined outcome variables, separate models were run and repeated on two datasets: A. Complete Cases (CC): All participants with data at baseline and 6-months. B. Per Protocol (PP): Individual was in iPA or iSED group and had attended 3 to 5 counselling sessions, or individual was in C group. For those individuals who did not have information on attendance available, it was assumed that they did not participate in at least 3 sessions.
Note: Numbers is for complete cases and per protocol analysis are based on if the participants have data on self-efficacy and motivation.