Participant recruitment and study design
Study participants were recruited from a large pharmaceutical company located in Flanders, the northern Dutch-speaking part of Belgium. The company employs over 3800 employees working across five worksites. Participants (n = 300) were recruited from two of the five worksites in two time waves, representing the first half year of 2013 and 2014, respectively. All employees of the company were invited by e-mail to attend an information session in which the purpose, organization, duration, benefits and registration of the PA intervention program was clarified. Eligible employees who were interested could subscribe to enroll in the intervention by an online registration tool.
The online registration tool consisted of a company’s health risk assessment (HRA) tool in which employees indicated the number of days and the accumulated number of minutes (in bouts of 10 min) during which they participated in moderate- and/or vigorous-intensity PA (MVPA) . More specifically, they were asked on how many days, over the last thirty days, they engaged in activities of moderate and vigorous intensity (separate questions for each intensity). Subsequently, respondents were asked to quantify the number of minutes they accumulated on an average day. To calculate the equivalent combination of MVPA, minutes of vigorous activity were weighted by two to account for their greater intensity. Employees who were classified as being insufficiently active (i.e., not meeting the health-enhancing PA recommendations ) were invited to participate in the PA intervention program.
Throughout the period of baseline measurements, the PA intervention program was promoted by means of internal communication (distributed by the Environment, Health and Safety office) including e-mails to all employees, leaflets, posters, and screen-based announcements.
Participants were measured using a quasi-experimental design, in which intervention group participants (n = 246) were recruited separately from participants in the reference group (n = 54). Both groups were recruited from the same two worksites and during the same recruitment period. The reference group was recruited by means of a general call for participants who were interested in monitoring their activity pattern and assessing their physical health.
Participants read and signed an informed consent and all procedures were approved by the local Medical Ethics Committee of the KU Leuven, Belgium (s54788). At baseline, participants of the intervention group were screened on contraindications to engage in PA using the Physical Activity Readiness Questionnaire (PAR-Q) . Participants were advised to visit a physician if they answered ‘yes’ to one or more of the PAR-Q items. They were, however, allowed to participate in the PA counseling intervention.
The 3-month intervention consisted of nine contact moments between participants and PA counselors (n = three): six face-to-face contact moments and three contact moments by e-mail or telephone (Fig. 1). Two of the six face-to-face contacts were explicitly planned as face-to-face counseling sessions, while the other four face-to-face sessions were mainly considered as measurement sessions. In addition, the three e-mail or telephone contact moments included PA counseling content and were part of the PA counseling (Fig. 1). The three PA counselors held Master degrees in Kinesiology, were educated in health-related PA counseling and were specialized in SDT-based counseling.
The initial session included an introductory talk in which participants expressed their general expectations and goals regarding the intervention program. After this short introductory talk, participants were asked to complete a questionnaire (on demographics, PA level and psychosocial variables; see Additional file 1). Thereafter, anthropometric measures (e.g., blood pressure, BMI, fat percentage) were collected by the PA counselors.
The measurement sessions consisted of two sessions that were separated by 1 week (Fig. 1). In the first session of each measurement period (session one, three and five), participants filled out the questionnaire and PA counselors collected anthropometric measures. Based on the participants’ anthropometry, a PA monitor (SenseWear Armband (SWA)) was prepared and participants were instructed to wear the SWA during the following week. In the second session of each measurement period (session two, four and six), results on the SWA and physical health measures (anthropometry) were discussed and compared with general recommendations for the specific parameters. In case of post- (session four) and follow-up (session six) measurement sessions, PA and physical health measures were compared to participants’ results on pre- and post-test, respectively.
In line with previously conducted PA interventions, the face-to-face counseling sessions lasted up to 60 min [6, 26]. During the first face-to-face counseling session (session 2), an individually tailored PA plan was designed based on participants’ goal, preferences for type of activities and current level of PA. Individualized PA programs were offered to facilitate common lifestyle PAs to improve cardiovascular health such as walking, cycling, running and swimming. Participants were allowed to choose one or more of the offered PA programs.
In addition, participants were encouraged to select at least one action out of a predefined list of actions to reduce their sedentary time and increase PA in multiple contexts (including home, transportation, workplace and leisure time). Proposed actions to reduce sedentary time included statements such as ‘I will stand during phone calls’ (work context) and ‘I will try to limit television viewing time’ (home context). The suggested actions to increase PA included statements such as ‘I will get off the bus one stop earlier’ (transportation context) and ‘I will take the stairs instead of the elevator’ (work/leisure time context). This opportunity to choose was created to support feelings of autonomy by enabling participants to individually select actions that they considered as relevant for themselves .
The three counseling contact moments between pre- and post-test were standardized and completed by e-mail or telephone depending on participant’s preference. The majority of participants (77%) chose to be contacted by e-mail, while nearly a quarter of participants (23%) preferred to be contacted by telephone. The content of the e-mail or telephone conversations was guided by the individualized PA plan and based on information on participant’s (baseline) PA behavior. More specifically, PA goals were evaluated and adjusted if necessary and participants were motivated to persist in and sustain their PA.
PA counseling was explicitly focused on fostering the three basic psychological needs outlined by SDT, (i.e., the need for autonomy, competence and relatedness) .
PA counselors intended to support the need for autonomy by allowing participants to choose from a number of options to facilitate lifestyle PAs (i.e., different types of individualized PA programs: walking, cycling, running and swimming programs). In addition, the need for autonomy was supported by providing participants with informational feedback (e.g., provide participants with individualized and personally relevant information regarding their PA preference) and by minimizing pressure (e.g., focus on facilitating rather than prescribing PA) during counseling sessions. These techniques encouraged participants to make informed decisions about the direction in which they preferred to proceed for the remainder of the intervention.
In order to support participants’ need for competence, PA counselors encouraged participants to consider how their intention(s) to become physically active might be implemented (i.e., implementation planning specified by the guiding questions ‘What, where, when, with whom will I be physically active?’; ‘What do I want to achieve?’; ‘How do I remind myself to be active?’). Implementing intentions based on these questions has been shown to be effective in previous studies [27, 38].
PA counselors intended to satisfy the need for relatedness by being empathetic (e.g., by demonstrating understanding), by providing positive feedback and by active listening during the counseling sessions. Moreover, participants were encouraged to engage in PA together with colleagues (e.g., form a group with colleagues to walk during lunch breaks) and seek support from colleagues. These techniques collectively aimed to stimulate participants to feel connected to both the PA counselor and colleagues alike.
Besides the need-supportive counseling, our intervention consisted of two additional behavior change techniques, namely barrier identification and self-monitoring . Participants were asked to identify barriers (e.g., environmental or social) that could potentially deter them to engage in regular PA. They were asked to formulate ways to overcome the self-formulated potential barriers. With respect to self-monitoring, participants received pedometers (Omron, Walking Style One 2.1) and PA diaries offering participants an opportunity to goal set and self-monitor their PA behavior [40, 41].
The main goal of the intervention program was to increase participants’ baseline PA level.
Ideally, participants would adopt more PA into their daily lives and more participants would attain the recommended PA norm of 30 min of daily MVPA .
In the second face-to-face counseling session (session four), 14 weeks after the start of the intervention, participants’ PA behavior change was evaluated and future challenges for PA were discussed in order to encourage participants’ maintenance in PA engagement.
Between post- and follow-up sessions, no contact occurred between intervention group participants and PA counselors.
During the final session at follow-up (session six), participants were provided with an information leaflet. This information leaflet centered on future maintenance of PA and included tips to remain physically active after the intervention has ended.
Participants in the reference group completed single pre-, post- and follow-up measurement sessions without receiving individualized PA counseling or a tailored PA plan. In the follow-up session (session three), the results with respect to their actual PA level were discussed and they were informed on the general PA recommendations.
Outcome measures were assessed at baseline, at post-intervention (immediately after the intervention) and at follow-up (6 months post-intervention; 9 months after baseline) (Fig. 1).
PA and sedentary behavior
PA behavior was objectively assessed by activity monitoring and assessed by self-report.
Objectively assessed PA and sedentary time
PA behavior was monitored by the SenseWear Pro3 Armband (BodyMedia, Inc. Pittsburgh, PA, USA) worn over the triceps muscle of the right arm. The SWA measures PA behavior using multiple sensors (i.e., tri-axial accelerometer, heat flux, skin temperature and galvanic skin response sensor) and was found to provide accurate, reliable and valid measures of PA [42–44]. Participants were instructed to wear the SWA for seven consecutive days (five weekdays and two weekend days) in order to provide reliable measures . SWA data were combined with participants’ gender, age, height and body weight and were analyzed using computer-based SWA software (SenseWear professional software, version 7.0). Valid data included SWA data monitored on at least three weekdays and one weekend day for at least 720 min per day. PA intensity and daily step count were calculated. PA intensity was determined using MET values. Activities with a MET-value ≤1.8 were considered as sedentary . Activities of light, moderate and vigorous intensity were defined as activities with a MET-value >1.8 and <3, ≥3 and <6, and ≥6, respectively . In addition, time spent in combined MVPA was calculated by combining moderate PA with vigorous PA, respectively (moderate PA: vigorous PA = factor 1:2). Minute by minute activities were averaged into daily totals (min/day). To allow comparisons with the International Physical Activity Questionnaire (IPAQ) measures, modified ten-min bouts of accumulated activities were introduced. Bouts were defined as ten or more consecutive minutes of activity, allowing for interruptions of 1 or 2 min below the MET threshold [47, 48]. Activities were averaged on weekdays and weekend days separately. Weekdays and weekend days were also combined into an average day using the following formula: average day = (weekday average*5) + (weekend day average*2)/7.
Besides the daily minutes of the different intensities of PA, we also integrated the daily minutes of MVPA and number of daily steps into variables that represent the health-enhancing MVPA and steps guidelines. More specifically, we determined the number of participants meeting the recommended number of 10,000 steps per day (for adults) . Furthermore, the number of participants achieving the recommended amount of 30 min of MVPA was determined . The guideline measures were determined for weekdays, weekend days and average days separately.
Self-reported PA and sitting time
The 7-item short version of the IPAQ was used to measure self-reported PA. Participants had to report on how many days per week and for how long they engaged in walking, moderate- and vigorous-intensity PA (in bouts of 10 min) in the past seven days. In addition, participants reported the average minutes they spent sitting in the preceding seven days . The IPAQ short form was found to have acceptable test-retest reliability . The minutes of moderate and vigorous PA were combined (moderate PA: vigorous PA = factor 1:2) to form an MVPA measure. This MVPA measure was translated to the number of participants meeting the recommended amount of 30 min of daily MVPA .
Psychological need satisfaction
Participants’ perceived satisfaction of basic psychological needs was assessed by the previously validated 12-item Basic Psychological Needs in Exercise Scale (BPNES) [51, 52]. Participants indicated, on a five-point Likert scale (ranging from ‘I don’t agree at all’ to ‘I completely agree’), the extent to which they agree with the statements on autonomy (four items; e.g., I feel that I have the opportunity to make choices with regard to the way I exercise), competence (four items; e.g., I feel I have made a lot of progress in relation to the goal I want to achieve) and relatedness (three items; e.g., My relationships with the people I exercise with are close). The autonomy, competence and relatedness subscales were found to be internally reliable at pre, post and follow-up (Cronbach’s alpha ranged from .73 to .93).
Degree of autonomy support
To evaluate the need-supportive character of the intervention, we also examined participants’ perceptions of autonomy support provided by the PA counselors. At post-intervention, participants of the intervention group were questioned on the degree of perceived autonomy support using the short (six-item) version of the Health Care Climate Questionnaire (HCCQ) (e.g., exemplary item: I feel that my physical activity counselor provides me with choices and options) . Intervention group participants indicated their agreement with the items on a seven-point Likert scale, ranging from totally disagree (=1) to totally agree (=7). The HCCQ items were marked by a high internal consistency (Cronbach’s alpha = .93).
Participants’ baseline level of motivation towards PA was compared using the Behavioural Regulation in Exercise Questionnaire-2 (BREQ-2) . Participants indicated on a five-point Likert scale (ranging from ‘not true for me’ to ‘very true for me’) to what extent they had extrinsic (e.g., ‘I exercise because other people say I should’), introjected (e.g., ‘I feel guilty when I don’t exercise’), identified (e.g., ‘I value the benefits of exercise’) and intrinsic (e.g., ‘I exercise because it is fun’) reasons for participation in PA. A fifth subscale included amotivation (e.g., ‘I don’t see why I should have to exercise’). Following the example of previous studies, we combined subscales into the composite variables autonomous motivation (intrinsic motivation and identified regulation) and controlled motivation (introjected regulation and external regulation) [55–57]. Cronbach’s alpha ranged from .62 to .90 indicating an acceptable internal consistency.
Data were analyzed using IBM SPSS (version 20; SPSS Inc., Chicago, IL, USA) and reported as mean and standard deviation in case of descriptives and as mean and standard error in case of outcome measures. Statistical significance was set at p < .05. Baseline differences between the intervention and reference group were examined by performing independent samples t-tests (for continuous measures) and chi-square tests (for categorical measures). Individual (at random) missing data on the BNPES questionnaire were imputed using the Expectation-Maximization procedure .
Outliers, defined as values exceeding three standard deviations from the mean, were identified in each outcome measure and excluded from analysis. Given the main advantage of mixed model analysis in handling with missing values, our longitudinal design allowed us to reliably assess the effects of the intervention with mixed model analysis . Linear mixed model analysis with an unstructured covariance structure was used to determine time and intervention effects of the PA counseling intervention on objectively assessed PA, self-reported PA and psychosocial variables. To distinguish between short- and long-term intervention effects, two separate mixed model analyses were performed for each outcome variable. Two time points (pre and post) were used to determine short-term intervention effects and three time points (pre, post and follow-up) were used to examine long-term intervention effects.
For categorical variables (% of participants meeting PA guidelines), generalized estimating equations with an unstructured covariance structure were conducted to examine the short-term and long-term intervention effects. In case of between-group baseline differences, the baseline value was included as covariate in the mixed model analyses.
Cohen’s d effect sizes (ESs) were computed based on F or χ
2 statistics and sample sizes of the intervention and reference group . ESs of < .30, .30-.80 and > .80 were considered small, medium and large effects, respectively .
Indirect effects of the PA counseling intervention (independent variable) on PA behavior (dependent variable) through perceived need satisfaction (mediators) were tested by the bootstrapping procedure using the SPSS PROCESS macro [62, 63]. Bias-corrected and accelerated confidence intervals (95% CI) of the indirect effects were generated with two thousand resamples. Bootstrapped CIs are preferred, as they make no unrealistic assumptions on the shape of the sampling distribution of indirect effects. Mediation analyses were conducted for change scores from pre- to post-test (short-term) and from pre- to follow-up-tests (long-term). Mediation was only tested in PA outcome measures which produced both short- and long-term intervention effects.