How Do Stages of Change for Physical Activity Relate to Employee Sign-Up for and Completion of a Worksite Physical Activity Competition?

Abstract

Introduction. This study evaluated whether stages of change for physical activity (PA) predict sign-up, participation, and completion in a PA competition. Method. Deidentified data were provided to evaluate a PA competition between 16 different institutions from a public university system. Employees who completed a health assessment (HA) prior to the start of the PA competition (n = 6,333) were included in the study. Participants completed a self-report HA and logged their PA throughout the competition. Multivariable logistic regression models tested whether stages of change predicted PA competition sign-up and completion. An ordinal logistic regression model tested whether stages of change predicted number of weeks of PA competition participation. Results. Stages of change predicted PA competition sign-up and completion, but not weeks of participation. The odds for PA competition sign-up were 1.64 and 1.98 times higher for employees in preparation and action/maintenance (respectively) compared with employees in precontemplation/contemplation. The odds for PA competition completion were 4.17 times higher for employees in action/maintenance compared with employees in precontemplation/contemplation/preparation. Conclusion. The PA competition was more likely to reach employees in preparation, action, or maintenance stages than precontemplation/contemplation. Most of the completers were likely participating in regular PA prior to the competition.

Keywords

physical activity/exercise worksite safety and health behavior change

Introduction

Approximately 50% of Americans live with a chronic disease (Ward, Schiller, & Goodman, 2014), the impact of which comprises 75% of health care system costs (Harris & Wallace, 2012). Physical activity (PA) is known to reduce risk for various chronic diseases including diabetes, cardiovascular disease, and cancer (U.S. Department of Health and Human Services, 2008). Therefore, there is a need to implement programs promoting healthy behaviors such as PA. Worksites provide an ideal setting for PA programs because (1) a large segment of the population can be reached and (2) it is a community of people who share a common purpose and culture (Carnethon et al., 2009; Goetzel, Roemer, Liss-Levinson, & Samoly, 2008).

Despite the potential positive impact of PA programs, results from a recent review suggest worksite PA interventions do not consistently engage participants (Ryde, Gilson, Burton, & Brown, 2013). Past studies consider engagement to be a continuum represented by program recruitment, participation, and completion (Grossmeier, 2013; Terry, Grossmeier, Mangen, & Gingerich, 2013). Poor engagement greatly diminishes worksite program impact, generating the need for more research in this area.

The RE-AIM framework identifies five elements for ensuring program impact: reach, efficacy, adoption, implementation, and maintenance (Glasgow, Vogt, & Boles, 1999). Reach, the most relevant for this study, refers to the percentage of eligible participants who take part in the program and how representative they are of the entire population. A highly efficacious program can have low impact if its reach is limited to a small segment of the priority population or does not reach those most at risk. Therefore, there is a need to better understand factors influencing employee sign-up, participation, and completion of worksite PA programs.

Past studies have evaluated correlates and determinants of employee engagement in worksite health programs (Grossmeier, 2013; Robroek, van Lenthe, van Empelen, & Burdorf, 2009; Terry, Fowles, & Harvey, 2010). Results suggest female workers have greater participation percentages compared with male workers. However, results were less consistent when evaluating other demographic, health-related, and work-related variables. Furthermore, few studies have examined the potential impact of behavioral sciences constructs across levels of engagement in worksite health programs (Ablah et al., 2015; Prochaska, Norcross, Fowler, Follick, & Abrams, 1992; Robroek, Brouwer, Lindeboom, Oenema, & Burdorf, 2010; Robroek, Lindeboom, & Burdorf, 2012). Past results suggest intentions and stages of change may be associated with worksite program participation (Ablah et al., 2015; Prochaska et al., 1992; Robroek et al., 2012).

Stages of change is a key construct in the transtheoretical model (Prochaska & Norcross, 2001). The transtheoretical model states behavior change is a process occurring through five stages: precontemplation (no intention to become physically active), contemplation (thinking about starting to become physically active), preparation (intending to become physically active in the next month), action (meeting a PA criterion but for less than 6 months), and maintenance (meeting a PA criterion for more than 6 months) (Prochaska & Norcross, 2001). Evidence suggests stages of change is a significant predictor of PA level among adults (Marshall & Biddle, 2001; Trost, Owen, Bauman, Sallis, & Brown, 2002; van Stralen, De Vries, Mudde, Bolman, & Lechner, 2009). In addition, stages of change were found to predict participation in a worksite pedometer-based intervention (Ablah et al., 2015). However, no studies have tested whether the stages of change model predicts different levels of engagement, from sign-up to completion, in a worksite PA program.

Team-based PA competition is one type of worksite PA program. This study evaluated whether the stages of change model predicts sign-up, participation, and completion in a PA competition. Based on the definitions of the stages, we hypothesize the following:

Employees in preparation and action/maintenance for PA will have higher odds of sign-up for the competition compared with those in precontemplation/contemplation.

Employees in preparation and action/maintenance for PA will participate for a greater number of weeks throughout the competition compared with those in precontemplation/contemplation.

Employees in action/maintenance will have higher odds of completing the competition compared with those in precontemplation/contemplation and preparation.

Method

Design

This retrospective cohort study used samples drawn from secondary deidentified data provided by the University of Texas System’s Office of Employee benefits. Data were from (1) a 2013 health assessment (HA) administered by a System contractor to all 16 member institutions and (2) a PA log each participant completed during the 6-week 2013 PA Challenge. This study was approved by the University of Texas Health Science Center at Houston Ethics Review Board.

Participants

The System oversees employee benefits and supports wellness programs across 16 member institutions. Member institutions consist of nine academic and six health institutions, and the System’s own administrative institution. There were about 74,000 active employees across the 16 institutions at the time of study. In February 2013, the System offered HAs to all benefits eligible employees ≥18 years of age. Employees were contacted through e-mail and directed to a website to complete the HA. The sample included employees who completed an HA in 2013 prior to the start of the 2013 PA challenge (May 1-June 12).

Measures

Three dependent variables were

PA challenge sign-up, a dichotomous (yes/no) variable defined by whether an employee signed up for the PA challenge.

Participation level, the number of weeks a participant logged his or her PA using the online platform (ranged from 0-6).

Challenge completion, a dichotomous (yes/no) variable defined by whether a participant logged ≥150 minutes of PA on the online platform for all 6 weeks of the challenge.

For participation level, employees did not need to meet the weekly goal of 150 minutes of activity to get credit for a week of participation.

The independent variable for all three hypotheses was stages of change for PA. Employees indicated their readiness to make changes or improvements in their health by being physically active on most days. Participants were categorized into their respective stages: “haven’t thought about changing” (precontemplation), “plan a change in the next 6 months” (contemplation), “plan to change this month” (preparation), “recently started doing this” (action), and “do this regularly (last 6 months)” (maintenance). To test Hypotheses 1 and 2, the stage of change variable was grouped into three categories: (1) precontemplation/contemplation, (2) preparation, and (3) action/maintenance. Even though each stage has distinct characteristics, our hypotheses predicted those in precontemplation/contemplation and action/maintenance would behave similarly in terms of their sign-up and participation. This prediction was supported by preliminary analyses and consistent with previous research (Taylor et al., 2004). Since Hypothesis 3 was based on challenge completion, the stages of change variable was grouped into two categories: (1) precontemplation/contemplation/preparation and (2) action/maintenance. This grouping was based on our hypothesis predicting the respective groupings would behave similarly with regard to meeting the challenge goals, support from preliminary analyses, consistent with previous research (Ablah et al., 2015), and due to small cell sizes.

Variables included to control for confounding were the following: gender, age, institution size, body mass index (BMI), number of chronic conditions, and general health status. Institution size was determined by number of employees: small (<999), medium (1,000-2,999), large (3,000-5,999), and extra-large (>6,000). BMI was calculated using self-reported height and weight values. BMI values ≥30 kg/m² were considered obese, 25.0 to 29.9 kg/m² were considered overweight, and <25 kg/m² were considered healthy weight. Participants also indicated the presence or absence of several chronic conditions, with answers used to create the following categories: none, one, two, or three or more chronic conditions. General health was assessed using an HA question asking participants to rate their overall health with answers being collapsed into two categories: (1) excellent/very good/good and (2) fair/poor.

Program Description

The PA challenge was a 6-week competition between institutions. Each institution competed as a team with participants representing their respective institution. The challenge goal was for each participant to complete 150 minutes of PA each week, for 6 weeks. Participants logged activities using an online platform where they entered physical activities ranging from a gym workout to household chores. Real time standings were provided on the online platform. The winning institution was determined by the sign-up and completion percentages. The winning institution received the PA challenge trophy. Participants who logged the most active minutes received individual recognition from the System.

The System provided information to institutions outlining program objectives, a timeline, promotion suggestions, communication ideas, delivery tips, and additional resources. The timeline provided details for when to distribute recruitment materials, when to send the registration e-mail, and what to do throughout the challenge. The System suggested promoting the program through newsletters, staff meetings, Intranet/Internet sites, video monitors, bulletin boards, posters, table tents, and e-mails. Communication ideas included predrafted e-mails to be sent to employees highlighting challenge aspects. Program delivery tips included strategies such as a “kick-off” event and a communication plan. Institutions were provided with preprepared flyers, posters, table tents, web-banners, and e-mails. The System also encouraged wellness coordinators to develop and distribute institution specific materials.

Statistical Analyses

Descriptive statistics were calculated for all variables across stage of change category. Multivariable logistic regression models tested Hypothesis 1 (sign-up) and Hypothesis 3 (completion). An ordinal logistic regression model tested Hypothesis 2 (participation). A Wald test was used to confirm no violations of the proportional odds assumption for the ordinal regression model. Gender, age, and institution size were included in all models as control variables. BMI, number of chronic conditions, and general health status were included in models if p values were <.05. Cell sizes for the stages of change variable were evaluated across each respective outcome variable to confirm adequate numbers (≥5 cases). Odds ratios and 95% confidence intervals were calculated for each stage of change grouping across program outcome variables. Likelihood ratio tests were used to compare each final model to the corresponding constant only model. Model fit for logistic regression models were tested using Hosmer–Lemeshow’s goodness-of-fit test. McKelvey and Zavoina’s R² was used to assess overall fit for the ordinal regression model. Analyses were performed using Stata 13 with p < .05 considered to be the statistical significance level.

Results

A total of 12,504 employees completed an HA during the 2013 plan year. Of the 12,504 HA respondents, 51% (6,333 individuals) completed an HA prior to the PA challenge start, making up the total study sample. All respondents who took the HA prior to the challenge had complete data across study variables. A total of 1,522 employees signed up for the PA challenge with 65% (985 individuals) completing an HA prior to the challenge. Descriptive statistics for employees in each stage of change grouping for the total study sample are presented in Table 1. About 16% of the sample was in precontemplation/contemplation, 28% in preparation, and 56% in action/maintenance. The majority of the sample was female. Trends for age category and institution size were consistent across the stages of change groupings. Most were from an extra-large institution, and there were more employees in the 35- to 49-year age category than other age categories. Those in preparation appeared to be in worse health with a greater proportion of employees with obesity, three or more chronic conditions, or reporting a health status of fair or poor compared with those in other stages.

Table 1

Descriptive Variables by Stages of Change Groupings

Variable	Stage of Change			Total Sample (n = 6,333)
Variable	PC/C (n = 1,010)	P (n = 1,750)	A/M (N = 3,573)	Total Sample (n = 6,333)
Gender
Male	28.9	19.1	26.3	24.7
Female	71.1	80.9	73.7	75.3
Age (years)
18-34	22.8	27.6	29.2	27.7
35-49	38.3	38.6	33.8	35.9
50-64	31.7	29.1	29.6	29.8
65+	7.2	4.6	7.4	6.6
Institution size
Small	4.8	5.8	5.1	5.3
Medium	9.8	11.9	10.9	11.0
Large	12.5	16.0	15.5	15.2
Extra-large	73.0	66.3	68.4	68.6
BMI category
Healthy weight	41.8	28.3	43.5	39.0
Overweight	28.9	28.7	30.9	29.9
Obese	29.3	43.0	25.6	31.1
Number of chronic conditions
None	24.2	17.6	28.6	24.9
One	23.1	25.1	28.9	26.9
Two	21.3	20.9	19.5	20.2
Three+	31.5	36.3	23.0	28.0
General health status
Fair/poor	13.3	17.5	6.7	10.7
Good/very good/excellent	86.7	82.5	93.3	89.3

NOTE: PC = precontemplation; C = contemplation; P = preparation; A = action; M = maintenance; BMI = body mass index.

Challenge Sign-Up

Odds ratios, confidence intervals, and p values from the model testing stages of change as a predictor of challenge sign-up are shown in Table 2. BMI category, number of chronic conditions, and general health status were removed from the model during covariate selection. Results from a likelihood ratio test comparing the final model to the constant only model were statistically significant, χ²(9) = 247.33, p < .001. Results indicated the odds for employees in preparation to sign-up for the challenge were 1.64 times higher compared with employees in precontemplation/contemplation while holding covariates constant. Additionally, the odds for employees in action/maintenance to sign-up for the challenge were 1.98 times higher compared with employees in precontemplation/contemplation while holding covariates constant. The Hosmer–Lemeshow goodness-of-fit test was not significant suggesting the model fit the data well.

Table 2

Physical Activity Challenge Sign-Up by Stage of Change and Covariates (N = 6,333)

Stage of Change/Covariate	OR [95% CI]	p
Stage of change
Precontemplation/contemplation (referent)	1	—
Preparation	1.64 [1.27, 2.10]	<.001
Action/maintenance	1.98 [1.57, 2.49]	<.001
Gender
Male (referent)	1	—
Female	1.35 [1.14, 1.60	.001
Age category (years)
18-34 (referent)	1	—
35-49	1.12 [0.94, 1.32]	.199
50-64	0.88 [0.74, 1.06]	.189
65+	0.23 [0.14, 0.37]	<.001
Institution size
Small (referent)	1	—
Medium	0.77 [0.57, 1.04]	.089
Large	0.46 [0.35, 0.62]	<.001
Extra-large	0.30 [0.24, 0.40]	<.001

NOTE: OR = odds ratio, CI = confidence interval. Hosmer–Lemeshow statistic model fit results: χ²(8) = 7.20, p = .515.

Challenge Participation

Ordinal logistic regression tested stages of change as a predictor for challenge participation for employees who signed up for the PA challenge and completed an HA prior to the challenge. Stages of change was the independent variable and gender, age category, institution size, BMI category, and general health status were covariates. The number of chronic conditions variable was removed from the model during the covariate selection process. Results from the Wald test revealed no variables were in violation of the proportional odds assumption. Results from a likelihood ratio test comparing the final model to the constant only model was statistically significant, χ²(12) = 123.63, p < .001. The McKelvey and Zavoina’s R² value was 0.11. Table 3 summarizes the odds ratios, confidence intervals, and p values for the final model. Results suggest the model explained about 11% of the variance. Furthermore, odds of participation for employees in action/maintenance were higher compared with employees in precontemplation/contemplation. However, this odds ratio was outside the level of statistical significance (p = .057).

Table 3

Physical Activity Challenge Participation by Stage of Change and Covariates (n = 985)

Stage of Change/Covariate	OR [95% CI]	p
Stage of change
Precontemplation/contemplation (referent)	1	—
Preparation	0.74 [0.48, 1.12]	.152
Action/maintenance	1.45 [0.99, 2.12]	.057
Gender
Male (referent)	1	—
Female	0.80 [0.60, 1.06]	.115
Age category (years)
18-34 (referent)	1	—
35-49	1.28 [0.98, 1.69]	.072
50-64	1.46 [1.08, 1.96]	.014
65+	0.87 [0.39, 1.96]	.743
Institution size
Small (referent)	1	—
Medium	0.61 [0.40, 0.94]	.027
Large	0.38 [0.24, 0.58]	<.001
Extra-large	0.29 [0.20, 0.43]	<.001
BMI category
Healthy weight (referent)	1	—
Overweight	0.69 [0.52, 0.92]	.010
Obese	0.55 [0.42, 0.72]	<.001
General health
Fair/poor (referent)	1	—
Good/very good/excellent	1.56 [1.08, 2.25]	.017

NOTE: OR = odds ratio; CI = confidence interval; BMI = body mass index. McKelvey and Zavoina’s R² = .11.

Challenge Completion

Odds ratios, confidence intervals, and p values from the model testing stages of change as a predictor of challenge completion are shown in Table 4. Number of chronic conditions and general health status were removed from the model during covariate selection because they did not significantly contribute. Employees in precontemplation, contemplation, and preparation were collapsed into one referent group because of low cell counts. Results from a likelihood ratio test comparing the final model to the constant only model were statistically significant, χ²(7) = 42.28, p < .001. Results indicated the odds for those in action/maintenance to complete the challenge were over four times higher compared with those in precontemplation/contemplation/preparation while holding other variables constant. The Hosmer–Lemeshow goodness-of-fit test was not significant suggesting the model fit the data.

Table 4

Physical Activity Challenge Completion by Stage of Change and Covariates (n = 985)

Stage of Change/Covariate	OR [95% CI]	p
Stage of change
Precontemplation/contemplation/preparation (referent)^a	1	—
Action/maintenance	4.21 [1.75, 10.13]	.001
Gender
Male (referent)	1	—
Female	0.50 [0.26, 0.98]	.043
Age category (years)
18-34 (referent)	1	—
35-49	2.48 [1.02, 6.02]	.046
50-64	3.46 [1.42,8.42]	.006
65+	1.17 [0.13, 10.67]	.886
Institution size
Small (referent)	1	—
Medium	0.64 [0.22, 1.92]	.43
Large	0.65 [0.22, 1.92]	.44
Extra-large	0.71 [0.29, 1.74]	.45
BMI category
Healthy weight (referent)	1	—
Overweight	0.25 [0.10, 0.58]	.001
Obese	0.43 [0.21, 0.86]	.018

NOTE: OR = odds ratio; CI = confidence interval; BMI = body mass index. Hosmer–Lemeshow statistic model fit results = χ²(8) = 8.39, p = .40.

This referent group was aggregated in part due to a small cell sizes.

Discussion

Results demonstrated stages of change were predictive of challenge sign-up and completion but not participation. Employees in preparation or action/maintenance had higher odds to sign-up for the challenge compared with employees in precontemplation/contemplation. Additionally, employees in action/maintenance had higher odds to complete the challenge compared with employees in precontemplation/contemplation/preparation. Results suggest the challenge was likely to reach employees already in the preparation/action/maintenance stages. Furthermore, employees who completed the challenge were more likely to have been in the action/maintenance stages, suggesting most of the challenge completers were already participating in regular PA.

Program developers used PA guidelines to inform program goals (U.S. Department of Health and Human Services, 2008). Program goals solely focused on meeting guidelines and not behavior change strategies could have affected sign-up and completion. Being physically active for 30 minutes/day for 5 days/week may not appeal to people in precontemplation or contemplation for PA. Past research suggests people intending to become active are interested in participating in worksite PA programs (Phipps, Madison, Pomerantz, & Klein, 2010). However, it is possible setting a goal that does not promote incremental PA increases may negatively affect engagement at multiple levels. Additionally, the challenge did not offer stage-matched strategies that can assist people in lower stages to progress to higher stages (Lippke, Schwarzer, Ziegelmann, Scholz, & Schüz, 2010).

Of the three hypotheses, only Hypothesis 2 was not fully supported. Stages of change approached statistical significance as a predictor of challenge participation (i.e., number of weeks a participant logged activity). Participants may have tracked their activity but did not enter it into the platform. Additionally, the participation variable treated all logged activity the same and did not account for the amount of activity logged. Results could have been influenced by these limitations; however, it is also possible stages of change is not predictive of logging activity during a PA competition.

Previous studies have reported stages of change is a significant correlate and predictor of PA (Marshall & Biddle, 2001; Trost et al., 2002; van Stralen et al., 2009). Our results extend the association between stages of change and PA to include multiple levels of engagement in a worksite PA competition. Other studies have also found stages of change to predict different levels of program engagement. One focused on program retention and outcomes with findings suggesting stages of change significantly predicted treatment sessions attended and pounds lost in a worksite weight control program (Prochaska et al., 1992). However, the weight control study provided individual treatment sessions for participants rather than a population level competition.

Another study evaluating predictors of weekly participation in a worksite pedometer-based intervention reported a greater percentage of employees in a higher stage of change participated for at least 1 week of the intervention than employees in a lower stage of change (Ablah et al., 2015). In addition, employees in a higher stage of change were more likely to achieve weekly step goals compared with employees in lower stages of change. These results are consistent with our study suggesting stages of change predict goal completion in a population level worksite PA program. However, our study also demonstrates stages of change predict program sign-up, a key component of program reach.

Limitations and Strengths

Limitations included self-report data for both the HA and PA logged during the challenge. Given the program was a competition, employees may have inflated their PA numbers to be more competitive. Additionally, the psychometric properties of the HA questions have not been formally tested so we cannot report the validity or reliability. The PA challenge had a low sign-up percentage (about 2% of the workforce). The low percentage could have been due to the promotion strategy, poor promotion strategy implementation, a lack of program interest, or some combination. The 2% sign-up greatly reduces program impact. However, the low percentage does not compromise the validity of results suggesting stages of change were predictive of challenge sign-up and completion for this sample.

We could not include all challenge participants in our analyses because many challenge participants completed their HA after the challenge began. Therefore, results may not represent the entire employee population and may be subject to selection bias. By only including people who completed the HA prior to the challenge, a time lag was introduced into the stages of change variable. Some participants completed their HA in February while the challenge began in May so it is possible participants moved stages by the time the challenge began.

To assess selection bias and determine whether the sample was representative of the total HA population (n = 12,504), we compared employees who completed the HA before and after the challenge across variables. This comparison demonstrated proportions for each respective variable were within a few percentage points of each other (<3%) between employees who completed the HA before versus after the challenge. Therefore, our sample appears to represent employees who took the HA throughout the System.

Despite limitations, the study has strengths. First, this study evaluated stages of change as a predictor across three levels of the engagement continuum for a PA competition. Therefore, results provide information across the engagement continuum that can be used to enhance program reach and impact. Second, we controlled for key confounding variables related to demographic and health information. Evidence suggests demographic and health-related variables can influence program recruitment and participation (Grossmeier, 2013; Robroek et al., 2009; Robroek et al., 2010). Third, we tested for differences across study variables between the study sample and the total HA population to address the potential selection bias. Results demonstrated our sample was representative of those who participated in the HA.

Conclusions

Based on our findings, PA competitions may be less appealing to employees not ready to change their PA behavior. Thus, PA competitions are more likely to reach an already active population. These findings are supported by a recent study reporting 92% of their competition participants were meeting PA recommendations (Macniven, Engelen, Rosen, & Bauman, 2015). Program organizers need to consider programs and program goals that are appropriate for all segments of the priority population. Moreover, programs focusing on smaller, incremental increases in PA as well as emphasizing social support rather than a competition may be more appropriate for people in precontemplation/contemplation/preparation. Additionally, adding program elements matched to participants’ stage of change rather than focusing on one goal could help engage more participants (Serxner, Anderson, & Gold, 2004). Finally, program organizers should use recruitment methods that target people in all stages of change. This objective can be accomplished by reaching out to employees in all stages through active recruiting (face-to-face contact) in addition to passive contact through newsletters and e-mails (Prochaska, Redding, & Evers, 2008). Recruitment materials should also include messaging to target people in all stages of change by highlighting stage-matched program elements.

This study demonstrated stages of change predicted sign-up for and completion of a worksite PA challenge. Sign-up and completion directly affect program reach and efficacy, which are two key components of the RE-AIM framework. Worksite program evaluations must extend beyond efficacy. We encourage future research to evaluate factors that influence sign-up and participation to reach all segments of the workforce. Improving knowledge in this area can lead to better program design, marketing, and implementation. Furthermore, additional research should evaluate worksite PA programs using all dimensions of the RE-AIM framework. Studies examining reach, efficacy, adoption, implementation, and maintenance will provide more comprehensive evaluations and inform program improvements.

Footnotes

Authors’ Note:

The authors would like to thank Laura Chambers, Faye Godwin, and Lance Rowell for their support and the University of Texas System, Office of Employee Benefits for providing funding support and access to the data. The authors disclose that Rolando Román is an employee of the University of Texas System’s Office of Employee Benefits (OEB). OEB organized and implemented the 2013 Physical Activity Challenge for the University of Texas System. OEB also provided the data for this article. This work was supported by a contract with the University of Texas System, Office of Employee Benefits (Jessica M. Tullar, PhD, PI).

References

Ablah

Dong

Konda

Armbruster

Tuttle

(2015). Early success is vital in minimal worksite wellness interventions at small worksites. Health Education & Behavior, 42, 500-509. doi:10.1177/1090198114564499

Carnethon

Whitsel

L. P.

Franklin

B. A.

Kris-Etherton

Milani

Pratt

C. A.

Wagner

G. R.

(2009). Worksite wellness programs for cardiovascular disease prevention: A policy statement from the American Heart Association. Circulation, 120, 1725-1741. doi:10.1161/CIRCULATIONAHA.109.192653

Glasgow

R. E.

Vogt

T. M.

Boles

S. M.

(1999). Evaluating the public health impact of health promotion interventions: The RE-AIM framework. American Journal of Public Health, 89, 1322-1327. doi:10.2105/AJPH.89.9.1322

Goetzel

R. Z.

Roemer

E. C.

Liss-Levinson

R. C.

Samoly

D. K.

(2008). Workplace health promotion: Policy recommendations that encourage employers to support health improvement programs for their workers. Retrieved from https://www.prevent.org/data/files/initiatives/workplacehealtpromotion-policyrecommendations.pdf

Grossmeier

(2013). The influence of worksite and employee variables on employee engagement in telephonic health coaching programs: A retrospective multivariate analysis. American Journal of Health Promotion, 27, 69-81. doi:10.4278/ajhp.100615-QUAN-190

Harris

J. R.

Wallace

R. B.

(2012). The institute of medicine’s new report on living well with chronic illness. Preventing Chronic Disease, 9, E148. doi:10.5888/pcd9.120126

Lippke

Schwarzer

Ziegelmann

J. P.

Scholz

Schüz

(2010). Testing stage-specific effects of a stage-matched intervention: A randomized controlled trial targeting physical exercise and its predictors. Health Education & Behavior, 37, 533-546. doi:10.1177/1090198109359386

Macniven

Engelen

Rosen

Bauman

(2015). Does a corporate worksite physical activity program reach those who are inactive? Findings from an evaluation of the global corporate challenge. Health Promotion Journal of Australia, 26, 142-145. doi:10.1071/HE14033

Marshall

S. J.

Biddle

S. J.

(2001). The transtheoretical model of behavior change: A meta-analysis of applications to physical activity and exercise. Annals of Behavioral Medicine, 23, 229-246. doi:10.1207/S15324796ABM2304_2

10.

Phipps

Madison

Pomerantz

S. C.

Klein

M. G.

(2010). Identifying and assessing interests and concerns of priority populations for work-site programs to promote physical activity. Health Promotion Practice, 11, 71-78. doi:10.1177/1524839908318165

11.

Prochaska

J. O.

Norcross

J. C.

(2001). Stages of change. Psychotherapy, 38, 443-448. doi:10.1037//0033-3204.38.4.443

12.

Prochaska

J. O.

Norcross

J. C.

Fowler

J. L.

Follick

M. J.

Abrams

D. B.

(1992). Attendance and outcome in a work site weight control program: Processes and stages of change as process and predictor variables. Addictive Behaviors, 17, 35-45. doi:10.1016/0306-4603(92)90051-V

13.

Prochaska

J. O.

Redding

C. A.

Evers

K. E.

(2008). The transtheoretical model and stages of change. In Glanz

Rimer

B. K.

Viswanath

(Eds.), Health behavior and health education (4th ed., pp. 99-121). San Francisco, CA: Jossey-Bass.

14.

Robroek

S. J. W.

Brouwer

Lindeboom

Oenema

Burdorf

(2010). Demographic, behavioral, and psychosocial correlates of using the website component of a worksite physical activity and healthy nutrition promotion program: A longitudinal study. Journal of Medical Internet Research, 12(3), e44. doi:10.2196/jmir.1402

15.

Robroek

S. J. W.

Lindeboom

D. E. M.

Burdorf

(2012). Initial and sustained participation in an internet-delivered long-term worksite health promotion program on physical activity and nutrition. Journal of Medical Internet Research, 14(2), e43. doi:10.2196/jmir.1788

16.

Robroek

S. J. W.

van Lenthe

F. J.

van Empelen

Burdorf

(2009). Determinants of participation in worksite health promotion programmes: A systematic review. International Journal of Behavioral Nutrition and Physical Activity, 6, 26. doi:10.1186/1479-5868-6-26

17.

Ryde

G. C.

Gilson

N. D.

Burton

N. W.

Brown

W. J.

(2013). Recruitment rates in workplace physical activity interventions: Characteristics for success. American Journal of Health Promotion, 27, e101-112. doi:10.4278/ajhp.120404-LIT-187

18.

Serxner

Anderson

D. R.

Gold

(2004). Building program participation: Strategies for recruitment and retention in worksite health promotion programs. American Journal of Health Promotion, 18, 1-6, iii.

19.

Taylor

W. C.

Hepworth

J. T.

Lees

Cassells

Gousse

Sweeney

M. M.

… Tobin

J. N.

(2004). Readiness to change physical activity and dietary practices and willingness to consult healthcare providers. Health Research Policy and Systems, 2(1), 2. doi:10.1186/1478-4505-2-2

20.

Terry

P. E.

Fowles

J. B.

Harvey

(2010). Employee engagement factors that affect enrollment compared with retention in two coaching programs: The ACTIVATE study. Population Health Management, 13, 115-122. doi:10.1089/pop.2009.0040

21.

Terry

P. E.

Grossmeier

Mangen

D. J.

Gingerich

S. B.

(2013). Analyzing best practices in employee health management. Journal of Occupational and Environmental Medicine, 55, 378-392. doi:10.1097/JOM.0b013e31828dca09

22.

Trost

S. G.

Owen

Bauman

A. E.

Sallis

J. F.

Brown

(2002). Correlates of adults’ participation in physical activity: Review and update. Medicine and Science in Sports and Exercise, 34, 1996-2001. doi:10.1249/01.MSS.0000038974.76900.92

23.

U.S. Department of Health and Human Services. (2008). 2008 Physical activity guidelines for Americans: Contents. Retrieved from http://www.health.gov/PAGuidelines/guidelines/default.aspx

24.

van Stralen

M. M.

De Vries

Mudde

A. N.

Bolman

Lechner

(2009). Determinants of initiation and maintenance of physical activity among older adults: A literature review. Health Psychology Review, 3, 147-207. doi:10.1080/17437190903229462

25.

Ward

B. W.

Schiller

J. S.

Goodman

R. A.

(2014). Multiple chronic conditions among US adults: A 2012 update. Preventing Chronic Disease, 11, E62. doi:10.5888/pcd11.130389