Eight-month postprogram completion: Change in risk factors for chronic disease amongst participants in a 4-month pedometer-based workplace health program

Authors

  • Rosanne Freak-Poli,

    Corresponding author
    1. Obesity & Population Health, BakerIDI Heart and Diabetes Institute, Melbourne, VIC 3004, Australia
    • Department of Epidemiology and Preventive Medicine, Faculty of Medicine, Nursing and Health Sciences, School of Public Health and Preventive Medicine, Monash University, the Alfred Centre, Alfred Hospital, Melbourne, VIC, Australia
    Search for more papers by this author
  • Rory Wolfe,

    1. Department of Epidemiology and Preventive Medicine, Faculty of Medicine, Nursing and Health Sciences, School of Public Health and Preventive Medicine, Monash University, the Alfred Centre, Alfred Hospital, Melbourne, VIC, Australia
    Search for more papers by this author
  • Margaret Brand,

    1. Department of Epidemiology and Preventive Medicine, Faculty of Medicine, Nursing and Health Sciences, School of Public Health and Preventive Medicine, Monash University, the Alfred Centre, Alfred Hospital, Melbourne, VIC, Australia
    Search for more papers by this author
  • Maximilian de Courten,

    1. Copenhagen School of Global Health, University of Copenhagen, 1014 K⊘benhavn K, Denmark
    Search for more papers by this author
  • Anna Peeters

    1. Department of Epidemiology and Preventive Medicine, Faculty of Medicine, Nursing and Health Sciences, School of Public Health and Preventive Medicine, Monash University, the Alfred Centre, Alfred Hospital, Melbourne, VIC, Australia
    2. Obesity & Population Health, BakerIDI Heart and Diabetes Institute, Melbourne, VIC 3004, Australia
    Search for more papers by this author

  • Author contributions: RFP takes responsibility for the integrity of the data and the accuracy of the data analysis. RFP, MdC and AP undertook the study design and oversaw the data collection for the project. MB contributed towards the data collection. RFP, RW and AP contributed to the statistical data analysis. RFP, RW, MB, MdC & AP contributed to the critical interpretation of the data. All authors contributed to the final version of the article and have read, as well as, approved the final manuscript

  • Disclosure: The authors declared no conflict of interest.

  • Funding agencies: The authors acknowledge the Australian Research Council (ARC) and the Foundation for Chronic Disease Prevention™ in the Workplace, which is associated with the Global Corporate Challenge®, for partially funding this study. RFP is supported by an Australian Postgraduate Award and a Monash Departmental Scholarship. AP is funded by a VicHealth Public Health Fellowship.

Correspondence: Rosanne Freak-Poli (Rosanne.Freak-Poli@monash.edu)

Abstract

Objective

To evaluate whether participation in a 4-month, pedometer-based, physical activity, workplace health program is associated with long-term sustained improvements in risk factors for type 2 diabetes and cardiovascular disease, 8 months after the completion of the program.

Design and Methods

A sample size of 720 was required. 762 Australian adults employed in primarily sedentary occupations and voluntarily enrolled in a workplace program were recruited. Demographic, behavioral, anthropometric and biomedical measurements were completed at baseline, 4 and 12 months.

Results

About 76% of participants returned at 12 months. Sustained improvements at 12 months were observed for self-reported vegetable intake, self-reported sitting time and independently measured blood pressure. Modest improvements from baseline in self-reported physical activity and independently measured waist circumference at 12 months indicated that the significant improvements observed immediately after the health program could not be sustained. Approximately half of those not meeting guidelines for physical activity, waist circumference and blood pressure at baseline, were meeting guidelines at 12 months.

Conclusions

Participation in this 4-month, pedometer-based, physical activity, workplace health program was associated with sustained improvements in chronic disease risk factors at 12 months. These results indicate that such programs can have a long-term benefit and thus a potential role to play in population prevention of chronic disease.

Introduction

Worldwide, workplace health programs that incorporate a pedometer are increasingly being utilized as a method of improving health, reducing absenteeism, and increasing productivity [1]. Short-term evaluations have reported immediate physical activity and chronic disease risk-factor improvements associated with participation in workplace health programs that incorporate a pedometer [1, 5]. However, it is unclear whether these improvements continue beyond the end of the program.

To date, there have been four long-term evaluations of workplace health programs that incorporate a pedometer, spanning between 12 weeks and 1-year postintervention [7, 11]. One evaluation assessed the effectiveness of a pedometer-based workplace health program in descriptive terms by assessing participant views and walking maintenance (via a “yes” or “no” answer) [14]. The other three evaluations assessed multicomponent health programs that incorporated a pedometer and reported improvements for physical activity and varying chronic disease risk-factors [7, 11]. However, the results from these three evaluations are not generalizable to the general population as each study restricts entry based on health status [7, 11]. A before and after analysis of a 4-month workplace pedometer-based health program, which did not restrict participation based on baseline health, found immediate improvements in self-reported physical activity, self-reported fruit intake, self-reported vegetable intake, self-reported takeaway dinner frequency, self-reported sitting time, independently measured blood pressure and independently measured waist circumference [1]. The aim of this article was to evaluate whether participation in this 4-month, team-based, pedometer-based, physical activity, workplace health program, results in long-term sustained improvements in risk factors for type 2 diabetes and cardiovascular disease, 8 months after the completion of the program.

Research Design and Methods

The Global Corporate Challenge® (GCC®) Evaluation Study is a prospective observational study. Here we assess whether there are sustained improvements associated with participation in the GCC®, 8 months after the completion of the program.

Study population

Melbourne workplaces were approached to be evaluation sites. Following receipt of the Workplace Consent, employees enrolled in the 2008 GCC® event were recruited.

A sample size of 720 participants was required to detect a 1 kg change in body weight and a 7 mmHg change in systolic blood pressure, with power of 0.90 and two-sided significance level of 5%, and allowance for 20% attrition.

Description of the program

The GCC® is a corporate organization that undertakes a pedometer-based workplace program [1, 15]. It is established world-wide, occurs annually and involves wearing a visible step-count pedometer for 4 months (May to September). The target is for each participant to achieve at least 10,000 steps per day, which is based upon the World Health Organization's recommendation [15]. Teams of seven employees enter their step-counts to virtually walk around a world map. Weekly encouragement emails are sent in the form of a newsletter which, in 2008, included the participant's personal best daily step count, a weekly health tip from a nutritionist, stories from participants, a “Dear GCC” answer to a participant's question, housekeeping notices and prizes awarded from sponsors. A website is used for logging daily steps, access to additional health information such as the number of steps required to burn off a hamburger, communication among participants and comparing team progress.

Recruitment

In 2008, 259 of the workplaces participating in the GCC® had an office located in Melbourne, 10 of which participated as evaluation sites for the GCC® evaluation study. Preference for approaching workplaces was based on early conscription to the Global GCC® 2008 event, a large number of employees, a variety of sedentary occupations, and the availability of a designated GCC® coordinator. Within the 10 evaluation sites, 4,138 people enrolled in the GCC®. Recruitment was targeted at adults who were employed at 10 Melbourne workplaces with relatively sedentary occupations. Participants were voluntary, aged 18 years and above and were enrolled in the GCC® 2008 Event at a participating workplace. Seven hundred and sixty two participants were recruited and assessed in April/May 2008 [2]. GCC® Evaluation study participants were similar to the other employees enrolled in the GCC® 2008 at the participating workplaces in terms of age and sex, but were more likely to comply with the step goal. Generally, the workplaces with more GCC® enrolees had lower recruitment rates into the evaluation (recruitment varied between 12 and 53% of GCC® enrolees). At 4 months, immediately after the completion of the GCC®, 79% (n = 604) returned [1].

Data collection

Data was assessed at baseline (prior to the GCC® 2008), at 4 months (immediately after completion of the GCC® 2008) and at 12 months (8 months after the completion of the GCC® 2008). An Internet self-report questionnaire incorporated:

Demographic information

  • Age, sex and education was assessed through the use of the WHO STEPwise approach [16] both core and expanded options.
  • Household status was assessed through the use of the Australian Diabetes, Obesity and Lifestyle Study [17] survey.
  • Marital status and parity was assessed through the use of the Melbourne Collaborative Cohort Study [18] questionnaire.
  • Occupation was assessed through the use of the Australian Standard Classification of Occupations [19].

Motivation and support

  • Specifically designed questions were developed to assess motivation for enrolling in the GCC®, the number of friends/work colleagues/family members enrolled in the GCC®, supports from workplace (for example time and encouragement) and whether the participant had enrolled in the GCC® in previous years.

Behavioral measures

  • Tobacco use, alcohol consumption and diet (fruit and vegetable) was assessed through the use of the WHO STEPwise approach [16] both core and expanded options.
  • Eating behavior was assessed through the use of Ball et al. [20] and the Melbourne Collaborative Cohort Study [18] questionnaire.
  • Physical activity was assessed through the use of the WHO STEPwise approach [16] both core and expanded options and the WHO mini-STEP [21] questionnaire. Participants who reported that they either undertook “30 min or more a day of moderate physical activities, 5 or more days a week" or "20 min or more a day of vigorous physical activities, 3 or more days a week" were considered to be reaching national Australian guidelines [22, 23].
  • Sedentary behavior was assessed through the use of the WHO STEPwise approach [16] core option and Ball et al. [20].

Health status

  • Blood pressure and diabetes was assessed through the use of the WHO STEPwise approach [16] both core and expanded options.
  • Coronary heart disease and high blood cholesterol was assessed using the formatting of the WHO STEPwise approach [16] both core and expanded options.

    • Physical Functioning was assessed through the use of the SF-12v2 [24].

Fasting anthropometric and biomedical measurements were collected by trained health practitioners during scheduled workplace visits between 7:30AM and 11AM. If participants were unwell on the day they were generally encouraged to reschedule on another day, either at their workplace or another participating workplace. Before testing, participants were asked to fast for 10-12 h, but no longer than 15 h, and to abstain from activities causing pain, smoking and exercise for 1 h before their appointment. Participants were asked to remove shoes, outer garments, such as jackets, and heavy items, such as jewelry and keys. For each participant, three blood pressure measures were recorded one minute apart using Omron IA1B Automatic blood pressure intellisense machines (Omron Corporation, Sydney, Australia) with Digitor AC 1000 mA DC output adaptors (Digitor, Sydney, Australia). Height was recorded to the nearest 0.1 cm up to 200 cm by using a stadiometer (portable height scale code PE087) and step ladder, ensuring that the top of the external auditory meatus (ear canal) was leveled with the lower margin of the bony eye socket. Salter electronic bathroom scales (model 913 WH3R 3007) during baseline and 4-month data collection and Seca digital scales (model robusta 813) during 12-month data collection were used to record weight to the nearest 0.1 kg up to 150 kg. Waist and hip measurements were recorded to the nearest 0.1 cm using a Figure Finder Tape Measure (Novel Products Inc 2005 code PE024) and a mirror to ensure that the tape was horizontal. Participants were asked to point out their lower rib margin and the top of the hip (iliac crest) so that the waist measurement could be recorded midway. Hip measurements were taken at the maximal circumference over the buttocks. Venipuncture samples of fasting glucose, total cholesterol, and triglycerides was assessed by Analytical Reference Laboratories, an independent pathology company. Analytical Reference Laboratories is accredited with the National Association of Testing Authorities and is in an independent quality-assurance program. Pathology testing was undertaken with Roche machines, which were tested twice daily using a third-party Qual from Biorad, at three levels. Measurement equipment was tested daily during data collection for differences in measurement and blood pressure machines were sent to Omron Corporation every 12 months for calibration. National guideline cut-offs are summarized in Supporting Information Table 1 [2]. The type 2 diabetes 5-year risk [25] and cardiovascular disease 10-year risk [26] prediction scores were calculated keeping age stable to assess the change associated with participation in the program and not being a year older.

Step information

GCC® enrollees entered daily step count into an online step diary as part of the health program. Bicycle ride length was also recorded and incorporated into the step-count as 6.4 km being equal to 10,000 steps.

Analysis

Analyses were performed using Stata version 11 (Stata Corporation, College Station, TX). Robust standard errors, clustered by workplace, were used in all statistical analyses including the calculation of P values and confidence intervals. P values <0.05 were interpreted as statistically significant. Pregnant participants (n = 4 at baseline, n = 9 new pregnancies at 4 months, n = 15 new pregnancies at 12 months) were excluded from all analyses.

Participants who did and did not return at 12 months were compared according to baseline variables and step data. For each chronic disease risk factor or risk outcome the change over time, between 12 month and baseline measurements, was analyzed. The primary analysis included participants who completed all behavioral, anthropometric and biomedical measures at the three data collection rounds. A sensitivity analysis was undertaken to evaluate the potential effect of missing data on the results using participants with complete data on the relevant variable/s at both baseline and 12 months. Differences in changes between baseline and 12 months in the primary analysis was assessed by age and stratified by sex and medical treatments.

Changes in continuous variables was assessed through linear regression. Pairs of baseline/12-month binary risk factor values was assessed for temporal change through conditional logistic regression with time-point as an explanatory variable. Changes between baseline and 12-month ordinal variables were also analyzed using conditional logistic regression, where an increase over time was represented by a 0, 1 pair of binary values, and a decrease was represented by a 1, 0 pair.

Ethics

The Monash University Standing Committee on Ethics in Research involving Humans approved this study.

Results

Twelve-month retention

Between April 20, 2009 and June 2, 2009, 76% (after restricting to non-pregnant participants: 558 of the original 734 at baseline) of participants returned for the 12-month data collection, Figure 1; 503 underwent anthropometric measurements, 478 biomedical measurements and 519 started the questionnaire (505 completed). Nearly 56% (n = 437) of participants completed all measures at 12 months, with greater retention observed in companies with more participants at baseline (retention range: 35-95%). About 315 (43%) participants completed all measures at baseline, 4 and 12 months. Those who participated in the 12-month data collection were more likely to be older, work in an inner city location, not moved workplace during the study, have a professional or associate professional position, participate in the GCC® “to look my best”, have a higher physical functioning at baseline, have a faster heart rate at baseline, not report having hypertension at baseline, have a higher CVD risk at baseline and undertake at least 10,000 steps per day on average during the health program, Supporting Information Table 2.

Figure 1.

Recruitment and retention. Restricted to participants who were not pregnant throughout the study period.

Change between baseline and 12-month

Between baseline and 12 months, improvements were observed for self-reported vegetable intake (an increase of 4.5% in the proportion meeting guidelines, P = 0.004), self-reported alcohol intake (7.0% more meeting guidelines, P = 0.01), self-reported nonsmoking (3.5% more meeting guidelines, P < 0.001), self-reported sitting time (weekday: mean −0.6 h day−1, P = 0.01; weekend: −0.4 h day−1, P = 0.01) and independently measured blood pressure (4.2% more meeting guidelines, P = 0.04; systolic: mean −3.8 mmHg, P < 0.0001; diastolic: −2.2 mmHg, P = 0.001), Table 1. Despite this, the proportion of participants that self-reported having hypertension increased (2.5%, P = 0.01, equates to eight participants newly reporting). The modest improvements from baseline to 12 months in self-reported physical activity (2.5% more meeting guidelines, P = 0.4) and independently measured waist circumference (−0.5 cm, P = 0.1) were not significant. Self-reported SF-12 mental health significantly decreased between baseline and 12 months (−6.3 SF-12 units, P < 0.001). Between baseline and 12 months, an increase was observed for independently measured total fasting cholesterol (mean 0.4 mmHg, P < 0.001) and independently measured triglycerides (mean 0.1 mmHg, P = 0.04), as similarly observed at 4 months.

Table 1. Twelve-month change in risk factors for and risk of cardiovascular disease and type 2 diabetes
 Baseline mean (SD) or percentageTwelve-month follow-up mean (SD) or percentageChange mean (SD) or percentageDifferencea (95% confidence interval)P value
  1. aOR denotes odds ratio.
  2. bAge kept stable to assess the change associated with participation in the program.
Baseline measures
Behavioral measures
Fruit Intake (meeting guidelines)33.037.24.2OR: 1.5 (1.0, 2.4)0.07
Vegetable intake (meeting guidelines)17.121.64.5OR: 1.7 (1.2, 2.4)0.004
Takeaway dinner
Once or less per month47.350.22.9OR: 1.5 (0.9, 2.6)0.1
About once a week40.041.31.3
More than once a week12.78.6−4.1
Alcohol (meeting guidelines)43.550.57.0OR: 1.9 (1.1, 3.1)0.01
Non smoker91.494.93.5OR: 6.5 (2.3, 18.4)<0.001
Physical activity (meeting guidelines)42.244.82.5OR: 1.2 (0.8, 1.9)0.4
Sitting time (hrs per day)
Weekday8.3 (3.7)7.6 (3.7)−0.6(−1.1, −0.2)0.01
Weekend5.4 (2.9)5.0 (2.7)−0.4(−0.7, −0.1)0.01
Psychosocial measures
Health related quality of life (SF-12)
Mental health component49.3 (10.1)43.0 (5.2)−6.3(−7.9, −4.7)<0.001
Physical health component51.4 (7.0)48.5 (4.9)−2.9(−4.0, −1.9)<0.001
Anthropometric measures
Systolic blood pressure (mmHg)119.2 (13.4)115.4 (12.6)−3.8(−5.3, −2.3)<0.001
Diastolic blood pressure (mmHg)80.0 (9.9)77.8 (9.9)−2.2(−3.1, −1.2)0.001
Blood pressure (meeting guidelines)82.286.44.1OR: 2.0 (0.8, 5.1)0.1
Heart rate (beats per minute)68.8 (10.1)68.3 (10.5)−0.5(−2.2, 1.2)0.5
Weight (kg)78.3 (15.8)78.2 (15.7)−0.1(−0.5, 0.3)0.7
Body mass index (kg m−2)26.9 (4.8)26.9 (4.8)0.0(−0.2, 0.1)0.7
Body mass index (meeting guidelines)39.738.4−1.3OR: 0.8 (0.4, 1.5)0.5
Waist circumference (cm)88.8 (12.8)88.3 (12.6)−0.5(−1.2, 0.1)0.1
Waist circumference (meeting guidelines)45.749.53.8OR: 1.6 (0.9, 2.8)0.1
Self-reported clinical status
Self-reported hypertension16.819.42.5OR: 1.5 (1.1, 2.0)0.01
Self-reported diabetes (type 1 or 2)5.46.40.9OR: 1.3 (0.8, 2.3)0.3
Biomedical measures (fasting)
Total cholesterol (mmol L−1)4.9 (1.0)5.3 (1.0)0.4(0.3, 0.5)<0.001
Total cholesterol (meeting guidelines)70.561.3−9.2OR: 0.3 (0.1, 0.7)0.005
Glucose (mmol L−1)5.1 (0.9)5.0 (0.7)−0.1(−0.2, 0.0)0.03
Impaired fasting glucose2.222.220.0OR: 1.0 (0.4, 2.3)1.0
Glucose (meeting guidelines)96.596.2−0.3OR: 0.8 (0.4, 1.7)0.6
Triglycerides (mmol L−1)1.2 (0.9)1.3 (0.8)0.1(0.0, 0.2)0.04
Triglycerides (meeting guidelines)78.172.1−6.0OR: 0.5 (0.3, 0.8)0.009
Predicted risk scoresb
Cardiovascular disease risk (next 10 years)
CVD risk (continuous)4.7 (4.8)4.5 (4.6)−0.1(−0.4, 0.1)0.2
High-risk1.01.30.3OR: 1.0 (0.7, 1.3)0.8
Intermediate-risk8.97.9−1.0
Type 2 diabetes risk (next 5 years)
Type 2 diabetes risk (continuous)7.7 (4.7)7.6 (4.6)−0.1(−0.5, 0.2)0.4
High-risk10.810.2−0.6OR: 1.0 (0.7,1.5)0.9
Intermediate-risk51.851.80.0

Sensitivity analyses

When complete case analysis was undertaken, the magnitude and significance of the changes between baseline and 12 months were generally similar to the primary analysis. Improvements in self-reported fruit intake reached statistical significance for sustained improvement (3.2%, P = 0.007) despite the estimated effect reducing slightly. The increase in the number of participants who self-reported that they had hypertension was sensitive to exclusion of individuals with missing data as it was not significant (0.8%, P = 0.6) in the complete case analysis, Supporting Information Table 3.

Change between baseline and 12 month in those at high baseline risk

When evaluating those at high risk at baseline, 25% of those not meeting physical activity guidelines (P < 0.001; Figure 2), 46% of those not meeting blood pressure guidelines (P < 0.001; Figure 3) and 19% of those not meeting waist circumference guidelines (P < 0.001; Figure 4) improved to meet guidelines at 12 months. For those not meeting guidelines at baseline, the mean 12-month change in systolic blood pressure (-8.5 mmHg, P < 0.001) and waist circumference (-1.85 cm, P < 0.001) were greater than the total sample (systolic blood pressure −4.5 mmHg difference, P < 0.001; waist circumference −0.99 cm difference, P = 0.02).

Figure 2.

Proportion meeting physical activity guidelines at baseline, 4 and 12 months.

Figure 3.

a: Proportion meeting blood pressure guidelines at baseline, 4 and 12 months. b: Systolic blood pressure at baseline, 4 and 12 months.

Figure 4.

a: Proportion meeting waist circumference guidelines at baseline, 4 and 12 months. b: Waist circumference at baseline, 4 and 12 months.

During the health program, 31% (n = 99) of participants increased their self-reported physical activity, 56% (n = 179) improved their independently measured blood pressure and 68% (n = 214) improved their independently measured waist circumference, all based on changes on a continuous scale. Within these participants who improved during the program, the proportion meeting guidelines at 4 months increased by 52% for physical activity (P < 0.001; Figure 2), 16% for blood pressure (P < 0.001; Figure 3) and 17% for waist circumference (P < 0.001; Figure 4). These improvements were sustained at 12 months for physical activity (31% improvement in proportion meeting guidelines from baseline P < 0.001; Figure 2), blood pressure (9% P = 0.04; Figure 3) and waist circumference (11% P < 0.001; Figure 4).

Sex and age comparisons

Forty-six percent (n = 146) of participants who completed baseline, 4- and 12-month data collection were male, Supporting Information Table 4. Changes in risk factors between baseline and 12 months were broadly similar between males and females, and between younger and older participants with only a few statistically significant interactions which are now described. Between baseline and 12 months, self-reported mental health decreased more for males (mean change −7.5 U vs. −5.2 U for females, P = 0.02 for interaction) and older participants (mean change −1.8 U per 10 years, P = 0.02 for interaction). Females improved in independently measured glucose (mean change −0.2 mmol L−1) while males remained stable (0.0 mmol L−1, P = 0.03 for interaction). Females improved in predicted type 2 diabetes risk (mean change −0.5 U) unlike males (+0.3 units, P = 0.03 for interaction).

Medical treatment

Of the 315 participants who attended baseline, 4 and 12 months, 32% (n = 101) self-reported they were taking at least one of the following medications; cholesterol (50 baseline, 57 twelve-months), blood pressure (23 at baseline, 30 at 12 months), diabetes (7 baseline, 6 twelve-months) and/or heart disease (4 baseline, 3 twelve-months). When analyses were restricted to participants not taking any of these therapies, the improvements in self-reported alcohol intake and self-reported sitting time reduced slightly from the observed change in the total population (4.7% increase in meeting guidelines and −0.3 h day−1, respectively) and lost statistical significance (Supporting Information Table 5). Additionally, the difference in the proportion meeting blood pressure guidelines at 12 months (6.5%, P = 0.007) was higher than in the total study population. When comparing outcomes between participants defined by their treatment status there was some evidence of a greater change in the proportion not meeting triglyceride guidelines between baseline and 12 months in those receiving medications when compared to those not on medication (−10.9% medicated participants verses −3.7% nonmedicated, P = 0.047 for interaction). Additional subanalyses assessing sustained change in independently measured blood pressure found benefits regardless of medical treatment status (systolic blood pressure 12 month mean ± SD change: medical treatment −3.9 ± 8.7 mmHg P = 0.001 versus nonmedical treatment −3.6 ± 10.9 P = 0.04; P = 0.9 for interaction), self-reported blood pressure specific medical treatment (medical treatment −3.6 + 13.0 P = 0.008 verses no medical treatment −3.8 ± 8.9 P < 0.001; P = 0.9 for interaction) or self-report hypertension status (self-reported −4.9 + 10.2 P = 0.001 versus not self-reported −3.4 ± 9.1 P = 0.001) although there was some evidence that the magnitude of benefit differs somewhat for the latter (P = 0.06 for interaction).

Discussion

In this evaluation of a 4-month, pedometer-based, physical activity, workplace health program, sustained improvements in self-reported vegetable intake, self-reported sitting time and independently measured blood pressure were observed 8 months after the completion of the program. Although modest improvements were observed for self-reported physical activity and independently measured waist circumference at 8-months postprogram, the significant improvements that had been observed immediately after the health program could not be sustained in the long-term. Self-reported SF-12 mental health, which had also improved immediately after the health program, had completely reversed to a decrease at 8-months postprogram. Long-term improvements in self-reported alcohol consumption and self-reported smoking, which were not observed immediately after the program, were observed at 8-months postprogram. The health benefits were present for males and females. From baseline the greatest improvements 8 months after the end of the program were seen in those who had shown improvements during the program itself. Sustained improvements in self-reported physical activity, independently measured blood pressure and independently measured waist circumference were greater for those who did not meet guidelines at baseline. In contrast to these benefits, a sustained increase was observed for independently measured total fasting cholesterol and independently measured fasting triglycerides. Although there were significant changes for individual risk factors, there was no evidence of corresponding improvements in composite cardiovascular disease and type 2 diabetes risk prediction scores.

As discussed in the introduction, there have been three long-term evaluations of workplace health programs that incorporate a pedometer and reported improvements for physical activity and varying chronic disease risk-factors [7, 11]. Two of these evaluations assessed long-term change in blood pressure associated with 12-week programs. Although Touger-Decker et al. [11] restricted eligibility to overweight and obese employees, they reported similar blood pressure benefits associated with the program, however, this was reported at a shorter period after the completion of the health program, at 26 weeks. Rush et al. [13] evaluated participants after a longer interval at 52 weeks, similarly to our study, however, they reported a significant increase in blood pressure. The lack of benefit in blood pressure may be due to Rush et al.'s [13] slightly healthier eligibility criteria of employees without type 2 diabetes, hypertension or medical use for cholesterol. Neither study [11, 13] reported a significant improvement in blood pressure directly after completion of the program, potentially due to power. Hence, our results are the first to demonstrate sustained blood pressure benefits associated with participation in a workplace health program that incorporates a pedometer. Additionally, if a control group reflecting Safar et al.'s [27] findings of an average yearly increase in systolic blood pressure between 0.2 and 0.5 mmHg was compared to our participants, it could identify a larger magnitude of benefit associated with the health program.

The sustained improvement in blood pressure correlates well with the lifestyle improvements associated with the health program [22]. Physical activity has been shown to significantly improve during workplace programs that incorporate a pedometer [7, 11, 13] and be sustained at 1-year postprogram [7]. In this evaluation although physical activity and waist circumference changed in a beneficial direction at 8-months postprogram, the significant improvements immediately after the health program were not sustained in the long-term for the overall population. Psychosocial mental health also significantly improved immediately after the health program, however scores significantly decreased at 12 months in the overall population. However, we did find indications that participating in this pedometer workplace health program was associated with long-term improvements in vegetable intake, reduced alcohol consumption, non smoking and reduced sitting time. Additionally, fruit intake was observed to significantly improve in the sensitivity analysis although with a smaller magnitude of improvement than in the primary analysis where the result was not statistically significant, which may simply reflect sample size. Positive program effects have been reported upon dietary behavior immediately after a workplace program that incorporates a pedometer and sustained in the long-term, including significant improvements in high-energy dense sandwich topping consumption [7], total calorie intake, total fat intake and carbohydrate intake [11]. In the present study the changes in self-reported alcohol, fruit and vegetable intake suggest an improvement in diet. The GCC® provides additional health information, including dietary information, which may have encouraged fruit and vegetable intake, as might a general increase in health awareness associated with participation [28, 29]. However, as dietary intake was not the primary aim of this study, limited dietary data were collected. Our findings of suggested sustained improvements in vegetable intake and reduced sitting time, as well as, the long-term improvements in alcohol consumption and nonsmoking associated with a pedometer workplace program are novel as these factors have never been assessed.

Although a sustained significant improvement was not observed for waist circumference at 12 months, the change was in the direction of benefit and was of an encouraging magnitude. Other long-term evaluations of workplace program that have incorporates a pedometer have reported significant sustained improvements in waist circumference of 3.7 cm at 14-weeks postprogram in overweight and obese employees [11] and 0.5 cm at 1-year postprogram [12]. Rush et al. [13] reported a nonsignificant increase of 0.2 cm at 10 months postprogram, however, there was no change directly after the program nor was adiposity improvement an aim of their program. Potentially without the health program there may have been an average increase or worsening in waist circumference of 0.6 cm in men and 1.2 cm in women per year as suggested by Ekelund et al. [30] or 0.3-0.5 cm per year as suggested by Kwark et al.'s [12] control group. Similarly to waist circumference, in our study weight and BMI changed in an encouraging direction but the improvements were not statistically significant. Only one long-term evaluation of a workplace program that incorporated a pedometer has reported a significant benefit in weight of 2.1 kg in overweight and obese employees [11]. The other two evaluations reporting weight or BMI at 10-12 months after the end of the health program observed similar results to ours, a reduction of 0.29 kg [12], 0.1 kg [13], and 0.11 BMI units [7] with the results not reaching significance. Given that annual weight gain is ∼0.4-0.5 kg [31, 32] and 0.2-2.4 BMI units [33, 34], no increase in adiposity may be a beneficial outcome for our study participants.

No change in blood glucose, composite CVD or type 2 diabetes risk scores and sustained increases in total fasting cholesterol and triglycerides were associated with participation in this program. Only two other long-term evaluations of workplace programs that incorporate a pedometer have assessed biomedical measures, reporting contrasting results. Similarly to our study, Touger-Decker et al. [11] reported no significant change in blood glucose directly after the program or at the post program follow-up in overweight and obese employees, however, they also reported no significant changes in total cholesterol and high-density lipoprotein. Rush et al. [13] reported very similar triglyceride results to our study; a 0.1 mmol L−1 increase at 10-months postbaseline. However, Rush et al. [13] reported a significant benefit to total cholesterol of 0.3 mmol L−1 and blood glucose of 0.2 mmol L−1 at 10 months associated with their program. Only one other pedometer workplace program evaluation has assessed CVD risk [11], also reporting a lack of change at 14-weeks postprogram in overweight and obese employees. There are a number of theories on the possible cause of the association between increased walking and increased cholesterol and triglycerides including, [1] increases in high-density lipoprotein cholesterol [35] which was not measured in this evaluation; [2] the requirement of a threshold level of activity to see decreases in these factors [9] which is not being met by walking; [3] changes in food access and eating behavior [36]. Having observed no immediate improvement in biomedical measures directly after the program in our study [1], a long-term benefit in biomedical measures associated with the program was not expected.

The main potential limitation of this article is the lack of a control group. As discussed previously [1], it is not possible to definitively conclude that improvements observed in this study were attributable to participation in the program. However, as discussed above, the typical population change in many of these anthropometric and biomedical measures is an increase or worsening over time [12, 30]. Hence, this article is possibly underestimating the anthropometric and biomedical health benefits associated with the program. Another potential limitation of this study is the influence of the evaluation itself. It is possible that by asking detailed questions and distributing individualized results our evaluation may have exposed or increased the interests of participants in regards to their health, which may have encouraged changes beyond that of the health program. For example, eight participants newly reported hypertension at 12 months, indicating that some participants identified their risk status during follow-up. However, when these individuals were removed from the analysis no difference in findings was observed.

An additional potential limitation of this study is the selection bias associated with workplace recruitment, individual recruitment and participant retention [1, 2]. As expected, participants who returned at 12 months were more likely to be older and have healthier baseline measures than those who did not return [1, 12, 39]. Our attrition rate (21% at 4 months and a further 3% at 12 months) was better than most other long-term evaluations of workplace programs that incorporated a pedometer; 22% (13), 28% (12), 38%(11), 77% (14). It is difficult to predict the effect of selection biases as a healthier, more motivated cohort would be more likely to undertake the goal of 10,000 steps (as observed), leading to overestimation of health benefits, while a healthier cohort would already be meeting health guidelines at baseline (as observed) with less room to improve, thereby leading to underestimation [1, 2]. By undertaking a primary analysis and the sensitivity analysis, we have attempted to assess some of the effect of response bias upon the results. The primary analysis, which included participants who completed all measures at the three data collection rounds, offers simplicity as all calculations proceed from a common base [40]. However, there are some disadvantages to this approach, especially as a high proportion of cases were deleted (59% in our database); the results may be biased if the deleted cases differ from the complete cases (some differences were found in our database, Supporting Information Table 2), and the precision of model estimates will be lower due to the smaller sample size. By undertaking the sensitivity complete case analysis we have increased the sample size and attempted to reduce the bias, however as there was 21% missingness in the database, bias and lack of precision may still be present. Limited differences were observed between the primary and sensitivity analyses; however we recognize that a large controlled study should be utilized to confirm these findings.

Another limitation to this study is the change from bathroom weight scales used at baseline and 4 months to validated weight scales at 12 months. Any consequent measurement error would reduce the likelihood of detecting a true difference in weight at follow-up.

The strengths of this long-term evaluation were the range and quality of measurements, the large sample size and the variety of sedentary occupations within the sample [1, 2]. An additional strength is that baseline and 8-month postprogram measurements were undertaken at the same time of year, which would reduce seasonal effects. Through demonstration of sustained improvements in a range of risk factors for CVD and type 2 diabetes, across a range of sedentary occupations, the current study suggests that a large controlled study, measuring a similar range of behavioral, anthropometric and biomedical chronic disease risk factors with long-term evaluation is worthwhile.

Conclusions

Participation in this 4-month, pedometer-based, physical activity, workplace health program was associated with sustained improvements in vegetable intake, sitting time and blood pressure which are risk factors for type 2 diabetes and cardiovascular disease. Sustained improvements were greater for those who were not meeting risk-factor guidelines at baseline and for those whose risk factor status improved during the program itself. These results indicate that such programs have a potential role to play in population prevention of chronic disease prevention.

Acknowledgments

The study design, analysis and interpretation of data, the writing of the manuscript, and the decision to submit the manuscript for publication were solely at the discretion of the Monash researchers, independent of GCC® or The Foundation's involvement. RFP, RW, MB & AP are affiliated with Monash University. MdC is affiliated with the University of Copenhagen. No further conflicts of interest were reported by the authors of this article.

Ancillary