Following the intervention there was a significant decrease in total cholesterol, LDL-cholesterol, HDL-cholesterol, and systolic BP compared to pre intervention levels. The sample was comprised mostly of women making a sex comparison difficult, but because of the known sex differences in lipid levels , the data was analyzed for possible pre-post intervention differences by sex. Men had significantly larger drops in total cholesterol and LDL-cholesterol compared to women.
The decrease in HDL-cholesterol following the intervention was unexpected. However, the replacement of saturated fats with carbohydrates or polyunsaturated fats has been found to be associated with reductions in HDL-cholesterol levels [14, 15]. However, it can not be confirmed that this is the case in this study due to a lack of detailed dietary data from participants. In a meta-analysis of randomized-controlled trials, Kodama et al. documented that a minimum exercise length and energy expenditure levels are needed to produce positive effects on HDL-cholesterol levels . They also demonstrated that exercise effects on HDL-cholesterol are more pronounced in subjects with low BMI. In the current study, only 17% of participants performed exercise that fell in the vigorous activity range (≥ 3000 MET m/wk), based on responses to the IPAQ. In addition, 70% of the participants in the current study were either overweight or obese. These issues together with seasonal variation in blood lipid levels, may explain this observed effect on HDL-cholesterol . Future annual follow-ups should be conducted in the fall to alleviate concerns about seasonal variations in risk factors.
The increase in self-reported physical activity following the intervention was not statistically significant. However, the change in self-reported physical activity levels was not correlated with the change in CVD risk factors. Self-reported compliance is an issue in and of itself because people tend to over-report changes in exercise levels. Thus, the possibility can not be excluded that the moderately positive changes that were observed on CVD risk factors were due to the effect of the repeated measurement itself as opposed to an actual change in physical activity levels. It has been shown that repeated population screening is associated with reductions in CVD risk factors in high risk individuals . A Hawthorne effect whereby the mere presence of the intervention, not the intervention itself, is associated with favorable changes in outcome measures can therefore not be excluded. This is because people tend to work harder to make lifestyle changes when they are participants in an experiment because of the attention they receive, not because of the experimental manipulation itself. In addition, BMI did not change following the intervention. However, favorable changes in CVD risk profiles have been documented in response to an increase in physical activity levels despite a lack of effect on body weight .
Certain limitations of this study are worth noting. The screenings were conducted during different seasons in the same school year. Due to the seasonal variation in blood lipids and blood pressure [17, 20], these results should be taken with caution. To alleviate concern regarding seasonal variation, there is a plan of continuing the wellness program while conducting an annual screening in the fall of each school year. However, seasonal variation is related to differences in temperature, which were not a concern in this study since both pre and post assessments were conducted in the spring and late fall when extreme temperatures were not an issue.
One strength of this study is that two-third of the employees of the public school system chose to participate in the wellness program. However, we do not have information on those who chose to not take part in the screenings or walking program. The external validity of findings from research studies rests on the assumption that the participants represent the populations from which they were drawn. However, biases associated with non-response and drop-out rates may often result in overestimation of the effect of the interventions. It is quite possible that the changes we observed were caused by a "healthy participant effect" whereby participants report healthier lifestyles than non participants and are thus more likely to comply with the study protocol. This issue could be addressed if we had information on the fraction of the public school system employees who chose not to participate. However, this information is not available and their choice not to participate prohibited us from accessing any information related to them such as socioeconomic status, race, age or any other demographic information from which we could draw conclusion about the characteristics of the non-responders.
A high non-response rate to the IPAQ questionnaire at post intervention was most likely caused by allowing participants to complete it at a later time and not during the actual in-school screening. Although completing it during the screening sessions would have increased compliance, as it did with the pre intervention IPAQ questionnaires, this was not possible because the nurses wanted the screening sessions to be as short as possible because of time constraints of the teachers towards the end of the school year. Nevertheless, this high non-response rate raises the suspicion of a nonresponse bias. However, there were no significant pre-post intervention differences between responders and nonresponders in any of the CVD risk factors of interest. In addition, we did not have enough men in the sample to permit an adequate sex comparison. However, the composition of the work force in that public school system is such that there are a larger number of women than men employees. Finally, objective data on physical activity were not collected. The fact that self-report of exercise creates a smaller burden on the participants made the IPAQ a more suitable assessment method for this study. In addition, self-reported physical activity predicts cardiorespiratory fitness [21, 22]. Finally, lack of a control group limits the validity of the current findings. However, because of the participatory nature of this work and the desire of the school community to improve the health of as many interested employees as possible, it was not possible to randomize participants to a non-intervention arm. In addition, limited funds hindered our ability to involve a non-intervention group in another community.
The main strength of this community-based participatory research project was in the ability to interest two thirds of the public school employees in a physical activity promotion program, while maintaining a 95% response at 6-months follow-up. It is well recognized that physical activity is effective in reducing CVD risk. However, getting individuals to increase their activity levels is a challenging task that, along with changes in dietary factors, has sustained the obesity epidemic in the U.S. Community-based programs are needed to increase awareness and compliance with physical activity recommendations. If we exclude salaries of University investigators from the project costs, approximately $60,000 would be needed to implement this study at the community level. However, consideration needs to be given to the fact that the work of the 5 investigators would need to be conducted by the school nurses which means that they would need 29 full days to screen the 202 participants. This is not feasible, since the 2 nurses are serving 5 schools. If implementation of health promotion activities is to take place in the "real-world" using available resources, serious consideration needs to be given to policies affecting the availability of personnel and funds to support these efforts.