EVALUATING THE EFFECTIVENESS OF AN INTERVENTION PROMOTING WALKING AND LIFESTYLE PHYSICAL ACTIVITY IN A PRIMARY CARE PRACTICE
Kirsten J. Black
B.A., University of California, Berkeley, 1982 M.P.H., University of California, Berkeley, 1989
A thesis submitted to the University of Colorado at Denver in partial fulfillment of the requirements for the degree of Doctor of Philosophy Health and Behavioral Sciences
This thesis for the Doctor of Philosophy degree by Kirsten J. Black has been approved by
Black, Kirsten Johnson (Ph.D., Health and Behavioral Sciences)
Evaluating the Effectiveness of an Intervention Promoting Walking and Lifestyle Physical Activity in a Primary Care Practice
Thesis directed by Professor Craig Janes
Sedentary behavior is a risk factor for many chronic diseases and conditions including cardiovascular disease, hypertension, diabetes, obesity, osteoporosis, colon cancer and depression. Nevertheless, only 25% of US adults achieve recommended levels of physical activity and 29% report no regular leisure activities.
This study used a sequential explanatory design to evaluate the effectiveness of a 13-week intervention to increase lifestyle physical activity among primary care patients. The primary quantitative outcome was change in average daily step count over time. Interviews were conducted after the intervention and 6 months later to explore participants experiences with the intervention.
Of 83 enrollees, 33 completed the intervention; 26 dropped out after submitting at least one week of data; and 24 never submitted data. Between group comparisons were significant for age F(2, 76) = 4.07,/? = .021, BMI F(2, 78) = 3.49,p =.035, race x (1, N =83) = 10.13,/? < .001, and baseline step count t(56.9) = 2.026,/? <.047 (two-tailed). Eight-eight percent (n = 59) of participants were in the lowest categories of activity (59% sedentary, 29% low activity) at baseline. Self-reported activity level was poorly correlated with baseline activity rs (n = 59) = .256,/? = .0502. Comparison of baseline steps with daily steps averaged over the last 4 weeks of the intervention showed study completers increased their average daily steps by 1,850.3 (95% Cl = 1,024.2, 2,676.4,/? <.001).
Wearing step counters increased activity awareness and motivated behavior change. Positive health benefits, including weight loss, improvements in chronic disease markers and subjective experiences of feeling better were associated with attitudinal changes that reinforced behavior change. Although step counters motivated physical activity behavior change during the intervention, different patterns of use were observed in the maintenance phase. During the intervention, participants developed strategies to deal with stable physical activity barriers, however the maintenance phase was often characterized by emergent situations that interrupted regular patterns of physical activity.
The clinical implications of this study are that personalized information coupled with educational messages encouraging self-monitoring and realistic goal setting can motivate physical activity behavior change for sedentary individuals.
This abstract accurately represents the content of the candidates thesis. I recommend its publication.
I dedicate this dissertation with love to my children, Kevin and Sarah Black.
Many friends and colleagues have been instrumental in supporting me through my dissertation. First, I would like to thank the members of my committee, Craig, David, Miriam, Chris and Wilson for giving generously of their time, sharing their expertise and guiding me through this process.
I would also like to acknowledge the support that I have received from the Department of Family medicine. If it takes a village to raise a child, the department has been my village. Thank you Colleen for believing in me and encouraging me to take this step. Thank you Wilson for your vision, wisdom, and mentorship and for marshalling resources to help with this project. Thank you Miriam for your patience in helping me deal with statistics. Mark, thanks for your enthusiasm for physical activity which was the spark for this endeavor. Thank you Sherry for guiding me through the administrative details. Thanks also to Deb, Laura, Kristen and Erica for being supportive friends.
I also appreciate the support that I received from Jim Hill and others at the Center for Human Nutrition. Thank you Jim for sharing your vision and expertise during the conception of this research, and for the resources you provided. One of the most enjoyable aspects of this study was collaborating with Helen, Joanie,
Mary and Martha. Thank you Helen and Joanie for your consistent encouragement. Thank you Mary for showing me the ropes and for always being available when I needed someone to talk to. Finally, thank you Martha for teaching me about qualitative research and for taking the time to help me during the course of this study.
I am also thankful for the funding that I received from the American Academy of Family Medicine, (Practice Based Research Network Research Stimulation Grant G0205PB) that supported the intervention phase of this research.
Significance of the Study...............................3
2. DEFINITIONS AND LITERATURE REVIEW..........................9
Definitions Used in the Study of Physical Activity......9
Activity and Obesity............................13
The Relationship between Weight and Health.......15
Fitness Recommendations Contrasted with
Weight Loss Recommendations......................16
Fitness in the Absence of Weight Loss Fitness as a Public Health Target......
Physical Activity Recommendations..................19
Exercise vs. Lifestyle.............................23
Walking for Exercise...............................26
Using Step Counters to Promote Physical Activity...29
Colorado on the Move...............................30
Establishing Step Counter Activity Guidelines......32
An Ecological Perspective..........................36
Promoting Physical Activity in Primary Care........40
Improving Behavioral Interventions in Primary Care.42
Social Cognitive Theory............................48
The Transtheoretical Model.........................57
Maintenance of Physical Activity Behavior Change...59
3. RESEARCH DESIGN AND METHODS..................................62
Mixed Methods Paradigm Conundrum..........................63
Sequential Explanatory Design.............................66
Human Subjects Review .................................68
Participants and Site..................................68
Dependent and Independent Variables....................70
Phase I: Protocol and Data Collection (Quantitative)..71
Phase II: Protocol and Data Collection (Qualitative)..73
Data Collection, Analysis and Integration.....................74
Integration of Data....................................85
4. QUANTITATIVE RESULTS..............................................86
Definition of Study Completion Groups..................86
Differences between Study Completers,
Non-completers and Non-reporters.......................92
Testing the Validity of Week 1 Data
as a Proxy for Baseline Activity.......................94
Relationships between Demographic Variables
and Baseline Step Counts............................94
Comparison of Weekly Step Counts between Completers and Non-completers........................96
Change in Step Counts................................98
No Relationship between Change in Step Counts
and Demographic Variables...........................99
Change in Activity Categories.......................100
Relationship between Self-Identified Activity
Level and Baseline Step Count.......................100
5. QUALITATIVE RESULTS............................................102
Qualitative Data Sample.............................103
Physical Activity Behavior Changes................ 116
Barriers to Increasing Physical Activity............122
Benefits of Being in the Study......................130
Future Plans for Physical Activity..................141
Long Term Impacts...................................142
A Conceptual Model.................................146
Evaluating the Study Using the
6. DISCUSSION AND STUDY LIMITATIONS.............................172
Interpreting Quantitative Findings.................172
Reaching People Who are Sedentary..................174
Patterns of Step Counter Use Over Time.............180
Addressing Practical Questions.....................187
Questions for Future Research.............................192
Study Delimitations and Limitations.......................194
Threats to Validity................................195
A. INTERVIEW GUIDES...................203
B. CODELIST.......................... 205
C. CROSS TABULATION OF QUALITATIVE
AND QUANITATIVE DATA MATRIX........216
D. STEP COUNTER USE MATRIX............219
3.1 Sequential Explanatory Design...............................67
4.1 Reasons for Dropout by Week.................................89
4.2 Change in Average Daily Step Count Over Time................96
4.3 Amount of Change in Average Daily Steps.....................98
4.4 Percent Change in Average Daily Steps.......................99
5.1 Relationship between Affective Outcomes....................137
5.2 Conceptual Model of Showing Relationships between
2.1 Linking SCT Constructs with Components of the
4.1 Participant Characteristics............................... 91
4.2 Baseline Activity Level.....................................93
4.3 Mean Daily Steps During Week 1..............................94
4.4 Comparing Demographic Factors with Baseline
4.5 Average Weekly Step Counts for Non-completers
4.6 Change in Activity Categories..............................100
B. 1. Code Definitions...........................................206
C. 1. Cross Tabulation of Qualitative and Quantitative
D. 1. Step Counter Use Matrix....................................220
Sedentary behavior is a strong risk factor for many chronic diseases and conditions including cardiovascular disease, hypertension, diabetes, obesity, osteoporosis, colon cancer and depression (Eden, 2002). Inactivity accounts for an estimated 200,000 excess deaths each year (Dishman, 1996). Consequently, Healthy People 2010 lists physical activity as one of five leading health indicators (United States Department of Health and Human Services [USDHHS], 2000). A review of 44 studies published between 1966 and 2000, reported that all cause mortality was 20-30% lower in people who met Healthy People 2010 guidelines for physical activity (30 minutes of moderate activity five or more times per week, or 20 minutes of vigorous activity three or more times per week). Although the review was based on observational studies and therefore subject to potential confounding by prior health status as well as other health behaviors, the data are consistent with other studies that have controlled for these confounders (Lee, 2001).
During the last twenty years, the average level of physical activity for Americans declined significantly. Today, most Americans are not sufficiently
active to promote health or maintain their weight. Furthermore, because norms for physical activity have declined, many people are oblivious to their own inactivity. Not surprisingly, reduced levels of physical activity have been associated with increasing rates of obesity and obesity related chronic diseases.
According to national data, only 25% of adults in the United States achieve recommended levels of physical activity and 29% of adults report no regular activity during leisure time (USDHHS, 1996). In response to these low levels of activity for adults, Healthy People 2010 targets the following three objectives. Along with each objective, national statistics from 1997 (HSDHHS, 2000) as well as Colorado data collected in 2001 are reported (Colorado Department of Public Health and Environment, 2003).
Objective 22-1: Reduce to at least 20% the portion of adults aged 18 and older who engage in no leisure-time physical activity (U.S. baseline 40%, Colorado 19%).
Objective 22-2: Increase to at least 30% the proportion of people aged 18 and over who engage in regular, preferably daily, moderate physical activity for at least 30 minutes per day (U.S. baseline 15%, Colorado 38%).
Objective 22-3. Increase to 30% the proportion of adults who engage in vigorous physical activity that promotes the development and
maintenance of cardiorespiratory fitness, three or more days per week, for 20 or more minutes per occasion (U.S. baseline 23%, Colorado 29%).
Based on data from 2001, Colorado has met two of the objectives and is very close to meeting the third. In comparison to other states, Colorado has a smaller percentage of overweight or obese adults. Furthermore, adults in Colorado report higher levels of activity compared to those in other states. However these optimistic findings mask the disparities among different population groups. Both within Colorado and nationally minorities are most at risk for sedentary lifestyles, and diseases associated with being sedentary. Nationally, sedentary lifestyle is more prevalent among women (30.7%) than men (26.5%), individuals with low incomes (41.5% for incomes < $10,000 yr. compared with 17.8% for incomes > $50,000), and among the less educated (46.5% for < 12 years of school vs. 17.8% for college graduates). People who have chronic diseases are also among those who report the lowest levels of physical activity (USDHHS, 1996).
Significance of the Study
Despite the wide body of evidence demonstrating the physical and mental health benefits of physical activity, the vast majority of Americans are inactive. Furthermore, efforts to promote physical activity during the past 15 years have
had virtually no impact on physical activity rates in the United States. One of the reasons for this is that previous approaches to physical activity have promoted exercise and by so doing have failed to engage the public. For example, on average half the people who start an exercise program drop out within the first few months (Dishman, 2001). The failure of traditional approaches to increase physical activity rates suggests the need for new strategies based on theoretical models that account for the complexity of behavior associated with physical activity.
One new strategy for reversing inactivity and improving health in the United States is to promote lifestyle physical activity (LPA). LPA consists of activities done within the context of ones everyday life. Promoting LPA involves encouraging people to find simple ways of incorporating activity into their lives. Associated with the concepts of LPA, step counters have been promoted as tools for measuring daily activity. Step counters are relatively inexpensive devices (costs are approximately $10.00-$30.00) that are worn inconspicuously on clothing to track the number of steps a person takes each day. They can be used for assessment as well as for monitoring ambulatory forms of physical activity.
Step counters have become popular in Colorado due to extensive media coverage of Colorado on the Move, an initiative targeting obesity through promotion of lifestyle physical activity and calorie reduction. Since its inception
in October of 2002, over 50 work sites and 12 communities across the state have participated in Colorado on the Move. The success of this initiative has led to the development of a national organization: America on the Move. In a very short time, interest in step counters has grown tremendously.
Despite the popularity of step counters there are few published studies examining peoples experiences using counters either short term or long term. Tudor-Locke (2001) one of the few published researchers in this field, states that one of the needs for future research is to better understand the experience of using a step counter:
If pedometers are to be used as motivational or intervention tools, information concerning participant compliance and experiences (though qualitative methods such as structured interviews or focus groups) should be provided to assist with interpretation.
The present study was conceived in early 2002 based on the positive
results and wide public interest in Colorado on the Moves community and work
site interventions. Recognizing the success of these interventions, there was
considerable interest in adapting the intervention strategy for use in primary care.
At that time there was no published research in this area. The purpose of this
study was to both evaluate the effectiveness of using step counters to promote
physical activity in a primary care setting as well as to provide information about
participants subjective experiences.
A team consisting of James Hill, PhD and his colleagues at Colorado on the Move, Wilson Pace, MD and Mark Cucuzzella, MD, two primary care physicians and myself developed the intervention strategy that was used for this study. Dr. Hill was instrumental in providing the educational resources used in the intervention, such as funding for development of the Step Your Way to Better Health video and supplying Colorado on the Move pamphlets. Dr. Pace organized the automated data collection activities including the database and telephony system.
This study uses mixed methods to evaluate the effectiveness of an intervention designed to increase lifestyle physical activity among primary care patients. The 13-week intervention consists of using a step counter to measure baseline activity level, setting individual goals and self-monitoring. Educational messages emphasize the value of small changes and incorporating lifestyle changes to increase physical activity. The primary quantitative outcome is change in average daily step count. Interviews are used to assess individual experiences both immediately following the intervention and again 6 months later. Information gained from this research will help to inform future interventions utilizing step counters.
The study utilizes a sequential explanatory design to address the following aims:
1. Compare and contrast characteristics of completers and non-completers of the 13-week intervention.
2. Determine if activity levels change during the 13-week intervention period and whether changes are maintained at 6 months.
3. Examine the accuracy of self-perceptions of physical activity and evaluate the extent to which the intervention changes perceptions.
4. Explore the subjective experiences of participants during the intervention and follow-up phase of the study.
5. Propose intervention strategies for using step counters in a primary care practice.
As a health educator and dietitian, I bring particular biases to this research. Foremost among these is a patient centered orientation, which characterizes my approach to both patient care and research. My training and clinical experience has taught me to begin with the individual and to tailor behavior change to their unique circumstances. For me it is difficult to study a single aspect of a persons life, such as their physical activity without
consideration of the larger context in which it occurs. The qualitative portion of this study was an opportunity to explore individual stories while at the same time looking for broad themes. I discovered however that the complexity of individual life stories does not always allow reduction to simple themes.
DEFINITIONS AND LITERATURE REVIEW
The field of physical activity research is complex and the terms used may be misunderstood. Therefore, this chapter begins by clarifying the meanings of terms and concepts that are used in this thesis. This is followed by a discussion comparing and contrasting exercise and lifestyle approaches for promoting physical activity with people who are sedentary. Recognizing that individual behavior change occurs within a larger social and environmental context, a social ecological framework is used in the next section to discuss proximal and distal factors that influence physical activity promotion. The final section explores some of the difficulties in applying current theories of behavior change to the study of physical activity.
Definitions Used in the Study of Physical Activity Physical activity refers to bodily movement produced by the skeletal muscles that results in energy expenditure that can range from low to high (Pescatello, 2001). Two subsets of physical activity are exercise and lifestyle. Exercise is a form of physical activity that is planned and structured. It is usually
performed during ones leisure time for the purpose of obtaining or maintaining fitness. In contrast, lifestyle physical activity (LPA) refers to everyday activities consisting of leisure, occupation and/or household as well as planned and unplanned activity. The high degree of variability in types of physical activity performed makes its assessment difficult (Dishman, 2001).
Exercise is categorized as either resistance or endurance. Although both types of exercise confer health benefits, endurance activities have been more widely studied because of their association with cardiovascular disease reduction. In endurance research, the units used to measure exercise intensity are metabolic energy equivalents (METs). One MET is defined as the resting metabolic rate (3.5 ml oxygen per kg body mass per minute). Low intensity activities such as light house work or leisurely walking have METs < 4; moderate intensity activities such as walking at a pace of 15-20 min per mile have METs of 4.0-5.9; and vigorous activities such as running have METs > 6. With increasing fitness cardiorespiratory efficiency improves. Consequently, people with low fitness levels experience more exertion and get more benefit from low to moderate intensity activities compared with those at higher fitness levels.
Physical fitness is an attribute that can be defined variously depending on whether the outcome goal is health or performance. Higher levels of physical activity are associated with increased levels of physical fitness. Components of physical fitness that contribute to health outcomes include: cardiorespiratory
endurance, muscular endurance and strength, body composition and flexibility. Higher levels of physical fitness promote cardiovascular benefits and reduce health risks. Studies that use physical fitness as their primary endpoint typically measure changes in cardiorespiratory capacity. The main measures used for this are maximum oxygen uptake (VO2 max), resting heart rate and endurance.
Sedentary is a term that does not have a specific definition but is frequently used to refer to individuals who do not meet the recommendations for physical activity (20 minutes of vigorous activity three days per week, or accumulating 30 minutes of moderate activity on five or more days of the week). There are no standard criteria for categorizing people as sedentary based on step counts, however Tudor-Locke and Bassett (2004) have suggested that sedentary be defined as achieving <5,000 steps per day.
Sedentary behaviors are defined as those activities that require very little physical activity to perform. Prevalent sedentary behaviors in the United States include sleeping and eating, television and video viewing, using the computer, reading, sitting and talking on the telephone.
Body Mass Index (BMI) is calculated based on a persons height and weight according to the following formula: weight (kg)/height (m ). The correlation between BMI and percent body fat is generally good (0.6-0.9).
Higher levels of body fat are associated with increasing health risks; therefore, BMI is used to predict health risk. Based on health outcome data, BMI
categories and their associated level of health risk are: < 18.5 underweight (low), 18.5-24.9 normal (average), 25-29.9 overweight (increased), 30-34.9 obesity I (moderate), 35-39.9 obesity II (severe), > 40 obesity III (severe).
A step counter is a mechanical device, usually battery operated that measures the number of steps taken. Although the term step counter is frequently used interchangeably with the term pedometer, the two are not the same. A pedometer is a battery operated mechanical device that calculates the distance traveled by multiplying the number of steps taken by the length of a persons stride.
Physical activity promotion addresses two distinct public health problems, namely inactivity and obesity. The inextricable relationship between physical activity and weight along with societys preoccupation with weight has resulted in physical activity being viewed as primarily a mediator of weight loss. This is unfortunate because the beneficial effects of physical activity are broader than simply weight loss and accrue at lower levels than are typically required for weight loss. Furthermore, people who are not overweight often do not appreciate their need for physical activity. Thus, there is a need to promote physical activity in a way that is appropriate to engage the general public. One potentially effective strategy emphasizes increasing lifestyle physical activity. In contrast to
traditional approaches to increasing physical activity that have promoted regular exercise, the lifestyle approach lowers the bar so that more people are encouraged to increase activity. This strategy may also be the most realistic way of addressing the obesity epidemic. According to Hill (2003), simply encouraging people to increase their activity by the equivalent of 100 calories a day would be sufficient to help reduce weight gain. Furthermore, this modest amount of activity may represent a necessary first step for helping sedentary individuals increase their fitness and self-efficacy for physical activity.
Activity and Obesity
Physical activity is closely associated with obesity. Low rates of physical activity coupled with over consumption of energy are believed to be the underlying causes of the obesity epidemic in the United States (Hill, 1998). Currently, over 56.4% of Americans are considered overweight or obese (Mokad, 2001). The factors associated with this epidemic are complex and deeply rooted in our society, however at the individual level, weight gain results from an imbalance between energy consumption and energy utilization.
Over the last 20 years, the amount of energy required for daily living in industrial countries has declined by approximately 800 kcal/day (James, 1995). However, energy consumption has not matched this decline, and as a result the average American adult gains approximately two pounds per year. While it is
necessary to intervene on both sides of the energy balance equation, there are
many who believe that targeting physical activity is essential for success. For
example, Kumanyika (2001) speculates that there is:
....a certain minimum level of physical activity needed in the society at large in order for the average person to maintain energy balance, and that the ability to compensate for low energy expenditure by maintaining a low level of energy consumption may be limitedparticularly if the eating pattern contains energy-dense, high fat foods.
Although I am not aware of any studies demonstrating that a certain physical activity level is required to maintain energy balance, this hypothesis is conceptually appealing. From an evolutionary perspective it is clear that historically humans have been an active species. It is only within the last 100 years that physical activity has declined so significantly. Consequently, it is likely that our physiological mechanisms are better adapted for activity than sedentary behavior.
The importance of physical activity to energy balance is also supported by data from the national weight loss registry. According to these data the people who are most successful at weight loss and weight loss maintenance are those who have high levels of physical activity (McGuire, 1998). The exact mechanisms involved in this relationship are not completely understood, however it is clear that higher energy expenditure helps to compensate for dietary indiscretions. People who are sedentary have a very small margin for managing
energy balance and this poses a problem since our society promotes energy dense foods. Thus while interventions aimed at both sides of the energy equation are necessary, I believe that physical activity may represent a threshold issue for reducing obesity.
The Relationship between Weight and Health
The association between weight and health risk is moderated by the percentage and distribution of body fat. Studies that focus on change in weight or body composition, either alone or in addition to fitness typically measure body mass index (BMI), abdominal girth and/or percentage of body fat. Although evidence links each of these measures with mortality and morbidity, taken together they provide a more comprehensive profile of health risk. For example, there is a small gradual increase in morbidity and mortality for BMI between 25 and 30. For BMI between 30 and 35 the risks increase more rapidly and above 35 they increase dramatically. These findings are the basis of the current definitions of overweight and obesity (overweight is defined as BMI 25-30; obesity class I is BMI between 30-35, class II is 35-40). However for an individual, BMI is insufficient for determining health risks, especially for people in the overweight category. This is because the metabolic effects of obesity are related to the amount, type and distribution of body fat. People who are at greatest risk from being overweight or obese are those with centrally distributed
body fat (apple shape) (Reaven, 2003). Central or visceral fat has unique metabolic characteristics that increase the risk for type 2 diabetes and heart disease. For this reason measures of abdominal girth and body fat are important corollaries to BMI.
When changes in weight or body fat are the outcomes of interest, physical activity is often measured in terms of calories burned. At the individual level, calories burned can be calculated from measures of oxygen consumption per minute, however this is impractical for many studies. Therefore calories burned are typically estimated based on the persons weight, the intensity of the activity and the amount of time spent doing the activity. For example, a 150 lb person who walks at a moderate pace for 18 minutes bums about 100 kcalories (kcals).
Fitness Recommendations Contrasted with Weight Loss Recommendations
Physical activity recommendations are written in terms of frequency, duration and intensity. Current recommendations from the Surgeon General and the American College of Sports Medicine (ACSM) state that Americans should accumulate 30 minutes or more of moderate intensity activity on most, preferably all days of the week (USDHHS, 1996). The goal of these recommendations is to improve fitness; they are not designed to promote weight loss. Although some people may lose weight at this level of activity, most people will need to do
significantly more activity in order to lose weight. Since weight is determined by both energy intake and energy expenditure, recommendations for weight loss usually consist of moderate calorie reduction (250-500 kcals a day) combined with increased energy expenditure (250-500 kcals per day) (Leermakers, 2000). This equates to approximately 60 minutes of moderate activity, which is significantly more than is recommended for fitness. In order to achieve the recommended activity level for weight loss a person needs to have a moderate to high level of fitness. Consequently, the first step in increasing activity for weight loss should focus on improving fitness rather than setting exercise goals that may be unrealistic for a sedentary person. For many people, this first step may be the only step and so we must consider the value of fitness as a goal separate from weight loss.
Fitness in the Absence of Weight Loss
Physical fitness is not directly associated with weight, therefore it is possible to be physically fit yet also overweight or obese. Similarly it is possible to have a normal body weight and be unfit. Fitness and weight loss are separate outcomes that independently impact health. In a consensus statement issued by the ACSM the following benefits of improving fitness in the absence of weight loss were noted: 1) improves insulin action and reduces insulin resistance in obese persons, 2) attenuates the progression from impaired glucose tolerance to
type 2 diabetes, 3) reduces blood pressure, and 4) reduces the amount of visceral and abdominal subcutaneous fat (Gundy, 1999).
Two studies performed at the Cooper Center have shown that fitness protects against mortality for men who are overweight and obese. In one study, men with BMIs >30 who were moderately or highly fit were compared with normal weight men who were unfit (measured as in the lowest 20% of their age group). Death rates for the unfit men were 52 per 10,000 man-years of observation while death rates for the fit but obese men were 18 per 10,000 (Barlow, 1995). A larger study that followed 21,925 men age 30-83 years over an 8 year period found similar results (Lee, 1993). According to the studys authors, Obese men who were fit had a much lower risk of cardiovascular disease (CVD) and all cause mortality than unfit men in the lowest body fat group. The authors caution however that these results should not be interpreted to suggest that physical activity eliminates all the health risks associated with obesity. For example, cardiovascular fitness will not lessen the effects of weight on osteoarthritis or sleep apnea.
He and Baker (2004) analyzed data from 7,867 adults aged 51-61 over a four year time period and found that regular physical activity was associated with better health outcomes even for people who do not lose weight. The health benefits of activity were also apparent for people who did light activity. For example, the rate of decline in overall health for those who reported doing no
activity was 20.8% whereas it was 8.4% for those who did light exercise on average three or more times a week.
Fitness as a Public Health Target
In the same way that a sedentary person must first develop fitness before s/he can expend sufficient energy to facilitate weight loss, our public health approach to activity goals for sedentary individuals should begin by emphasizing fitness rather than weight ioss. Although weight loss is an optimum goal, it is often unattainable. Lack of motivation and inability to attain adequate activity levels prevent many people from losing weight; however there are also people who are highly fit yet whose weight is intractable no matter how hard they work at losing. One of the advantages to promoting fitness is that the amount of activity necessary for fitness is less than is recommended for weight loss. Most importantly, there is strong evidence to show that fitness confers significant health benefits even in the absence of weight loss. Focusing on activity rather than weight may motivate a greater percentage of the population to change sedentary behaviors.
Physical Activity Recommendations
Regular physical activity reduces the risk of cardiovascular disease, type 2 diabetes, colon cancer and several major chronic diseases and conditions. It is
also provides mental health benefits and is inversely associated with mortality. These relationships are dose dependentgreater benefits accrue to people with the highest activity levels (USDHHS, 1996; Lee, 2001). Historically this dose response relationship has been the basis for recommending high intensity activity, however in recent years there has been a growing consensus that physical activity need not be vigorous in order to provide health benefits.
Morris (1958) did some of the earliest research linking moderate physical activity with health. Following 9,376 male workers over 9 years he demonstrated a relationship between higher levels of occupational activity and lower mortality. Although this study has been criticized because of the likelihood of confounding by other factors associated with occupation such as diet, the relationship has been corroborated by other epidemiological studies. The Harvard Alumni Study demonstrated that men who regularly expended 500-35,00 kcal per week had significant risk reduction for premature mortality from coronary artery disease (CAD). The relationship between activity and health benefits was dose dependent, and the study demonstrated that even low levels of activity conferred health benefits (Paffenbarger, 1986). In the Multiple Risk Factor Intervention Trial (MRFIT), men who averaged 30 minutes a day of moderate intensity activity had one third fewer deaths from CAD and a 20% lower overall mortality rate compared to inactive men (Leon, 1987).
Two large cohort studies found similar results for women. In the Womens Health Initiative Observational Study, a diverse sample of 73,743 postmenopausal women (aged 50-79) was followed over a 5.9 year period. In the study, coronary events were inversely associated with energy expenditure.
Women who reported walking briskly or engaging in vigorous exercise for 2.5 hours per week had a 30% lower risk of coronary events. The Nurses Health Study followed 80,348 women over a 20 year period (Rockhill, 2001). The authors report that moderate intensity activities conveyed approximately the same mortality benefit compared to vigorous activities. They found that increasing levels of physical activity were inversely associated with mortality. Interestingly, mortality risk for women in different categories of activity showed the greatest incremental benefit between women in the lowest level of activity (<
1 hour of activity per week) and those only slightly more active (between 1 hour and 1.9 hour per week). Thus, for sedentary individuals, even small increases in activity can provide health benefits.
Lee and Skerrett (2001) reviewed 44 studies to determine whether there was a linear dose response relationship between the volume of physical activity or fitness and mortality. Although they used only observational studies, they did find a dose response relationship. Furthermore, the greatest benefit from increased activity appeared at the lowest levels of activityat higher levels of activity there appeared to be a threshold effect. They estimate that minimal
adherence to physical activity recommendations leading to energy expenditure of 1000 kcal/week results in decreased all-cause mortality rates with risk reductions on the order of 20-30%. They also found evidence to suggest that even lower volumes of activity may have health benefits.
Experimental studies have compared multiple short bouts of physical activity with an equivalent amount done continuously. DeBusk, Stenestrand, Sheehan, and Haskell (1990) found that improvements in cardiovascular fitness were similar for 40 men who were randomized into either short bout or long bout activity groups. Jakicic, Wing, Butler and Robertson (1995) compared both fitness improvement and adherence rates between intermittent and continuous exercisers in a 20-week study. He found that improvement in fitness were similar although adherence was better for the group assigned to intermittent exercise (234 min/week compared to 188 min/week).
In 1995, an NIH consensus panel was commissioned to review new evidence and recommend changes in the Surgeon Generals physical activity recommendations. The committees report is summarized in the first Surgeon Generals Report on Physical Activity and Health published in 1996 (USDHHS,
1996). The report made three significant changes from earlier recommendations. First, it reduced the intensity of activity recommended from 60% VO2 max to 50% VO2 max and recognizes that 40% VO2 max is an appropriate level for people with low fitness. Second, it increased the frequency of activity by
encouraging activity on most if not all days of the week. Third, it included the option of accumulating activity over the course of a day rather then at one time. The following section considers two different implementation strategies for increasing activity among sedentary adults.
Exercise vs. Lifestyle
There are two main approaches to improving physical fitness. The first is a structured approach that emphasizes regularly planned physical activity exercise. The second utilizes a lifestyle approach. The lifestyle approach differs from exercise because it is not based on prescribed regular structured activity. Instead it emphasizes accumulating 30 minutes of daily activity through everyday activities. Both methods provide similar health benefits; however, from a public health perspective, the lifestyle approach has advantages that suggest it would be a better strategy for increasing activity among people who are sedentary. This section compares and contrasts these two approaches with regard to health benefits, motivation and sustainability.
The dominant paradigm for promoting physical activity has historically been exercise. To promote fitness, people were taught that they must exercise at 60% of maximum age-predicted heart rate. Tables and formulas were published so people could calculate target heart rates. However this approach has not been effective for increasing activity among the majority of the population.
Epidemiological studies show that only 15% of the population is active at this level, 60% of the population is inadequately active and 25% were completely non-active (USDHHS, 1996). One of the reasons for these low levels of activity is that people interpreted the recommendations as promoting vigorous activity, which many find both unappealing and unrealistic. Even after the recommendations were broadened in 1996, researchers discovered that most people were not exercising at the amounts prescribed by the ACSM Exercise Prescription guideline because of a misperception that vigorous exercise was their only alternative (Dunn, 1998).
High intensity activity is one of the barriers to exercise that is frequently identified. Intensity can be a barrier both for initiating as well as sustaining physical activity. For sedentary people who intended to exercise, but did not the most common barriers were beliefs that exercise would be too physically demanding and lack of time (Godin, 1986). And, of the relatively small number of people who begin a vigorous physical activity program, at least 50% of them drop out within one year (Dishman, 1988).
Other barriers to exercise include: lack of time, too tiring, lack of social support, inclement weather, disruptions in routine, and lack of access to facilities (Dishman, 1994; Trost, 2002). Many of these barriers are minimized or alleviated by the lifestyle physical activity approach. Moderate rather than vigorous activity is promoted. People are encouraged to accumulate at least 30
minutes of moderate intensity activity each day. This can consist of short bouts or one long bout. Activities are self-selected. They can consist of everyday activities such as housework as well as structured activities. The lifestyle approach therefore offers flexibility in meeting ones goals. Furthermore it allows people to perform activities that are purposeful. Purposeful activities are shown to be associated with greater long-term adherence among sedentary individuals (Morgan, 2001).
Two controlled studies comparing lifestyle physical activity with structured exercise interventions found that participants in both groups had similar health outcomes. Dunn, et al. (1999) conducted a two year controlled trial in which 235 healthy sedentary people aged 35-60 years were randomized into either lifestyle physical activity or structured activity. For each group there was a six month intensive intervention period followed by an 18 month less intensive period. The main findings for the study were that there were similar changes in physical activity and cardiorespiratory fitness over the 24 months. In a study by Anderson et al. (1999), 40 sedentary obese women were randomized into either lifestyle or structured physical activity interventions and were followed for one year. All women received similar dietary recommendations. At the end of the study period both groups showed similar improvements in systolic blood pressure, serum lipids, and lipoprotein levels. After 6 months the exercise group had slightly higher weight loss (8.3 kg compared with 7.9 kg); however, at
one year the amounts were similar because the aerobic group had a higher amount of weight regain (1.6 kg in the aerobic group compared to 0.08 kg in the lifestyle group). These studies help to show that lifestyle activity confers similar health benefits as traditional exercise programs.
Other evidence supporting the value of lifestyle activity comes from a meta-analysis of 127 physical activity studies by Dishman and Buckworth (1996). They state that effect sizes for interventions promoting active leisure (similar to lifestyle physical activity) are larger compared with exercise programs. Similarly, they found that studies with low intensity activities reported larger effects compared with studies using higher intensities of activities. They also note that behavioral intervention studies had better effect sizes.
Walking for Exercise
Walking is the most popular form of physical activity and it an integral part of lifestyle physical activity. The following paragraphs will briefly summarize some of the extensive literature devoted to walking. This research not only supports the health benefits of walking, but also demonstrates the utility of promoting walking with sedentary people.
In a review of data from the 1990 Behavioral Risk Factor Surveillance System (BRFSS), Siegel, Brackbill, and Heath (1995) found that walking was the
most popular form of physical activity. Furthermore the relative prevalence of walking was highest among population subgroups that report the lowest levels of participation in leisure time physical activity, such as low income, less educated, minority, elderly and overweight. The authors suggest that promotion of walking for physical activity might be an underused tool for reaching the most sedentary groups in the population. These findings were corroborated by Salomon, et al. (2003) who found that 86% of survey respondents reported high enjoyment of walking while only 31% reported high enjoyment of structured physical activity.
Henderson and Ainsworth (2003) performed qualitative interviews to identify perceptions of walking with African American and American Indian women. They found that attitudes toward walking were positive. Women stated that it is something that can be done anywhere and most women interviewed did not consider walking as exercise. The authors state that: walking was noted as an important physical activity undertaken by a range of women of different ages and from different economic backgrounds and family situations.
The health benefits of walking have been consistently demonstrated. In the US Nurses Health Study, walking was inversely associated with coronary events and the relationship was stable for women irrespective of race, age or BMI (Manson, 1999). Similar results for men were found in the Honolulu Heart Study. Men who walked >1.5 miles a day had half the risk of heart attack as those who walked < 0.25 miles a day (Hakim, 1999).
Hartman (2001) summarized the literature on walking in elderly populations stating:
Walking is especially suitable for older people and can lead to improved quality of life. Among older people regular walking has been associated with lower rates of hospitalization, lower plasma triglycerides and higher bone mineralization.
She also argues that many sedentary people are physically unable to undertake
more vigorous activities because of poor cardiorespiratory fitness and/or excess
weight and that for them walking is sufficiently vigorous to promote fitness.
One of the hypothesized benefits of promoting walking for physical
activity is sustainability. Since moderate intensity activities like walking are
easily incorporated into ones daily routine, there are fewer barriers to
participation, therefore maintenance should be enhanced. In a 32-week study
where 32 sedentary adults were randomized to three groups: 1) 30 minutes of
continuous brisk walking, 2) three 10 minute bouts of walking and 3) any
combination of activity equaling 30 minutes, Coleman, et al. (1999) found that
participants in all groups made similar gains in activity and cardiovascular
improvement. Across all groups those who were able to make walking part of
their daily lifestyle where more successful at meeting activity goals. In
particular, participants reported that to be successful, they needed to develop the
habit of choosing walking instead of sedentary activities. Another aspect of the
study was that participants stated that they were better able to meet their walking
goals during the week because they encountered more unexpected time demands on weekends.
Using Step Counters to Promote Physical Activity
The concept of a device to measure steps may have originated from drawings by Leonardo de Vinci (Wilson, 1999). Thomas Jefferson is reported to have purchased a pedometer in France and brought it back to the United States. Then in the late 1930s a popular radio program named Jack Armstrong offered hike-o-meters (pedometers) as a promotion. During the promotion, approximately 70,000 daily orders, requiring 10 cents and 2 Wheaties box tops, were received (Norms Radio, 2003).
The first large-scale commercial production of pedometers occurred in Japan in the 1960s. These were given the name manpo-kei, which means ten-thousand steps meter (Tudor-Locke, 2004). It is likely that the concept of walking 10,000 steps originated from this name. In Japan, walking 10,000 steps a day continues to be a popular goal. In the United States, some of the earliest uses of pedometers in the medical field were to measure ambulation for patients in post surgery or post injury situations. In the late 1990s, step counters were reintroduced to the American public as a tool to motivate physical activity.
At the forefront of the current wave of step counter popularity is Colorado on the Move. The primary objective of Colorado on the Move is
reducing obesity. James Hill, the founder of this movement believes that the current obesity epidemic originates from an energy gap. He defines the energy gap as the extent to which energy balance must be modified to prevent weight gain (Wyatt, 2004). According to his calculations, using data from residents of Colorado, he hypothesizes
... .the energy gap could be closed by modifying energy balance by 100 kcal/d... Assuming 50% efficiency of energy storage by the body, a reduction of positive energy balance of about 80 kcal/d form the Colorado population would be sufficient to close the energy gap for most people. (Wyatt, 2004).
The premise of Colorado on the Move is that this amount of energy
expenditure can be obtained from increasing lifestyle physical activity.
Colorado on the Move
Colorado on the Move is a statewide intervention that uses step counters to encourage physical activity. Jim Hill, PhD at the Department of Human Nutrition at the University of Colorado Health Sciences Center initiated the program in 2001 after pilot data showed that people who used step counters increased their activity levels by an average of 2,000 steps over a 12 week period. Over 75,000 people throughout the state have participated in Colorado on the Move interventions in various setting such as worksites, schools, churches and even whole communities. Participating groups include: the Metro Black
Church Initiative, the Colorado Department of Health, Coors Brewing Company, the city of Peetz, and the Southern Ute Tribe. The success and popularity of Colorado on the Move has led to its transformation and expansion into a national programAmerica on the Move.
Results recently published by Colorado on the Move showed that participants in both work site and church interventions increased their steps by approximately 2,000 steps per day (Wyatt, 2004). According to Hill and colleagues, a 2,000 step increase is hypothesized to be sufficient, on a population level, to help prevent weight gain, however the study did not collect outcome data on weight.
Although qualitative data has not yet been published, preliminary summaries of focus group data provide insights into different ways people use step counters and raise concerns about sustainability (M Tenney, personal communication, April 2003). Some of the unpublished findings from focus groups of participants in Colorado on the Move are:
People use step counters different ways. After wearing step counters each day during the 12 week intervention period, people transitioned to different patterns of use: every day, intermittent, temporal and non-users.
Despite the fact that people had different ways of utilizing step counters they agree that the step counter helped them to pay more attention to their activity level.
Although the main benefits of the step counters were health related, people identified non-health related benefits as well such as increasing moral at work, feeling more in control of their health.
In a community or work site setting, certain factors were identified that enhanced the program. These included having organizational support, incentives, and publicity for the program.
When the program is advertised as a 15-week program, participants are less likely to maintain behavior changes compared to if it is advertised as a lifestyle program.
These findings reinforce the importance of learning more about peoples motivations, expectations and actual experiences with the intervention before getting caught up in the current wave of enthusiasm. Although people are initially very interested in using a step counter to learn about their activity, one of the major issues to be addressed is how to keep people motivated over the long run. Taking the time to explore peoples experiences with step counters is essential to planning a successful clinical intervention.
Establishing Step Counter Activity Guidelines
Evidence suggests that step counters can help motivate people to increase their activity, however there are currently no evidence-based guidelines on which to base step count recommendations. As described above, the recommendation
of 10,000 steps per day has historical roots, and there is evidence showing that
10.000 steps is associated with important health outcomes such as lower blood pressure, and less body fat (Tudor-Locke, 2004). However, studies have also demonstrated that there are health benefits at step levels below 10,000 steps. Sugiura, et al. (2002), reported improvements in lipid profiles for middle-aged women who had steps counts of approximately 9,000 steps/day. Although more data from studies evaluating the relationship between step counts and health outcomes are needed, one concern about establishing a single number as the recommendation is that it would not cover the entire population. Although
10.000 steps/day is believed to be a reasonable goal for healthy adults, it probably overestimates the amount of activity needed for older adults and those with chronic disease. It also underestimates the activity needs of children whose step equivalent for activity has been estimated as 15,000 steps/day (Tudor-Locke, 2004).
One of the difficulties in establishing recommendations for activity based on steps is determining what method to use. For example, the Surgeon Generals report recommends that Americans get 30 minutes of physical activity each day. Thirty minutes of walking at a pace of three miles per hour is equivalent to approximately 3,000 steps. This amount of activity should be in addition to the activity associated with daily living. Tudor-Locke (2004) reports that for healthy adults in the United States, activities of usual daily living account for between
6,000-7,000 steps. Adding 3,000 extra steps then boosts activity to 9,000-10,000 steps. This reasoning supports the 10,000-step recommendation, however it is based on the premise that the populations average activities of daily living are between 6,000-7,000 steps. Unfortunately, many of the people who are most in need of physical activity are likely to fall significantly below average. People whose activities of daily living are 4,000-5,000 steps a day would then need to increase their steps by 6,000-5,000 in order to achieve 10,000 steps. This would require about an hour of walking. Although this level of activity would probably be good for them, it may not be a realistic or sustainable goal and it conflicts with the Surgeon Generals recommendation.
Another approach to establishing physical activity recommendations is to encourage people to increase their activity by a specific amount. James Hill (2003) is a proponent of recommending that people increase their steps by 2,000 over baseline. As discussed above, this recommendation is designed to target the energy gap. Walking 2,000 steps takes about 20 minutes; therefore this amount of activity is lower than the Surgeon Generals recommendations. Arguments in favor of 2,000 steps are that it is probably more realistic and sustainable than
10,000 steps and that it appreciates the fact that people start from different baselines. One argument against recommending that people increase their steps by an average of 2,000 steps per day is that people who start out with very low activity levels (4,000 steps/day) would still only reach 6,000 steps and this may
not be sufficient to promote health. The two approaches to establishing recommendations could be combined by encouraging people to increase their steps gradually by 2,000 step increments until they reached an appropriate longterm goal. The long-term goal would be individualized based on the persons age, and health.
Realizing that currently there are no standards to evaluate step amounts, Tudor-Locke and Bassett (2004) have proposed preliminary indices to use to classify pedometer-determined physical activity in health adults. The criteria that she proposes are:
(i) < 5,000 steps/day may be used as a sedentary lifestyle index;
ii) 5,000-7,499 steps/day is typical of daily activity excluding volitional sports/exercise and might be considered low active;
iii) 7,500-9,999 likely includes some volitional activities (and/or elevated occupational activity demands) and might be considered somewhat active;
iv) > 10,000 steps/day indicates that point that should be used to classify individuals as active.
v) individuals who take > 12,500 steps/day are likely to be classified as highly active.
Currently these are the only indices available, consequently they have been adopted in this thesis.
An Ecological Perspective
There is an inherent contradiction between promoting lifestyle physical
activity when lifestyle is also hypothesized to be the cause of inactivity. If
lifestyle is to blame for Americans inactivity and obesity, then perhaps changing
ones lifestyle to incorporate everyday activities may not be as easy as it is
thought to be. This question will be addressed by discussing some of the social
ecological factors that influence activity. According to Green (1996):
The ecological or transactional view of behavior holds that the organisms functioning is mediated by behavior-environment interactions. This view has two implications for behavioral and social change:
The environment largely controls or sets limits on the behavior that occurs in it
That changing environmental variables results in the modification of behavior.
Although there are a variety of well-founded theories linking environmental influences with physical activity, the concept of the built environment is currently receiving a great deal of attention (King, 2002). Examples of some of the more traditional social-environmental theories include theories of environmental stress, social disorganization, and availability of community resources. Environmental stress theory posits that neighborhood crowding, noise, and traffic congestion negatively affect physical activity. Theories of social disorganization emphasize the harmful effects of
neighborhood crime and violence on social support and health promoting behaviors such as physical activity. Physical factors that are positively related to activity include access to open space, side walks, recreation centers, and physical features such as those that increase social cohesiveness and social capital (Giles-Corti, 2003).
Employing a perspective from the Healthy Cities movement, researchers have begun examining the impact of the built environment and city planning on activity. It is clear that over the last century, changes in urban planning have paralleled the increase in obesity. Examining the association between obesity and urban sprawl, Lopez (2004) found that the risk for being overweight increased by 0.2% and the risk for being obese increased by 0.5% with each point rise in the urban sprawl index. Furthermore, historians note that during the last century there was an intentional separation of the environment into living areas and commerce areas. Although this was done for the purpose of creating healthier living environments by moving people away from polluted manufacturing sectors, separating housing from work increased dependence on the automobile (Perdue, 2003). This has had a number of negative effects on physical activity. For instance, most people comminute to work by carthey live too far away to walk or ride a bike. The time they spend commuting takes away from the time they have for physical activity and commuting itself frequently increases stress. Even if people want to be more active, there are
often environmental barriers since most cities and suburbs are designed to accommodate cars rather than pedestrians or bicycles.
Studies show that most people who walk do so for either transportation or recreation (Giles-Corti, 2003). The idea of walking for transportation is closely linked with lifestyle physical activity. However, the ability to implement lifestyle physical activity is highly dependent on the built environment. People who live in communities where they are in close proximity to places where they need to go can more easily chose to walk or ride a bike compared to those living in the suburbs where things are too spread out to get to by walking or bike riding.
Although there is increasing recognition of the importance of the built environment on physical activity (and health in general), little is known about how to change the environment to enhance physical activity. One recommendation is that public health perspectives should be included in community development and planning activities. Also there is the recognition that implementing changes in the built environment will require changes in zoning and other legal ordinances. Economic incentives must also be considered to counteract societal trends favoring sedentary activities (Sturm, 2004). Thus, it is apparent that there must be a multidisciplinary approach to creating the environmental changes that are required to support a more active community.
One of the important, although not surprising, findings from focus groups of Colorado on the Move participants is the role of the micro-environment in
supporting behavior change. Participants in work site interventions stated that a supportive work environment helped to facilitate increased activity. Factors that were mentioned as helpful included: healthy competitions, incentives, and creating a step culture. Research has also shown that placing cues to action such as a sign encouraging the use of stairs can help increase activity (Marcus, 2003).
Creating health-promoting environments is essential for sustainability of individual level interventions. As a community and work site intervention, Colorado on the Move has strongly advocated for environmental changes that facilitate activity. With the programs increasing popularity both locally and nationally, the leaders of Colorado on the Move are a powerful voice on policy issues. For example, the group is working with the Stapleton Redevelopment project to design a walking and activity friendly community. This is in contrast to typical suburban developments, which tend to accommodate automobiles rather than walkers. Colorado on the Move is also working to increase the exposure of walking opportunities such as trails, walking routes and organized walks.
Colorado on the Move is attempting to impact social interactions by creating a stepping culture in the state. Some of the ways they are working to do this are by increasing media exposure and creating partnerships with community organizations. Colorado on the Move has successfully increased the publics
attention and motivation for walking and this is should continue diffusing through the population.
Promoting Physical Activity in Primary Care
From an ecological perspective, environmental engineering and community based interventions offer a great potential for increasing physical activity because of the large number of people they affect. However, clinical interventions in primary care settings can be an important complement to policy and community interventions.
Primary care settings provide opportunities for individualized counseling not often possible in community programs. Since adults typically visit physician offices several times a year, providers have multiple contacts with patients and can tailor messages according to each patients stage of change (Whitlock, 2002). Studies show that the majority of patients expect to receive counseling about behavior change from providers and that counseling increases patient satisfaction. Patients who receive counseling are also more likely to initiate behavior change (Pescatello, 2001). Even when change is not initiated, patients who receive counseling may be primed so that they are more receptive to the message at a later time (Whitlock, 2002).
Although evidence supports the role of clinical interventions to promote physical activity, rates of counseling about physical activity are low. Nationally,
rates of counseling have been estimated to be between 26% and 34% (Ma, 2004, Wee, 1999). Patients who were most likely to receive physical activity counseling were those younger than 75 years old and those with risk factors for cardiovascular disease (Ma, 2004). Data collected within a Colorado practice based research network, CaReNet, revealed that in 2,971 clinic visits with adult patients, physicians provided exercise counseling to only 20.3% of patients. Among patients diagnosed as obese, rates of exercise counseling were somewhat higher37% (unpublished data).
When questioned, providers usually identify lack of time, poor counseling skills, and little or no reimbursement as the main barriers to behavior change counseling (Pinto, 1998; Ma, 2004). At a more fundamental level, physicians are also concerned with the issue of effectiveness. Whitlock, Orleans, Pender and Allen (2002) note:
....although physicians increasingly agree that most health promoting behaviors are important to patients health, they report skepticism about patients willingness to change behaviors and their own ability to intervene successfully in these areas.
Questions about effectiveness are also raised in the US Preventive
Services Task Force (USPSTF) updated guidelines for physical activity
counseling (Eden, 2002). Although the guidelines recommend physician
counseling for physical activity, the strength of evidence to support this
recommendation is rated C (inconclusive evidence for or against).
Improving Behavioral Interventions in Primary Care
In order to provide a stronger evidence base on which to inform future guidelines, the USPSTF convened The Counseling and Behavioral Interventions Work Group. In a recent publication, this group proposed two analytic frameworks to guide research and evaluation of behavioral counseling in clinical practice (Whitlock, 2002). Analytic framework #1 asks the question: Does changing individual health behavior improve health outcomes? Analytic framework #2 asks: Can interventions in the clinical setting influence people to change their behavior? While there is strong evidence that increasing physical activity is associated with improved health, there is less evidence to guide implementation of successful interventions in primary care.
Analytic Framework #2 focuses on design and evaluation criteria for clinical interventions. Embedded in this model are 13 key questions concerning both short term behavior change and long term maintenance of behavior change. The two questions that are most relevant to this research are:
Q6: what are the essential elements of efficacious interventions (what, how, when, where, to whom, by whom, for how often and for how long?).
Q9: what type of ongoing assistance or support is needed to achieve or maintain targeted behavior changes? (Whitlock, 2002).
The analytic frameworks suggested by the USPSTF are designed to guide research efforts in order to build an evidence base for clinical interventions, but they do not explain how to go about doing this research. Traditional research methods have emphasized efficacy trials where interventions are tested under controlled conditions. According to Glasgow, Lichtenstein and Marcus (2003), these trials do not address many of the practical issues that are essential for actual implementation of interventions in clinical practice. They distinguish between efficacy and effectiveness trials and propose a five-phase model for intervention research. Early stages are devoted to developing intervention hypotheses (Phase I) and methodologies (Phase II), and testing these for efficacy (Phase III). Building on this research, Phase IV trials test the interventions in real life settings and among populations where results can be generalized to the intended audience. Finally, Phase V consists of large-scale demonstration projects.
The model proposed to guide effectiveness evaluation is the RE-AIM framework (Glasgow, 2001; 2003). The RE-AIM framework takes into consideration factors that operate at the level of both the individual and the setting. The five evaluation dimensions that compose RE-AIM are Reach, Efficacy, Adoption, Implementation, and Maintenance. These terms are described below:
Reach refers to the extent to which eligible participants take part in the intervention and how representative they are.
Efficacy or Effectiveness measures the impact (both positive and negative) of the intervention on participants who began the program.
Adoption considers the percentage of potential settings that will participate in the intervention and their representative ness.
Implementation tests the extent to which the intervention was delivered as intended.
Maintenance is evaluated on both the individual and setting level. At the individual level, long-term (6-12 months) effects are considered. For settings, maintenance is defined in terms of the extent to which the setting continues to provide the intervention and whether or not it becomes institutionalized.
The RE-AIM framework addresses both external (reach and adoption) and internal validity (efficacy and implementation) as well as individual and setting level outcomes. This information obtained from application of RE-AIM should inform translation of efficacy research into actual clinical settings. Many of these issues have been previously ignored. While the model provides a useful framework for conducting evaluation, its breath may be beyond the scope of smaller research projects such as the one discussed in this paper.
Similar concerns about the transferability of research conducted under
conditions that maximize internal validity that were issues raised by Glasgow are
also under discussion among behavioral science theorists. At the forefront of
these dialogues is Alexander Rothman (2004) who argues that theoretical models
that have been rigorously tested in controlled settings need further development
through application and evaluation as interventional strategies in real life
settings. Rothman states that, If investigators are more receptive to the
opportunities interventions afford for theory testing, there will be a dramatic
increase in data that can reveal the adequacies and inadequacies of a given
theory. He postulates that there is a gap between knowledge produced by
theorists and the knowledge that comes from testing theory in real life situations:
From the perspective of an interventionist, the accuracy of the relations specified in a theory is an important but not sufficient determinant of its value. Interventions need theories that are accurate and applicable; that specify not only the relation between two constructs, but also whether that relation does or does not change across contexts (e.g., does the impact of risk perceptions on behavior differ whether one is examining decisions to test of radon or to start smoking?
One of the opportunities afforded by intervention testing is the ability to identify moderators and mediators of behavior change.
Another theoretical concern that is pertinent to physical activity research
is the extent to which behavior change theories are applicable across different
types of behaviors. Increasingly physical activity researchers have recognized
that physical activity behavior change requires multilevel and multidimensional
models. Most behavior change theories use a fairly linear model and focus at the
level of the individual behavior change. These theories may not be
comprehensive enough for more complex behaviors such as physical activity or
nutrition. Recognizing the deficiency in current theories, one idea is to work
toward better integration of different theoretical approaches (Epstein, 1998; King
2002). Others believe that the inability of traditional theories to guide significant
long-term change in physical activity behavior is evidence that what is really
needed is a complete paradigm shift (Morgan, 2001).
Confusion over the application of theory to physical activity behavior
change is complicated by the fact that:
There are hundreds of behavioral studies on physical activity, with great diversity in research designs, measurement approaches, populations studied, theories used, variables tested, and physical activity outcomes. This diversity makes it difficult to integrate the findings and summarize the status of the field, thus limiting the ability of subsequent research to build on previous findings.
These issues are particularly relevant to this research study since it represents a fundamentally different approach to physical activity promotion compared with the majority of previous studies. The theoretical models that have
been evaluated in studies that promote exercise may not be applicable to those
that promote lifestyle physical activity. In fact one of the gaps in the physical
activity literature is in evaluation of the effect of type and intensity of activity on
adoption and maintenance (Dishman, 2001).
One of the ways that theoretical models have been tested is by evaluating
the extent to which their constructs are associated with physical activity.
Dishman, Sallis and Orenstein (1985) performed a meta analysis of over 300
physical activity studies, to determine the correlates of physical activity.
Commenting on this research Bauman (2002) notes:
... .the most remarkable finding was the large number of variables that were not associated with a specific theory but were still found to be consistently related to physical activity. No current theory, or even a combination of theories, accounted for 15 variables that were associated with physical activity.
These findings suggest both the inadequacy of current theoretical models of behavior change for the study of physical activity as well as the complexity of interacting factors that influence physical activity. Authors of this study conclude that physical activity involves a complex causal web of intrapersonal, interpersonal, social/cultural, and physical environmental correlates. Such an interpretation suggests the need for a multilevel model (Bauman, 2002).
Social Cognitive Theory
While a multilevel model is necessary for understanding the complexity of interacting variables that influence physical activity, the intervention used in this study is more narrowly focused at the level of the individual. Social Cognitive Theory (SCT) is used to guide this research because it appreciates the complexity of factors affecting behavior while supporting an emphasis on the level of the individual. Social Cognitive Theory explains human behavior as the result of triadic, reciprocal determinism between behavior, personal factors and environment. Within each domain specific constructs are identified that are associated with behavior change.
Within the personal factor domain, the main constructs are: knowledge, skills, self-efficacy, outcome expectations and personal goals. Behavioral constructs include: frequency, consistency and other aspects of behavior that are specifically relevant to particular circumstances. Environmental constructs include social, institutional and physical factors.
Studies evaluating physical activity determinants have found strong support for constructs from social cognitive theory, especially self-efficacy and social support (Dishman, 1985). Lewis, Marcus, Pate, and Dunn (2002) also found good support for SCT constructs in a review examining theory based physical activity interventions. This review looked at studies that measured change in constructs that were hypothesized to mediate physical activity.
Unfortunately, although many studies use a theoretical model few specifically measure the extent to which interventions change mediators. The following paragraphs outline the various constructs that make up SCT and describe how they apply to this study.
Personal Factors. The first group of constructs are those categorized as personal factors in the model.
Knowledge. Correlational studies have not found a relationship between knowledge about the health outcomes associated with exercise and behavior change to increase physical activity (Trost, 2002). This may be interpreted to mean that although people know that they should exercise they dont. If this is true, then we could hypothesize that knowledge about the health benefits of exercise is not sufficient to overcome the barriers associated with exercise. However, the outcome might change if the message substitutes the concept lifestyle physical activity for exercise.
When applied to studies of physical activity, knowledge is typically defined as extrinsic knowledge about the health risks associated with lack of exercise and the health benefits that accrue for increasing vigorous activity. Information, such as the Surgeon Generals recommendations, is widely publicized in the media. As with many public health messages, most Americans, even those who are sedentary, are at least somewhat aware of the health risks and benefits associated with physical activity. However, because of the
predominance of messages about moderate-to-vigorous activity, people who are sedentary may not be aware of the fact that even small changes in activity at low-to-moderate intensity can provide health benefits.
Although knowledge has not been correlated with physical activity, it may have more to do with the educational message rather than the construct. In this intervention, education about lifestyle physical activity may help motivate sedentary people because it promotes a level of activity that is more easily obtained.
In this study, knowledge is conveyed in a variety of ways. The first opportunity to provide education occurs between patient and provider. Flyers for the study instruct patients to ask their providers about the study. Also, providers can use the study as an opportunity to discuss activity with patients. In both instances, provides can individualize the message to patients, which increases the credibility of the message. The second place where education is provided is through watching a nine-minute video produced for healthcare settings by Colorado on the Move. The video is both motivational and instructional. It provides information about how to use the step counter (e.g. how to position it, and re-set it), and ways to increase daily activity (using stairs instead of elevators, parking farther away from stores, etc.). This information is supplemented by a small booklet that participants receive. The booklet contains
information about increasing activity with the step counter and has a journal in the back for recording daily steps.
The extrinsic knowledge discussed above is complemented by individualized knowledge that enables participants to evaluate and monitor their own activity levels. Having objective information that is both personal and readily available may empower participants to make sustainable changes in physical activity behavior.
Skills. The skills required for this study can be categorized in terms of skills related to 1) performing the activity, and 2) self-regulatory skills such as problem solving, and self-monitoring. One of the advantages of this study is that the type of activity that it promotes, namely walking requires very little skill. However, to be successful with the intervention (e.g., to increase activity) participants need to develop self-regulatory skills. The step counter is a tool for self-monitoring because it provides real time feedback about activity. Selfmonitoring throughout the day may lead to increased motivation to make small changes in activity. However, success with behavior change also requires development of problem-solving skills. These skills may involve planning time to take a walk and/or developing the habit of using the stairs instead of the elevator.
Self-efficacy. Self-efficacy is one of the most consistent correlates of physical activity behavior (Trost, 2002; Dishman, 1985). Self-efficacy for
physical activity refers to ones confidence regarding participating in specific types of physical activity or specific amounts of physical activity, or both. The concept of self-efficacy is especially useful for understanding the initial appeal of this intervention to the target audience. Most people have self-efficacy for walking and lifestyle physical activity. Efficacy is enhanced by the fact that people are encouraged to set their own goals. Participants are taught that setting realistic goals is more important then setting overly ambitious goals. By setting and achieving realistic goals, participants can increase self-efficacy.
Outcome Expectations. One of the theoretical constructs used to study determinants of physical activity is outcome expectancy theory. Outcome expectations are the individuals estimate of the likelihood that performing a behavior will result in a particular outcome. Researchers have found that there is a predictive association between participation in physical activity and perceived outcomes. The most commonly reported expected outcome for engaging in physical activity is health benefits. Other reported outcomes are decreased stress, weight control and appearance (Steinhardt, 1989). In a review of mediators of physical activity, Lewis et al. (2004) report support for the construct from one study that specifically measured changes in outcome expectations
Outcome Expectancies. Outcome expectancies are the values that a person places on a particular outcome. According to Baranowoski, Perry, and
Parcel (2002) expectancies influence behavior according to the hedonic principle: if all other things are equal, a person will choose to perform an activity that maximizes a positive outcome or minimizes a negative outcome. For physical activity it is important to realize that short-term outcomes, such as how you feel after activity can have an effect on behavior apart from long-term outcomes such as risk reduction. Sallis and Owen (1999) note that: The punishment of vigorous exertion remains immediate and salient, while the reinforces of improved health or weight loss are greatly delayed and silent. The importance of short term expectancies is also demonstrated is by the fact that enjoyment is consistently found to be a strong correlate of physical activity.
Personal Goals. Personal goal setting is an important aspect of this intervention. Although participants are not required to state their goal(s), they receive instruction and encouragement for goal setting. The first week of the study is devoted to collecting baseline data so that participants can determine their average weekly step amounts. Using their baseline levels participants are taught to set realistic goals for themselves. Although many people have heard that the recommended number of steps per day is 10,000 this study downplays this recommendation so that people do not become discouraged. For example, people whose baseline activity is below 3,000-4,000 steps per day are encouraged to try to increase by 500 to 1,000 steps per day. The goal of 10,000 steps is probably only appropriate for people who are already fairly active (8,000
steps). The qualitative part of the study will ask people whether or not they met their person goals (short term and long term).
Environmental Factors. In SCT, environmental factors are factors that are external to the individual and can be classified as micro and macro factors. Macro level factors include the availability of enablers of physical activity such as walking and bike paths, as well as cues to action. Micro factors include social support and social learning. As discussed previously, most people in the Denver area have some awareness of Colorado on the Move and many people know friends who have participated in this program. This provides a social learning experience. Another aspect of the intervention that was hoped to encourage social learning is that the staff at the clinic were given step counters and many providers purchased them prior to launching the study. It was hoped that staff and provider experience with step counters would increase enthusiasm for the study. In fact, there are a number of providers who regularly wear a step counter and this models the behavior for patients.
Behavior. In SCT, behavior is reciprocally determined by both environmental and personal factors. Although the components of this study target personal factors, behavior is also likely to be influenced by environmental factors because of the impact of Colorado on the Move. In the study the behavior of interest is physical activity and it will be measured as steps taken per day over a 13-week period. Unfortunately step counters do not measure all types
of activity such as biking or swimming. However it is unlikely that the target population for this intervention will be active in any of these other forms of activity since they are primarily sedentary. Nevertheless, they will have an opportunity to describe other activities that they are (or become) involved in during the qualitative interviews. Step counts will be used to determine whether participants increased their walking activity over time. One of the goals of the intervention is to encourage participants to at least make small lifestyle changes such as taking the stairs instead of elevator as a means of increasing their activity. During interviews participants will have the opportunity to describe the specific behavior changes they made, if any.
Linking SCT Theory with Intervention Components. Each of the preceding paragraphs discusses a construct from SCT and shows how it applies to this research intervention. In Table 2.1 this information is reorganized to show the links between components of the intervention, their relations to theoretical constructs and the potential mechanism by which they might influence behavior change.
Linking SCT Constructs with Components of the Intervention.
SCT Behavior Change Mediator Intervention Objective Intervention Components
Knowledge Teach lifestyle physical activity (LPA) Health benefits of LPA How to increase LPA Teach correct techniques for step counter use Video, booklet, provider-patient communication, study coordinator instruction
Goal setting Collect baseline activity level. Set realistic goals. Step counter, video, booklet
Self-regulatory skills: Self-monitoring Problem solving Develop self-regulatory skills to support behavior change. Use step counter to self-monitor activity. Keep daily activity records. Learn problem-solving strategies for LPA. Video, step count log Step counter Video, booklet, activity journal, weekly reporting
Self-efficacy Setting and accomplishing realistic goals increases self-efficacy. Video, booklet, activity journal
Outcome expectations Develop realistic expectancy of the outcome that will be achieved. Video, booklet
Outcome expectancies People will find that LPA is enjoyable and achievable. Video, booklet
Social variables Behavior modeling Physician and clinic staff model wearing step counters.
The Transtheoretical Model
One model of behavior change that has been applied to studies on physical activity is the transtheoretical model (TTM) (Prochaska, 2002). The TTM is a stage-based model that postulates that people move through specific stages prior to actually making a behavior change. One of the attractive features of this model is that it represents behavior change as a dynamic process. The theory discusses the various cognitive processes that apply to each stage of change and provides useful constructs for designing interventions at various stages. In the literature, this model is often combined with SCT for physical activity studies. Although the model is popular and has received good support in studies looking at behavior change for exercise, there is evidence suggesting that it may not be a useful model for studying lifestyle physical activity. In an evaluation of construct validity by Schumann et al. (2002), stages of change was found to only be valid for strenuous and moderate intensities of activitiesnot for moderate to low level activities (except in elderly).
Although this finding argues against the application of TTM to low-moderate levels of physical activity, the underlying logic of the model is valid. However, the criteria used to define the various stages need to be flexible and reflect goals that are realistic for the target population. In most instances, individuals are categorized as being in an active stage only when they meet specific criteria. These criteria are typically defined by professionals based on
health outcome data. In physical activity studies the criteria usually reflect the Surgeon Generals recommendations. Anyone who meets these criteria is considered to be in an action phase, while those who do not meet the criteria are judged to be contemplative or precontemplative. However, the Surgeon Generals recommendations may not be a realistic immediate goal for someone who has been sedentary for his or her entire life. For this person, it might be more appropriate to define their stage according to what is realistic for them to accomplish. Although they may not be in the Action Phase with regard to meeting the Surgeon Generals recommendations, they may be in the Action Phase with regard to taking a 10-minute walk each day. The utility of the TTM could be enhanced by taking a less rigid, more individualized approach to defining the behaviors associated with the various stages of the model. Such an approach would be consistent with evidence showing the value of individualized goal setting. Furthermore, physical activity requires a complex assortment of behavior changes and looking at only a single outcome can obscure the fact that people have different behavior change trajectories. While there may be argument about how to apply this theory, the general concepts and principles underlying TTM have pervaded the field of health promotion. The author recognizes that TTM concepts have influenced the development of this intervention on one level even though the primary model on which the intervention is based is SCT.
Maintenance of Physical Activity Behavior Change
While there are many theories that deal with the initiation (adoption)
phase of behavior change, there are few theories that attempt to cover the
maintenance (adherence) phase. Although maintenance of behavior change is
necessary to achieve health improvements, relapse remains the norm, regardless
of the behavior in question (Orleans, 2000). Many studies of exercise
adherence, especially early ones, typically had a disappointing drop out rate of
50% (Marcus, 2000). In more recent studies using different criteria to measure
maintenance and with different types of interventions, Marcus reports that results
are consistent with earlier conclusions that adherence decays over times. One of
the reasons for poor maintenance of behavior change is that most interventions
focus almost exclusively on strategies to promote adoption. These interventions
assume that short-term change will result in long-term change, but there is no
evidence to support this assumption (Dishman, 1996).
Consistent with the focus of intervention studies on initiation of change, there has been little attention paid to developing theoretical models of maintenance. According to Orleans (2000), most past research has tended to treat maintenance as a fixed or static state. As a result, the concept of maintenance has not been well studied and many today feel that the theoretical models that we have are not adequate to guide research in this area (Bauman, 2002; Baranowski, 1998).
In a supplemental issue of Health Psychology devoted to the topic of
maintenance of behavior change in cardiorespiratory risk reduction, authors
consistently express the need for better theoretical models to guide research in
the study of long-term behavior change. Rothman (2000) states that:
Given that the repeated application of intervention strategies that facilitate short-term success does little to improve rates of longterm success, the premise that there are important differences in the psychological processes that govern behavioral initiation and maintenance appears worthy of consideration.
These authors unanimously agree that the theoretical models which have aided
researchers in designing effective programs for initiating change, are insufficient
for providing theoretical knowledge on which to base maintenance strategies
(Orleans, 2000; Rothman, 2000; Marcus, 2000).
Even the Transtheoretical model (TTM), one of the few that explicitly
identifies the stage of maintenance, takes a narrow view of maintenance
conceptualizing it only in terms of relapse prevention. By focusing exclusively
on preventing relapse, the model assumes that maintenance is a final destination
rather than a stage on a longer journey. Applying the metaphor of a journey
suggests that maintenance is itself a process that requires the ability to persevere
with behaviors despite changing attitudes, and circumstances (Orleans, 2000).
Viewed in this way, it is apparent that relapse prevention is an insufficient
construct for studying and intervening in the maintenance phase of behavior
change. Since the concept of maintenance has been understudied, this research
uses an exploratory approach to understand the factors that influence long-term physical activity behavior change.
RESEARCH DESIGN AND METHODS
This study uses mixed methods to evaluate the effectiveness of an intervention designed to increase lifestyle physical activity among primary care patients. The primary quantitative outcome is change in average daily step count. Interviews are used to assess individual experiences both immediately following the intervention and again 6 months later.
1. Compare and contrast characteristics of completers and non-completers of the 13-week intervention.
2. Determine if activity levels change during the 13-week intervention period and whether changes are maintained.
3. Examine the accuracy of self-perceptions of physical activity and evaluate the extent to which the intervention changes perceptions.
4. Explore the subjective experiences of participants during the intervention and follow-up phase of the study.
5. Propose intervention strategies for using step counters in a primary care practice.
Mixed Methods Paradigms Conundrum
Mixed methods research designs present unique challenges for discussing paradigms because they combine research methods that are traditionally associated with divergent epistemologies. The theoretical and paradigmatic difficulties implied by mixed methods have lead many mixed methods researchers to avoid the issue of identifying a particular paradigm of inquiry. In reviewing mixed methods research in the social science, Greene and Caracelli (2003) note:
It appears that paradigms are not the primary organizing framework for mixed methods practice. Rather applied social inquirers appear to ground inquiry decisions primarily in the nature of the phenomena being investigated and the contexts in which studies are being conducted.
These authors caution that despite the apparent discrepancy between theory and practice, paradigms are important and practitioners must be reflexive as well as cognizant of the underlying epistemologies associated with the methods they chose.
In their book, Handbook of Mixed Methods in Social and Behavioral Research, Tashakkori and Teddlie (2003) identify six perspectives on the issue of how paradigms should be used in the development of mixed methods research. The a-paradigmatic viewpoint argues that methods and paradigms are independent consequently there is no inherent epistemological conflict. A
second perspective is that the paradigms associated with quantitative (QUAN)
and qualitative (QUAL) designs are incompatible and therefore mixed methods
designs are impossible. The third perspective, complementary strengths states
that the two paradigms have different strengths, which can only be realized if the
methods are kept separate within the study design. The fourth position argues
that there should be a single universal paradigm that unifies mixed methods
research. The fifth view, named dialectic holds that mixed methods research
designs intentionally utilize multiple paradigms, along with their underlying
assumptions, for the specific purpose of examine (ing) the tensions that emerge
from the juxtaposition of these multiple diverse perspectives (Teddlie, 2003).
The final perspective is similar to the dialectic position except that it prescribes
the paradigms that must be used for different types of inquiry.
Although the supporting arguments for each viewpoint have merit, the
position that mostly closely reflects the purpose for choosing mixed methods in
this study is the dialectic perspective. This perspective recognizes that
different paradigms make unique contributions and that combining multiple
paradigms provides greater understanding of the phenomenon under
investigation. According to Teddlie and Tashakkori (2003),
An important component of this position is the ability to think dialectically. This involves consideration of opposing viewpoints and interaction with the tensions caused by their juxtaposition.
The tensions come from the differences in the assumptions of the different paradigms.
QUAN and QUAL methods are associated with epistemologies that have
divergent assumptions. Quantitative inquiry is usually identified with a positivist
or post positivist approach to research. Positivists believe that:
.. .social research should adopt scientific method, that this method is exemplified in the work of modem physicists, and that it consists of rigorous testing of hypotheses by means of data that take the form of quantitative measurements. (Atkinson, 1994)
The positivist viewpoint is that truth can be discovered through
experimentation. Post positivists also uphold scientific method, but realize that
the results are only approximations of truth. In contrast, constructivists focus on
meanings that are constructed by individuals or groups. They do not attempt to
discover a single truth, but instead explore peoples differing perceptions of
reality. Constructivism supports qualitative methods of inquiry.
A fundamental difference between QUAN and QUAL methods concerns
the role of the researcher. In QUAN research, the researcher should have
no/minimal impact on the study, whereas QUAL research acknowledges the
researcher as a participant in the research and applies reflexivity as a tool to
Although theorists argue over the implications of blending epistemologies, there are many pragmatic advantages to combining QUAN and QUAL methods. In this study, a mixed methods design was chosen in order to conduct a broad study of the phenomenon of lifestyle physical activity
promotion. The QUAN data provides an objective assessment of baseline (week 1) activity and measures change in activity over a 13 week time period. While this data can help show whether or not the intervention was associated with increased number of steps, it provides very little insight into the reasons why the intervention was or was not successful. Nor do QUAN methods fully explain which types of people are more and less successful with the intervention. These questions can be more easily addressed with QUAL methods, especially in the early exploratory phases of the research. In addition to providing insight into the intervention, combining methods also allows triangulation of data, which increases the validity of QUAN inquiry.
Sequential Explanatory Design
The study utilizes a sequential explanatory design in which quantitative data (QUAN) is collected in the first phase of the study and qualitative data (QUAL) in the second phase. As defined by Creswell, Clark, Gutman, and Hanson (2003) this design
... .is characterized by the collection and analysis of quantitative data followed by the collection and analysis of qualitative data.
Priority is typically given to the quantitative data, and the two methods are integrated during the interpretation phase of the study.
In this mixed methods study the two strands are independent and each is guided by a unique yet complementary set of research questions. The data from each
part of the study are analyzed separately and integrated to create meta-inferences as illustrated in Figure 3.1.
Figure 3.1. Sequential Explanatory Design.
The Steps for Health study enrolled 83 participants who agree to wear a step counter daily for 13 weeks. Participants recorded their step counts each day and reported these values each week using either an automated telephone system or e-mail. At enrollment, participants viewed a 9 minute video that educated them about how to use the step counter, set personal goals and gave ideas for ways to increase daily physical activity. At the conclusion of the 13 week
intervention, telephone interviews were conducted with participants and follow up interviews were conducted 6 months later.
Human Subjects Review
Human subjects approval was granted from the Colorado Multiple Institutions Review Board/COMIRB (Protocol #03-281). Two amendments to the original protocol were filed and approved to enable the author to conduct the qualitative portions of this study. The research was also approved under HIPPA guidelines, which were implemented by COMIRB shortly after the study was approved. The research was submitted to the University of Colorado at Denver, Human Subjects Committee (protocol #2004-129) and this IRB accepted the approval granted by COMIRB.
In compliance with the study protocol to protect patients and ensure confidentiality, written informed consent and HIPPA authorization were obtained from all participants at the time of enrollment. Signed consent forms were stored in a locked file cabinet. Five digit randomly selected study numbers were used to identify study participants. Quantitative data were stored on a secured server.
Participants and Site
Research Site. AF Williams Family Medicine Center, Denver, Colorado.
Sample Size. The study was approved for enrollment of up to 85 patients.
Sample size was calculated based on data from the Denver Black Church Initiative (unpublished data from Colorado on the Move; J. Hill, P.I.). It was determined that a sample size of 60 would provide 80% power to detect a .6 effect size pre-post difference, assuming 33% dropout and a .2 correlation between baseline and final observations. Since dropout rates for exercise studies tend to be high, we planned for an initial sample of 85 to accommodate a dropout rate of 50% or slightly higher and still retain sufficient power to detect a .6 effect size difference.
Recruitment. Participants were recruited through self, nursing and provider referral. Posters and flyers were displayed throughout the clinic to advertise the study. Additionally, providers were asked to discuss the study with patients. To ensure that the study did not pose a medical risk to participants, each patients primary care provider (PCP) was asked to complete a medical waiver. The enrollment period was anticipated to take approximately two months, however it was necessary to extend it to six months. Participation was voluntary and participants were free to drop out of the study at any time. The interviews at 13 weeks and 9 months were optional and only those participants who consented to be interviewed were contacted.
Inclusion/exclusion Criteria. Participants were required to be over age 18, not pregnant, without medical problems that would be adversely affected by
participation in the study, not already exceeding the Surgeon Generals recommendations for activity, able to understand the study protocol, and willing to fulfill the data reporting requirements. Potential participants completed a brief health survey that included height, weight, age, gender, and self-assessment of current level of activity. Patients with medical conditions that in the opinion of their primary care provider would be adversely affected by participating in this study were excluded. Originally, patients whose physical conditions severely limited their mobility (e.g., morbid obesity, severe osteoarthritis, and severe COPD) were to be excluded however, providers felt that some of these patients could benefit from being in the study and referred them. Although the inclusion of people with activity limitations negatively impacted the quantitative results, these participants made important contributions to the qualitative portion of the study.
Compensation. Study participants received a complementary step counter with a safety leash (valued at $15.00) along with a Colorado on the Move educational booklet (valued at $1.00) when they joined the study. They did not receive any monetary compensation.
Dependent and Independent Variables
Dependent Variables. The dependent variables are average daily step count by week and change in activity over 13 weeks.
Independent Variables. Age, race/ethnicity, sex, BMI, current level of
activity, and average daily step count during week 1 (a proxy for baseline activity).
Phase I: Protocol and Data Collection (Quantitative)
Prior to launching the study, an in-service was done with clinic staff to describe the study, educate them about the importance of physical activity and distribute step counters to each person who wanted one. The purpose of the in-service was to create a step counter culture in the clinic in order to increase enthusiasm for the study and model lifestyle physical activity behaviors.
Patients initiated the enrollment process by either calling the study coordinator, completing a card provided by the nursing staff or giving their names to their provider. Patients who were in the office could complete the health questionnaire at that time. The study coordinator followed up with patients who expressed an interest in the study but did not complete a health questionnaire. The study coordinator either interviewed the patient over the phone using the health questionnaire or mailed the questionnaire for patients to complete and return. This questionnaire was used for preliminary screening regarding activity and health status. For those patients who met inclusion criteria, the patients PCP was asked to review the patients self-reported health questionnaire and sign the medical waiver. After the PCP cleared the patient for
participation, the study coordinator contacted the patient and scheduled a time for them to complete the enrollment process.
At the time of enrollment, patients viewed a brief video, Step Your Way to Better Health explaining the purpose of the study and requirements for participation. The video served as an educational and motivational tool. Patients who choose to participate were given a step counter, a Colorado on the Move booklet and written instructions for using the telephony system. Each step counter was checked for accuracy before it was distributed. The study coordinator also gave verbal instructions regarding how to wear the step counter and emphasized the importance of recording steps daily and reporting them weekly.
Participants were told that during the first week of the study they should not make any behavior changes, but should instead simply assess their baseline activity. Once they knew their baseline, they were encouraged to set realistic, goals for increasing their activity. Both the study coordinator and the educational booklet provided information about goal setting. Although the literature reports that a baseline can be calculated with as few as three days of data, a full week was chosen because it seemed easier to understand and it also would allow for the fact that participants may forget to wear their counters during the first week (Tudor-Locke, 2001).
Quantitative Data Collection. At enrollment, participants completed a
short questionnaire that asked about demographic information and selfperception of current level of activity. Participants were instructed to record their steps daily in a journal provided and to report these amounts on a weekly basis. The two methods for reporting data were a computerized telephone system and e-mail. For participants who used e-mail a weekly reminder was sent. Participants who neglected to report data were contacted by phone or by e-mail.
Phase II: Protocol and Data Collection (Qualitative)
Participant Interviews. Two sets of interviews were conducted, one immediately after completion of the study and the other 6 months later. In order to get good representation, 25-30 initial participant interviews were targeted. Interviews were conducted over the phone and expected to take about 15 minutes. Since the QUAL portion of the study was added after the QUAN intervention began, an updated consent form asking about willingness to be contacted for interviews was mailed to participants who were already involved in the intervention phase of the study (about 40 people). Participants were instructed to check the appropriate box indicating their willingness to be interviewed, sign the form and them return it in the stamped envelope provided. Those who enrolled later completed updated consent forms.
The main purpose of the interviews was to gather information about the intervention. All participants who consented for interviews and who completed the intervention were contacted. Participants who dropped out of the study typically were not interviewed although they were contacted to discuss their reason(s) for dropping out. There were however, three women who dropped out of the study due to time constraints but intended to continue trying to increase their activity with their step counters. These women were purposefully chosen for interviews because their experiences represented a naturalistic application of the intervention. They were each willing to participate in the interviews.
Data Collection, Analysis and Integration In this sequential explanatory design study, QUAN data consisting of a short questionnaire and weekly step reports were collected first. These data were analyzed after all participants completed the 13-week intervention period (the intervention period is considered the 13 weeks during which participants agree to wear the counters and record their steps). Interviews were conducted when each person completed the intervention phase of the study. Consequently, some people were in the intervention phase while others were in the follow up (interview) phase of the study. Thus, data from the QUAN strand of the study could not be used to generate questions for the initial QUAL portion of the study.
The findings from each portion of the study were combined in the following manner. Data from both QUAN and QUAL strands of the study are triangulated to address the first three specific aims. The remaining specific aims of the study are addressed using the data from QUAL interviews alone. The following paragraphs describe the methods for QUAN data collection and analysis followed by a similar discussion of the QUAL methods.
The first three study aims are listed below along with the QUAN research question(s) that address the aim. Following this, the specific analytic methods used to address each question are identified. Statistical analysis was conducted using SPSS Version 12.0 for Windows.
When this study was designed, a power analysis was conducted based on data from Colorado on the Move. The drop out rate was estimated at 25%; however, the actual rate of non-reporting and non-completion was much higher (60%). Although this rate is higher than anticipated, it is not unusual for similar studies (Wyatt, 2004). Nevertheless, the low completion rate limits the types of analysis that can be done.
Study Aim 1. Compare and contrast characteristics of completers and noncompleters of the 13-week intervention.
QUAN Question 1. Are there significant differences between study completers, non-completers and non-reporters with regard to BMI, gender, age, ethnicity, self-reported activity level and baseline activity level?
QUAN Question 2. Are demographic variables associated with significant differences in baseline steps for completers and non-completers?
Descriptive analysis was used to identify differences between study completers, non-completers and non-reporters with regard to demographic variables and baseline steps. The following statistical tests were used to determine if the differences between groups were significant:
One-way ANOVA and Student Newman-Keuls post-hoc tests at a = .05 were used to test differences between groups for BMI, age, and baseline activity.
Independent sample t-tests (two tailed) with a = .05 were used to test differences in baseline steps between completers and non-completers in QUAN question 1.
Independent sample t-tests (two tailed) with a = .05 with the files split were used to test for significance differences in baseline steps for QUAN question 2.
Chi-squared tests with a = .05 were used to evaluate differences in gender, ethnicity (Caucasian or non-Caucasian) and self-reported activity.
Study Aim 2. Determine if activity levels change during the 13-week intervention period and whether changes are maintained.
QUAN Question 3. Descriptive analysis of daily step averages per week with confidence intervals for completers and non-completers.
QUAN Question 4. For study completers, is there a significant mean change in step counts during the 13-week study?
QUAN Question 5. For study completers, what is the percent change in step counts during the 13-week study?
The following methods were used to address the study questions:
Average steps per week with percentiles (2.5, 97.5) for completers and noncompleters are presented in a table. This information is also displayed graphically.
One of the primary QUAN aims of this study is to determine whether participants increase their activity over the 13 week period. Two different methods were used to make this assessment.
The first uses a paired sample t-test (two-tailed) with a =.05 to compare average daily steps at baseline (week 1) with average daily steps at week 13.
An alternative approach to calculating this change compares the mean daily step count in week 1 with the average of the values during the last four weeks of the study. This method was used in the article that reported on two community interventions conducted by Colorado on the Move however
the reasons for selecting this method were not explained (Wyatt, 2004). It is used here to allow comparison with their data.
Percent change in average daily step count was calculated for each person who completed the study using the formula, (average steps in week 13 minus average steps in week 1, divided by average steps in week 1) x 100%. This information is displayed in a frequency table.
Study Aim 3. Examine the accuracy of self-perceptions of physical activity and evaluate the extent to which the intervention changes perceptions.
QUAN Question 6. To what extent do participants self-reported activity levels (sedentary, moderate or active) correlate with their baseline step counts?
Spearmans rank-correlation coefficient (rs) with a two-tailed test for significance at a = .05
The main objective of this QUAL portion of the research is exploratory. Two sets of semi-structured interviews were conducted with approximately 25 participants to assess their experiences with the 13 week intervention as well as explore the long-term impacts of the intervention on behavior change. The following research questions helped to guide the QUAL portion of this study:
1. In what ways did the intervention impact attitudes toward physical activity?
2. What types of behavior change, if any were associated with the intervention?
3. Were behavior changes maintained 6 months after the intervention period? Why or why not?
4. What strategies did people use to increase activity while in the study and did these change over time?
5. Identify the ways in which social and environmental factors influence physical activity behaviors.
6. What aspects of the intervention were most helpful for participants?
7. Did participants set activity goals and what was this experience like?
Data Collection. Two data collection methods were considered, interviews
and focus groups. The decision to use individual interviews was based on the following considerations: 1) the unit of analysis would be the individual rather than a group, 2) interviews can provide more depth about a topic, 3) logistics, and 4) finances. A semi-structured format was selected to allow specific areas to be targeted as well as probing of other topics that might emerge. The question guides used for the interviews are found in Appendix A. Since the qualitative portion of the study was not funded interviewees were not reimbursed for their time. In consideration of this, the interviews were designed to be relatively short
(approximately 10-15 minutes) and convenientthey were conducted by telephone. Although the interview questions could be easily answered in 10 minutes, most interviews took longer because people wanted to discuss their views and go into depth talking about their experiences. Handwritten notes were used to document the interviews. This method was chosen because the interviewer was comfortable taking notes and this seemed less intrusive compared with tape recording. The information that was collected was not particularly sensitive and the benefit of getting complete transcripts of interviews was less important compared with the potential reluctance of participants to talk freely because they were being tape recorded over the telephone.
The first set of interviews occurred after participants completed the 13 week intervention and the second 6 months later. Questions for the first set of interviews are more structured compared with the second set in order to allow more individualized follow up and discussion of themes that emerged during the first interview. Interviews were documented using hand written notes and typed immediately after each interview. During the second interview, the interviewer read a summary of the previous interview to the participant and asked if the account was accurate. In this way, member checking was used to verify the accuracy of the interviewers documentation of the previous interview and to provide a foundation for discussing changes in activity patterns.
During the entire study the author kept a study journal in which she recorded events that occurred, thoughts and perceptions, interactions with participants. This provided a place to keep track of thoughts and impressions.
Sampling. Participating in interviews was an optional part of the study; nevertheless only one person opted not to participate. Since the main purpose of the interviews was to learn about participants experiences with the intervention, people who did not report any data and most of those who dropped out were not considered for interviews. However, these people were contacted in order to ascertain their reason(s) for dropping out. Three women who did not complete the study were exceptions. These women dropped out of the study stating that they did not have enough time, but each intended to maintain the principles of the study. These women were purposefully chosen because they represented a naturalistic application of the intervention. All participants who completed the study (n = 33) were considered interview candidates and attempts were made to contact each. People who participated in the first set of interviews (n = 29) were contacted for the second set. No one who completed the study was intentionally excluded from interviews.
Interview transcripts were analyzed with ATLASti. 4.1 for Windows 95. The data organization and analysis method that best describes the methods used is
immersion/crystallization (I/C). According to Borkan (1999), immersion/ crystal lization
....consists of cycles whereby the analyst immerse him-or herself
into and experiences the text, emerging after concerned reflection
with intuitive crystallizations, until reportable interpretations are
He identifies the elements of I/C as:
Initial engagement with the topic/reflexivity
Crystallization during data collection
Immersion and illumination of emergent insight from collected data and texts
Explication and creative synthesis
Corroboration/legitimation and consideration of alternative
Representing and account/report
One of the main characteristics of I/C is recognizing the researchers
initial engagement/understanding of the topic and addressing this issue through reflexivity. In addition to acknowledging these biases to the reader it is important for the researcher to be in a continual reflexive dialogue.
The process whereby analysis occurs involves becoming immersed in the data and seeing it from multiple perspectives over time. This can begin during the data collection process when particular patterns are identified. In this study, these thoughts, interpretations and questions about the data were recorded as notes following each he interview. Later when the transcripts were entered into ATLASti. these notes were entered as memos. After the first set of interviews
was collected, the immersion process began. The author read and reread the transcripts and began looking for themes using an inductive approach.
While I/C describes the primary qualitative approach for this research, a template organizing style was used to develop codes. One of the advantages of the template style is that it allows the researcher to focus on particular aspects of the text early in the process. The main difference between using a template approach and an editing approach is that, When using a template, the researcher defines a template or codes and applies these to the data before proceeding to the connecting and corroborating/legitimating phases of the analysis process (Crabtree and Miller, 1999). This was appropriate for this research because of my familiarity with the subject matter. It was also consistent with the semi-structured nature of the interview questions.
Both a priori identification of categories as well as reviewing the text were used to develop a multilevel code list. This code list is found in Appendix B. Applying axial coding methods, I developed a coding frame from which to compare and contrast the data looking for relationships and patterns. Within the context of immersion, this allowed me to discover themes and connections between concepts.
One of the techniques used to look for relationships between concepts involved constructing case ordered matrices. According to Miles and Huberman (1994),
A case-ordered descriptive matrix is usually a fundamental next step in understanding whats going on across cases. .. .The matrix forces you to be grounded in the case data and does not allow escape into vagueness or premature abstraction.
Examining the cross tabulation of variables at the individual level was a
powerful tool for exploring relationships. This method was also used to compare
and contrast data from both the qualitative and quantitative strands of the study.
Qualitative research is subject to a number of challenges to validity. The primary concern for the qualitative portion of this study is that the researcher worked alone both in collecting and analyzing the data. Some of the techniques that were employed to reduce bias were peer review, triangulation, field journal, member checking and reflexivity. Peer review was used in constructing the interview questions, developing the codebook and during data analysis. Multiple data sources and different techniques for comparing them allowed for triangulation. I kept a journal during the course of the study and recorded issues that came up as well as thoughts and impressions. Member checking was used with participants to ensure that I had accurately captured the essence of the first interview. Finally, and most importantly was my own reflexivity during the entire process.
Integration of Data
Analysis of QUAN and QUAL data was done separately as described above. QUAL analysis is used to explain the QUAN findings as well as expand these findings by giving voice to the diverse experiences of participants. The results for each portion of the study are presented separately, although QUAN findings are integrated into the QUAL themes as appropriate. The study was not specifically designed to implement a RE-AIM evaluation and the scope of this study does not allow for all RE-AIM questions to be addressed, however a modified version of RE-AIM is used to help summarize the main QUAN and QUAL findings of the study.
Definition of Study Completion Groups
The Steps for Health study enrolled 83 patients at AF Williams Family Medicine Clinic in Denver, Colorado between June 2003 and January 2004. Of this group, 33 people are considered completers because they completed all 13 weeks of the intervention. Twenty-six people are classified as non-completers because they began the study and reported at least one week of data but dropped out prior to the studys completion. Finally, there were 24 people who did not submit any data and they are classified as non-reporters. Within the completer group, three peoples data for the last weeks of the study were lost. Where it is appropriate, sensitivity analysis is used to account for this.
Study completion rates (39.8%) were lower than anticipated, although not significantly different from other studies with similar designs (Wyatt, 2004). The