Abstract

Physical activity has emerged as an important risk factor for many chronic diseases, such as coronary heart disease and diabetes. As part of an effort to improve public health by increasing physical activity levels of the nation, the Centers for Disease Control and Prevention and the American College of Sports Medicine (26), the U.S. Public Health Service (35), the National Institutes of Health(23), and the U.S. Department of Health and Human Service(34) have included activity recommendations. Recognition of the importance of physical activity to the nation's health has also influenced all public health research; most population studies that examine chronic diseases incorporate the assessment of physical activity into their study design. Physical activity was defined by Caspersen et al. (5) as "any bodily movement produced by skeletal muscles that results in energy expenditure." Components of total energy expenditure include basal metabolic rate, which typically encompasses 50%-70% of total energy expended; the thermic effect of food, which accounts for another 7-10%; and physical activity (28,30). This last component, physical activity, is the most variable component and is comprised of activities of daily living (bathing, feeding, and grooming, for example), sports and leisure, and occupational activities. The share of total energy expenditure accounted for by physical activity is obviously greater for active individuals. Measurement of Physical Activity Valid and appropriate measurement of physical activity is a challenging task, because the relative contribution of each of these components can vary considerably both within and among individuals and populations. Measurement is further complicated because there are several health-related dimensions of physical activity, such as caloric expenditure, aerobic intensity, weight bearing, flexibility, and strength (4). Differences in these aspects of physical activity may have implications for the prevention of specific diseases. For example, 100 calories expended swimming may be particularly beneficial to cardiovascular health and the prevention of related diseases, whereas 100 calories expended weight training may have a more favorable effect on bone mass or osteoporosis risk. In examining the relationship between physical activity and a disease or condition, it is therefore important to focus on the dimension (or dimensions) of physical activity most likely to be associated with the specific outcome of interest. The quality of the measure of physical activity is another important concern. Poor quality measures can obscure important associations, as shown by a methodologic critique of coronary heart disease studies conducted prior to 1986. The critique found that 40% of the measures were"unsatisfactory," 40% were "satisfactory," and only 20% were "good" (4,27). As the quality of the physical activity measure increased from unsatisfactory to good, the proportion of studies revealing a statistically significant association increased from approximately 50% to 88% (4,27). The critique noted that the quality of the physical activity measures was poorer than that of the coronary heart disease measures(4,27). The difficulty of ensuring the quality and accuracy of measurements of various health-related dimensions of physical activity will also limit the ability to detect significant associations between physical activity and disease outcomes (17). The Survey Approach versus Other Assessment Tools Physical activity assessment tools have been used to measure various dimensions and attributes of physical activity. Most assessment tools used to measure physical activity have focused on the amount of energy expended(16). The advantages and disadvantages of these different approaches depend upon the population being studied and the research objectives. Epidemiologic studies have typically used subjective measures, such as the questionnaire, to assess physical activity in populations. Such studies then used objective measures to validate the subjective activity measures. Popular objective activity assessment tools include measures of total energy expenditure, such as the doubly labeled water technique and the respiratory chamber; movement counters, which initially measured frequency of movement and have progressively been modified to detect differences in speed and direction of movement; and measures that estimate physical fitness, such as heart rate monitoring and graded exercise testing. The physical activity questionnaire is typically chosen for population studies because it possesses the characteristics of non-reactiveness (it does not alter the behavior of the individual being surveyed), practicality (there are reasonable study cost and participant convenience), applicability (the instrument can be designed to suit the particular population in question), and accuracy (it is both reliable and valid) (17,20). By contrast, objective measurements of energy expenditure, some of which have the advantage of providing more precise estimates of energy expenditure (the respiratory chamber or the doubly labeled water technique) are not practical for most epidemiologic studies, but they have been used to validate the physical activity questionnaire (17,20,29). However, the estimates obtained by the activity questionnaire are valuable in relative terms and can be used to rank individuals or groups of subjects within a population from the least to the most active. The ranking can then be examined with respect to physiologic parameters and disease outcomes(13). Complexity of surveys. The survey approaches used to measure physical activity vary in their complexity, from self-administered, single-item questions to interviewer-administered surveys of lifetime physical activity. Single-item questionnaires may ask individuals whether the person surveyed is more active than others of their age and sex(22) or whether the person exercises long enough to break a sweat (39). Such simple single-item questions have been used to adjust for the confounding influence of physical activity when exploring associations of more primary interest (4). More complex questionnaires attempt to survey a wide range of popular activities over a selected time frame. Time frame of surveys. The activity questionnaire can either ask about usual activity or ask about activity done within the past week, month, year, or even over a lifetime. Participants may be asked to use diaries and logs to record activities over 1 day, 3 days, or the past week. Recall surveys may sometimes query the frequency and duration of activities performed over the past week. Questionnaires focusing on a longer time frame, such as 1 year, are more likely to reflect usual activity patterns and have been used extensively in epidemiologic studies. Surveys with short time frames have two advantages over those with longer time frames: the estimates are less vulnerable to recall bias and more practical to validate with objective tools. However, assessment over a short time period is less likely to reflect "usual" behavior, as activity levels may vary with seasons or as a result of illness or time constraints. To obtain the best estimate of physical activity levels, some questionnaires include assessment over both a short and a long time period. Some studies have attempted to assess lifetime physical activity patterns, because chronic diseases such as osteoporosis and cancer tend to have a long developmental period, and it is potentially the long-term, chronic exposure to physical inactivity that increases risk for disease. Early measures of lifetime or historical physical activity categorized people according to employment history (11,36,37) or participation in high school or inter-collegiate athletics(9,10,18,32,33). More comprehensive approaches attempt to evaluate the extent to which leisure, occupational, or both forms of physical activity were performed during specific age periods (14,15,19). Although limited by recall and validation problems, historical physical activity assessment as part of case-control studies of rare diseases is often preferred over other assessments used in more expensive longitudinal studies. Types of activities surveyed. Early studies in physical activity epidemiology estimated physical activity performed at work(21,24). Such studies typically did no more than inquire about job titles. In contrast, more recent occupational physical activity questionnaires query the frequency, duration, and intensity of activities performed by individual workers on-the-job(14,19). However, since physical activity levels at work have continued to decline in most industrialized countries(27), assessment of leisure-time physical activity is often assumed to be the best representation of physical activity in a population. For this reason, most contemporary physical activity surveys only assess leisure-time activities that require an energy expenditure above that of daily living. A few recently designed questionnaires include both leisure and occupational activity to be used in situations in which the homogeneity of energy expenditure related to both of these components of activity within the study population is not known or cannot be assumed(13,31). A focus on leisure (in a particular sport) and occupational physical activity may be valid for younger and healthier populations. Some have suggested, however, that differences in activities of daily living (bathing and eating) and low-level leisure activities may best represent energy expenditure and physical activity in an older or diseased population(16). Accordingly, questionnaires have been developed that assess these types of physical activities characteristic of populations at the low end of the activity spectrum(7,38,40). Scoring Physical Activity Data A more extensive physical activity questionnaire will measure the type(leisure, occupational, household, etc.), frequency (average number of sessions per given time frame) and duration (average number of minutes per session) of physical activity performed during a particular time frame(typically 1 wk, mo, or yr) and estimate the intensity (degree of vigor or metabolic cost) of the activity. The researcher can analyze the data at several levels (see Fig. 1). The two most common estimates for questionnaire data are derived from summing (1) time spent in physical activity; or (2) time weighted by an estimate of the intensity of that activity. Total time is derived by multiplying frequency (3 times per wk in the figure) by duration (in this case, 2 h per time). A summary estimate of energy expenditure can then be derived by multiplying the average hours per week of reported activity (6 h/wk) by the average intensity, expressed as metabolic cost or METs (5 METs, in this example). One MET represents the metabolic rate of an individual at rest (1,6) and is set at 3.5 ml of oxygen consumed per kilogram body mass per minute, or approximately 1 kcal/kg/h. Thus, these summary estimates of energy expenditure are calculated without consideration of the individual's body weight. Ten METs of activity participation, for example, would require 10 times the resting metabolic rate. Comprehensive lists of the energy requirements for specific physical activities are widely available (2,8,25,41). Each activity is, therefore, expressed in "MET-hours per week" or kcal/kg/wk (30 MET-h/wk or 30 kcal/kg/wk). In turn, this estimate can be converted to kilocalories per week (2100 kcal/wk) if one knows the body weight of the individual (70 kg, in this example). However, this involves making an assumption about the weight of the individual throughout the time frame which is being assessed. The researcher makes several additional assumptions in incorporating intensity into the analysis. First, a MET value in a list is assumed to be representative of the manner in which the individual performed the activity. However, because sporting activities can be performed at a range of skill levels and the pace of any physical activity (especially walking, jogging, and cycling) can vary, the actual energy expenditure across subjects who report the same amounts of time in a particular activity may vary considerably. Second, weighting physical activities by intensity also assumes that body weight is proportional to resting metabolic rate, and that the relative increase in the metabolic cost of a specific activity above resting is constant from person to person, regardless of body weight. Reliability and validity measures. Reliability and validity studies help ensure the accuracy and quality of physical activity assessment. A reliable questionnaire consistently provides the same results under the same circumstances. Reliability studies typically use test-retest reliability coefficients or intraclass correlation coefficients. Validity studies assess how well the questionnaire measures what it was designed to measure. An accurate questionnaire is both reliable and valid. This collection of questionnaires deliberately offered the reliability results first, because without instrument reliability there is no validity. Early studies in the field of physical activity epidemiology often published significant results with the presumption that such results automatically implied that the instrument was reliable and valid. Unfortunately, such studies ran the risk of having found a spurious result and may have been published under the influence of positive publication bias. Today, evidence of instrument reliability and validity is becoming a scientific norm. Reliability and validity are affected by cognitive factors such as a person's ability to store and retrieve information (3). Reliability and validity of the data collected can also be influenced by interviewer or respondent bias, the day of the week being probed, and the sequence of administration of the questionnaire within the battery of other measures collected. Future methodological research examining the reliability and validity of physical activity questionnaires should focus on these issues, as well as on sociodemographic and cultural issues(12,17). Selecting a Questionnaire to Assess Physical Activity Researchers faced with the task of designing a study often find themselves in a quandary. Because physical activity can be defined in several ways, there is no single standard for measuring physical activity. Moreover, time considerations often require researchers to choose a brief survey that measures the most common physical activities of a population. Regardless of the constraints of a particular study, however, the characteristics of the population being studied (such as culture, gender, and age) and the outcome of interest are critical considerations in the choice of a physical activity assessment tool. In any evaluation of the relationship between physical activity and a particular disease outcome, the assessment tool must elicit accurate information on the components of energy expenditure (whether leisure, sporting, or occupational activity) that encompass the greatest proportion of total energy expenditure in the study population. Perhaps because of the historic tendency to conduct epidemiological research on men, physical activity questionnaires have been oriented around the types of leisure and occupational activities typically performed by men. Yet, the physical activity patterns of men and women have traditionally differed, with men engaging in more intense physical activity than women (34). Also, women engage in substantial amounts of child care and household activities, each of which are difficult to assess. Therefore, many of the questionnaires currently in use may be less sensitive to differences in physical activity levels in populations of women. Using This Publication This publication contains the latest versions of most of the popular physical activity questionnaires, along with descriptions of their use. In the section entitled "Primary source of information," the person and/or institution that we contacted for information for each questionnaire is listed. Where possible, this information is followed by tables of the results of reliability and validity studies of the questionnaire, the questionnaire itself, instructions for administration, and details on how to calculate the summary estimates from raw data. Finally, example calculations of summary estimates that use hypothetical data are provided for most of the questionnaires. The tables of reliability and validity studies include a summary of the population in which the questionnaire has been tested. By carefully examining this information, researchers can decide whether the survey is suited to the study population under consideration. The reference list for each questionnaire provides citations for research studies that have utilized the questionnaire either as cited in the text or, as applicable, in "Other studies using the questionnaire." Finally, because all good research should begin with a careful review of the available literature, we feel that the information provided within this collection will expedite this necessary part of the research enterprise. This collection is not inclusive of all questionnaires in the field, nor does it represent all the published information on the questionnaires that are included. We nonetheless believe that this collection will help you find the best ways of assessing physical activity in populations of interest to you.Figure 1-Computation of summary estimates of physical activity.