Individual Physical Activity
MEASURES REGISTRY USER GUIDE
Behavioral Epidemiology Framework
Research advances occur over time through systematically using the scientific method and incorporating new ideas and approaches to answer progressively more complex questions. Medical and public health research is inherently driven by the need to develop and evaluate more effective methods to promote health and well-being in the population. A classic definition of epidemiology (which underlies all public health research) is “the study of the distribution and determinants of health-related states in the population, and the application of this study to the control of health problems.” This definition captures all health-related states but specific “behavioral epidemiology frameworks” have been proposed to facilitate the progressiona of research needed on specific health-related behaviors.7 This Guide adapts a behavioral epidemiology framework developed for physical activity research.8
Description of the Behavioral Epidemiology Framework
The conceptual model in Figure 2 presents different types or categories of research needed to understand physical activity behavior and how to promote it more effectively.
Five specific types of research are depicted around the perimeter of the figure, with each level building sequentially on the others to systematically advance behavioral research on physical activity. Basic Research provides the foundation for understanding physiological and biochemical mechanisms influencing disease risk and health. Health Outcomes Research then establishes specific associations with health indicators and facilitates the establishment of physical activity guidelines. Surveillance Research provides insights about secular trends with regard to physical activity as well as information about patterns, trends, and disparities in the population. Theory and Correlate Research seeks to understand the causes and correlates that influence the behavioral patterns and evaluates theoretical models designed to study and influence physical activity behavior. Intervention Research applies these insights to plan and evaluate intervention methods to influence physical activity and sedentary behavior in different settings and populations. The various types of research are linked to the center box labeled Physical Activity and Sedentary Behavior because these are the central behaviors of interest. The line between this center box and Intervention Research is a double-headed arrow because this is the only form of research designed specifically to promote change in these behaviors.
Figure 2: Behavioral Epidemiology Framework
Source: Welk, 2002.8
A prerequisite to advance research in these areas is to have accurate estimates of the underlying physical activity and sedentary behaviors. Therefore, assessment strategies have been placed in the center of the model. A novel adaptation in this adapted version of the framework is the depiction of three overlapping circles capturing the major categories of assessments:
- Report-based measures include various types of self- or proxy-report measures (e.g., questionnaires, diaries, logs).
- Monitor-based measures include various measures that directly or indirectly evaluate movement (e.g., accelerometer-based activity monitors, pedometers, multi-sensor monitors (e.g., devices that combine monitor-based measures such as accelerometer and heart rate), heart rate monitors, various smartphone apps (see Section 8 for more detail), and global positioning system [GPS] devices).
- Criterion measures include the doubly-labeled water technique, indirect calorimetry, and various direct observation measures that involve direct coding of behavior (e.g., time, intensity, type, location).
The independent but partially overlapping portions of the circles illustrate the fact that the measures provide somewhat different views of the physical activity construct.9-10 The overlapping circles also reinforce the advantages of employing multiple measures to triangulate outcomes.
The large space outside of the circles illustrates the fact that the available methods only capture estimates of the actual (true amount of) physical activity conducted. Although it could be argued that criterion measures explain more of the variance in the true physical activity (compared with report-based or monitor-based measures), the point is that all measures provide incomplete views of physical activity behavior. Thus, the methods represented in the circles should be viewed as surrogate measures intended to provide estimates of what actually occurs. Thus, all measures (even criterion measures) inherently contain errors of estimation, which is commonly called measurement error. It is important to take steps to minimize error during data collection, to control the amount of error introduced during processing, and to interpret all results in the light of overall measurement error.
The focus of the NCCOR Measures Registry is on field-based measures that are widely used in research applications. Other summaries have distinguished these types of measures as either objective or subjective3 but this categorization infuses an inherent, but unintentional, bias. Large discrepancies in prevalence rates and levels of physical activity have been observed when objective and subjective measures are compared.11-13 However, it is important not to assume that one category is better than the other. Monitor-based (e.g., activity monitors) measures are certainly more objective than report-based measures (e.g., diaries), but the key distinction is that they capture amounts of movement as opposed to a person’s perception or recall of physical activity experiences. An unfit individual may perceive a certain activity as vigorous in intensity while a fit person may perceive the same activity as being light intensity. This example describes physical activity intensity expressed in relative terms (i.e., accounting for one’s individual level of fitness) and highlights the need to differentiate between relative and absolute intensity of the physical activity performed. Report-based measures capture relative intensity (i.e., individual perception of intensity) while monitor-based measures capture the absolute intensity and volume of activity without considering perceived or actual effort. It is true that report-based measures contain a considerable degree of subjectivity that can contribute toward sources of error and bias, but different sources of measurement error also plague monitor-based measures. The U.S. Physical Activity Guidelines14 emphasize the distinction between absolute and relative indicators, and this needs to be considered when evaluating physical activity outcomes. The labels of “monitor-based” and “report-based” measures reinforce the message that physical activity behavior can be monitored or reported.
The key point of the framework is that the accurate assessment of physical activity is a priority in all facets of research. In making decisions about the most appropriate instrument, it is first important to fully consider the distinctions between movement and behavior. These distinctions are more fully explained in the next section.
a Not to be confused with classic experimental (e.g., clinical trials) or non-experimental (i.e., cohort, case-control, cross-sectional, and ecologic studies) study designs.