A range of methods are used in the assessment of physical activity, including self-report, systematic observation, motion sensors, cardiorespiratory fitness and free-living indirect calorimetry. Most are moderately correlated at best. Each method has its own strengths and limitations. Due to the complexity of physical activity, outlined previously, researchers and practitioners need to consider what is most important for their purposes when deciding to use a particular technique. Although no one measurement tool can be singled out as the most appropriate, recommendations for the most useful and practical measurement tools can be made after considering factors such as context, type, duration, frequency and intensity of physical activity as well as the constraints under which the research or treatment programme is operating. These techniques are presented in order of increasing feasibility but decreasing validity. Table 3.2 at the end of this section summarises the key advantages and limitations of each measurement technique.
Doubly Labelled Water
Doubly labelled water (DLW, or free-living indirect calorimetry) is a method commonly used to increase the precision and accuracy of physical activity measurement. DLW, which measures EE over several (4-21) consecutive days in a free-living person under normal life conditions (Schoeller, 1988), is the most widely accepted gold standard method by which to measure EE (Aadahl & JÃ¸rgensen, 2003). DLW measures EE of free-living, unrestricted subjects using water labelled with stable isotopes of oxygen and hydrogen (Schoeller & van Santen, 1982). It calculates activity-related EE by combining measurement of total EE with basal metabolic rate. The utility of the DLW method in measuring total EE is demonstrated by its use in a variety of settings, including all age groups, premature infants, hospitalised patients, pregnant women and the elderly (Ainslie, Reilly & Westerterp, 2003). As such, researchers who use this method to assess the relationship between physical activity and other psychological variables or to detect changes in physical activity as a result of interventions can be confident that the results produced regarding EE are accurate. Schoeller (1988) fully describes the validation of the DLW method.
Despite its level of precision, the DLW method is not without disadvantages, which include high cost, limitations for assessing brief periods of EE (Ainslie, Reilly & Westerterp, 2003) and additional demands on participants in terms of time and tasks required. The cost and availability of isotopes and the requirement for analysis by isotope ratio mass spectrometer prohibit DLW from being widely used in studies of large populations. The use of this method among populations with more severe mental health problems may likely present a set of ethical issues. Furthermore, although this technique provides an accurate measure of total EE, it cannot provide information about patterns of physical activity in terms of type, frequency, duration, intensity or context. Consequently, such an approach would not be feasible in a large natural-field experiment or in a test of the effect of interventions on changes in physical activity in both healthy populations and populations that suffer from mental health problems. However, this gold standard measure is believed to offer the most precise estimate of EE and is often used to validate many other types of tools that assess physical activity.
Direct observation, one of the most basic approaches for acquiring information about behaviours, provides information about how people exercise and play, how environments shape the activities individuals participate in and how people use specific facilities (e.g., parks, leisure facilities, walking or cycling paths). It can take place using basic observation methods, systematic forms or sophisticated technology (e.g., lasers). Direct observation often involves a trained observer who codes physical activity behaviours (e.g., sitting, walking, running) undertaken by participants over time in various settings (e.g., playground, park, home). A number of observation systems are available, such as SOPLAY (System for Observing Play and Leisure Activity in Youth), SOFIT (System for Observing Fitness Instruction Time), and the Systematic Pedestrian and Cycling Environmental Scan. The trained observer may either observe participants in person or review video media. Because it is time consuming, direct observation might be used only with small groups in specific settings (Dugdill & Stratton, 2007).
Advantages of direct observation include that self-report bias is eliminated, participants do not need to recall behaviour and the level of detail regarding behaviour patterns and context can be extremely accurate. Furthermore, practitioners may find some of these tools accessible. For example, clinical psychologists working with institutionalised individuals diagnosed with severe mental health conditions may find this type of measurement technique useful for assessing relationships between physical activity patterns or patterns of inactivity and specific mental illnesses because it can combine both context and behaviour to provide powerful data.
However, disadvantages include the time involved in recording and coding behaviour, the time involved in consolidating the plethora of data collected and the costs associated with training coders. Furthermore, the subjective bias of coders may lead to incorrect judgments about the intensities of specific activities, which would affect estimates of EE. As such, it would be important to consider whether information regarding patterns of behaviour combined with context or specific EE data are required to fulfill the aims of a research project or treatment programme.
Motion sensors, such as pedometers and accelerometers, can be used to detect body movement and estimate physical activity (Spruijt-Metz et al., 2009). Accelerometers are devices that measure bodily movements in terms of acceleration. This measurement can then be used to estimate the intensity of physical activity, and therefore EE, over time (Burton et al., 2005; Chen & Basset, 2005). Accelerometers can measure human activity on vertical (uniaxial accelerometers), anterior - posterior and medial - lateral (triaxial accelerometers) planes. EE can then be then estimated from vector magnitude counts using a proprietary algorithm, which is a composite of counts from these planes of motion (Howe, Staudenmayer & Freedson, 2009). Accelerometers can be used repeatedly on numerous participants, are more accurate than pedometers and are less expensive (approximately Â£200 per device; ActiLife, 2009) than other objective methods such as DLW (approximately Â£200 per participant per procedure; Friedman & Johnson, 2002). However, they remain too expensive to use in studies that assess large numbers of people (Wood, 2000). Nevertheless, these devices are a more feasible and participant-friendly method of attempting to validate a self-report questionnaire than is DLW. They also provide the data necessary to allow researchers to distinguish between light, moderate and vigorous physical activity as well as between continuous and intermittent activity modes (Crouter, Clowers & Bassett, 2006). A recent review of accelerometers against DLW found ActiGraph (previously named CSA/MTI) models to be of the most valid types tested; they produce an average correlation of r = .57 (Plasqui & Westerterp, 2007). Limitations of accelerometers include the increased time required to analyse the large amount of data provided, participant burden of wearing the device, that they cannot be feasibly used to test large-scale interventions, and that they cannot provide information about the specific type of activity (e.g., playing football, going to the gym) or the context in which it is performed.
Some studies have attempted to validate accelerometers among populations diagnosed with mental health conditions. For example, Sharpe and colleagues (2006) conducted a study to assess the validity of the RT3 accelerometer against DLW in people with schizophrenia. They found that the accelerometer overpredicted energy expended on physical activity by an average of 148 kcal/day (standard deviation = 413 kcal/day); this varied from an underestimation of 614 kcal/day to an overestimation of 582 kcal/day. The authors suggested that the RT3 accelerometer is a poor tool for measuring activity EE in sedentary men with schizophrenia. As such, the results of studies that have used this measurement tool in mentally ill populations may be questionable. For example, the recent intervention study of Jerome and colleagues (2009) used the RT3 accelerometer to measure physical activity in persons with mental illness. The authors of this study concluded that participants were undertaking approximately 120 min/wk of moderate-intensity physical activity on average, which would equal approximately 70 kcal/day. Given that the results of the study by Sharpe and colleagues (2006) indicated that the RT3 accelerometer overpredicted EE, it is possible that the results found by Jerome and colleagues (2009) overestimate the amount of physical activity undertaken by participants. Consequently, this might affect the validity of the relationships found between physical activity and various mental health variables measured in this study. Sharpe and colleagues (2006) did, however, indicate that the RT3 appeared to be a valid measure of physical inactivity in men with schizophrenia. Therefore, it could be used for research or clinical purposes to quantify the contribution of sedentary behaviour to medical conditions associated with inactivity. Sharpe and colleagues (2006) also recommended that using the RT3 to validate questionnaires may not be appropriate until it is more robust.
Indirect Objective Measures
Indicators of the physiologic response to physical activity include heart rate and pulmonary gas exchange. Heart-rate monitoring is a promising measurement method because heart rate is a physiological parameter that correlates well with, and strongly predicts, EE (Strath et al., 2002). Most heart rate monitors include software that converts heart rate data into an estimate of EE. However, one of the limitations of heart rate monitoring is that training state and individual heart rate characteristics can affect the relationship between heart rate and oxygen consumption. Higher levels of accuracy can be obtained through a graded submaximal exercise test that calibrates participant heart rate to simultaneous oxygen consumption. This information allows for the construction of a calibration curve that estimates EE at moderate and strenuous levels of exercise. (A linear relationship exists between increasing heart rate and oxygen consumption; Freedson & Miller, 2000.) Measuring heart rate is a common method used to describe intensity and duration of physical activity and is a relatively inexpensive method of measuring EE. However, heart rate is affected by factors other than physical activity, such as emotional stress, temperature, humidity, dehydration, posture and illness (Ainslie, Reilly & Westerterp, 2003). These factors can influence heart rate without causing associated changes in oxygen consumption. Given the potentially increased fluctuations in emotions in individuals diagnosed with mental health conditions, this method may not be the most appropriate for obtaining accurate physical activity levels. Furthermore, the relationship between oxygen uptake and heart rate is weak at low levels of activity (Keim, Blanton & Kretsch, 2004). Evidence suggests that individuals suffering from mental health problems tend to perform less intensive physical activity than do members of the general population (e.g., Brown et al., 1999). Therefore, the heart rate method may not provide accurate information about the physical activity of mental health populations. Also, the heart rate method is unable to identify types of activity or the context in which physical activity is performed. Although heart rate is a physiological marker for physical activity and may provide a general picture of physical activity patterns, it may not be the best method available for obtaining an accurate estimate of EE.
Pedometers, which measure steps on a single axis as well as calories expended, are less expensive than some other types of motion sensors. As such, these devices are an attractive alternative to self-report in large observational or intervention studies. Public health campaigns have also promoted pedometers as a motivational tool for achieving the goal of 10,000 steps daily. A range of pedometers is available. The pedometer that is most suitable for a particular study may depend on factors such as the research question (outcome of interest), population, available funds, context and validity of the model. For example, when considering context, researchers or practitioners who wish to understand the types of physical activity being undertaken in a variety of settings would not gain this information using pedometers alone. When considering population, some pedometers are validated in healthy adults but not in other populations such as children, the elderly or those diagnosed with mental illness. Pedometers have demonstrated reduced accuracy in elderly populations because of the slow pace or shuffling nature with which elderly people walk (Cyarto, Myers & Tudor-Locke, 2004). Tudor-Locke and colleagues (2002) evaluated the validity of pedometers in a review of 25 studies and found a strong correlation between pedometer counts and accelerometer output (median of reported correlations is r = .86). However, evidence suggests that the validity of pedometers for measuring EE and distance in normal populations is questionable. The findings of a study testing the validity of 10 pedometers (Crouter et al., 2003) indicated that these devices overestimated distance at slower speeds and underestimated distance at faster speeds. Furthermore, in 8 of the pedometers tested, it was unclear whether the device was measuring gross EE (all the energy expended by an individual during a specific activity) or net EE (the energy expended by an individual during a specific activity minus the resting EE for the equivalent amount of time). The Yamax Digiwalker SW-200 was found to be the most reliable and accurate pedometer available (Crouter et al., 2003). This pedometer was used in a recent study by McKercher (2009) that assessed relationships between physical activity and depression in young adults. This study found that low levels of depression were significantly correlated with moderate levels of physical activity, as measured by the Yamax Digiwalker SW-200, among females. However, this device has not been validated in specific mental health populations, which may limit the validity of these results. It is important to validate physical activity devices among specific populations because patterns of physical activity in these populations may differ from those in the general population. For example, individuals with serious mental illness are significantly less active than the general population (Brown et al., 2004). As such, using a physical activity measure that has been validated with mental health populations in research on physical activity and mental health will increase the strength of the results produced.