Journal Logo


Detection of Lying Down, Sitting, Standing, and Stepping Using Two ActivPAL Monitors


Author Information
Medicine & Science in Sports & Exercise: October 2014 - Volume 46 - Issue 10 - p 2025-2029
doi: 10.1249/MSS.0000000000000326
  • Free


There are numerous methods of classifying physical activity behaviors. Exercise physiologists are interested in categorizing physical activities by intensity (e.g., light, moderate, and vigorous). Epidemiologists often subdivide physical activity into various domains (e.g., occupation, transportation, home based, and leisure time). More recently, researchers have become interested in classifying physical activities by the posture assumed (e.g., lying down, sitting, standing, and stepping).

Prolonged sedentary behavior has long been believed to have deleterious effects on human health. In the 1700s, the Italian physician Bernardino Ramazzini noted that “chair workers” such as cobblers and tailors experienced diseases not typically seen in those with more active occupations (24). The landmark studies of Morris et al. (16–19) and Paffenbarger et al. (21–23) were helpful in providing scientific evidence of increased mortality risk in individuals employed in sedentary occupations. Since then, other studies of more diverse adult populations have confirmed that prolonged sedentary behavior is associated with increased all-cause and/or cardiovascular mortality (4,9,14,29) and reduced life expectancy (10). People who accumulate more sitting time and who fail to break up prolonged sitting with bouts of physical activity are at increased risk of chronic diseases (9,20).

Recently, Levine et al. (12) reported that obese individuals spend an average of 2 h or more per day of sitting down compared with that in lean individuals. Moreover, obese individuals in that study spent 2 h less per day engaging in light- to moderate-intensity physical activity. Hamilton et al. (5) showed that muscle activity as measured by EMG recordings of postural muscles is diminished in the seated position, relative to the upright position. They also demonstrated that physical inactivity in rats decreases lipoprotein lipase activity, plasma triglyceride clearance by skeletal muscle, and HDL cholesterol levels in plasma (5,6). Several investigators have shown that prolonged sedentary time is associated with a higher risk of metabolic syndrome, obesity, and inflammatory markers (2,4,8,26,28).

The recent focus on sedentary behaviors has sparked interest in objective methods of quantifying it. Sedentary behavior has been defined as any waking behavior with body posture in the seated or reclining position and a low metabolic rate in the range of 1.0–1.5 METs (25). It is possible to measure sedentary behavior and postural allocation by using a body suit instrumented with accelerometers/inclinometers (12) or the Intelligent Device for Energy Expenditure and Physical Activity that uses five accelerometers hard-wired to a small computer (30,31). However, a method that is less intrusive and less costly could have broader applications.

The activPAL (PAL Technologies, Glasgow, Scotland) is a small device that classifies a person’s behavior into one of three categories (sitting/lying, standing, and stepping). The activPAL is an accelerometer-based monitor that is attached to the thigh with double-sided tape or a breathable, transparent, adhesive bandage. It discriminates between sitting versus upright posture by measuring static acceleration and determining the orientation of the thigh, and it detects ambulatory activity (steps) by measuring dynamic acceleration. The activPAL has been validated against direct observation performed in laboratory settings (3,11,13) and was found to be highly accurate.

A limitation of the activPAL is that the orientation of the thigh (i.e., the angle of inclination) does not discriminate between the sitting and lying postures because the thigh is horizontal in both of these conditions. When worn for 24 h, a single activPAL would classify time in bed as sitting/lying, which could have significant implications for sedentary behavior researchers because sitting and sleeping are believed to have different effects on human health. To obtain more precise information on postural allocation over a 24-h period, it is necessary to distinguish between the sitting and lying postures.

Thus, the purpose of this study was to determine whether placing a second activPAL on the torso (in addition to one placed on the thigh) could allow the detection of seated versus lying postures. We hypothesized that placing an activPAL monitor on the torso would allow us to correctly classify sitting versus lying down. This distinction is currently not possible using a single thigh-worn activPAL monitor. A semistructured routine that called for participants to perform specific amounts of activity for various amounts of time was designed.


Fifteen volunteers (mean ± SD: age, 25.0 ± 9.4 yr; height, 176.3 ± 10.5 cm; weight, 71.58 ± 13.76 kg) agreed to take part in the study. Upon their arrival, they received a consent form approved by the university’s institutional review board, and the study protocol was explained to them. They were given an opportunity to read the consent form and ask questions about anything that was unclear to them. All participants provided a written informed consent before taking part in the study. The participants performed a semistructured routine consisting of lying on a bed (supine, prone, and on the side), sitting in a chair, standing, sweeping floors, walking at 3 mph, and jogging at 6 mph. Each activity was performed for 3 min. A 2- to 3-s transition period occurred between activities.

Body posture and activity monitoring

The activPAL (Professional) is a small (53 × 35 × 7 mm) rectangular device weighing 16 g. It contains a microelectromechanical system piezo-capacitive accelerometer with a sensing range of ±2g. The accelerometer senses both static acceleration due to gravity (i.e., 9.8 m·s−2) and dynamic acceleration due to body movements. Our empirical bench tests suggested that when the angle of inclination exceeds approximately 20° from horizontal (0°), the device predicts the upright position. The newest version of the device (activPAL3) can record acceleration continuously over 7–14 d. It is worth noting that the period needed to obtain reliable measures of habitual physical activity ranges from 3–5 d in adults and 4–9 d in youths (27).

One activPAL device was affixed to the midline of the right thigh midway between the hip joint and knee joint using a Tegaderm® waterproof, breathable bandage. A second activPAL device was attached to the chest (over the lower right rib cage) in the same manner. Both devices were synchronized to an external timepiece and were initialized to begin data collection at the same time. Data were reported at 15-s intervals.

The data were downloaded to an Excel file, and the data from the two activPALs were aligned using the activPAL time stamps for each device. The activPAL software divides the body posture into one of three categories: sitting/lying, standing, and stepping. The activPAL was designed to be worn on the thigh, and thus, “sitting/lying” could indicate either sitting or lying down. “Standing” indicates an upright posture without continuous ambulation. “Stepping” indicates an upright posture with continuous ambulation. For each device and epoch, the activPAL software category was examined. The category in which most time for that 15-s epoch was spent was used to determine the body’s position. If the trunk and thigh devices were both classified by the activPAL software as sitting/lying, we could assume that both devices were in the horizontal position. Thus, we classified the body position as “lying down.” If the trunk device was classified by the activPAL software as standing and the thigh device was classified as sitting/lying, we could assume that the chest device was in the vertical position and the thigh device was in the horizontal position. Thus, we classified the body position as “sitting.” If the trunk and the thigh device were both classified by the activPAL software as upright, we could assume that both devices were in the vertical position without movement. Thus, we classified the body position as “standing/light activity in the upright position.” If the thigh device indicated continuous stepping (i.e., walking or running), then we classified the body position as “stepping.”

Direct observation

Direct observation of body posture during each activity was used as the validation criterion. A trained investigator observed the laboratory trial of semistructured bouts and recorded the time at which each posture change took effect. (Because our pilot work had established that both standing and sweeping resulted in the accumulation of little or no steps, the observer coded these activities as “standing/light activity in the upright position.” The observer coded walking and running activities as “stepping.”) Thus, the body posture for each 15-s interval was determined by a criterion method.

Statistical methods

A confusion matrix was developed to show the classification accuracy of the two activPAL methods. Statistical analyses were conducted using IBM SPSS Statistics version 21.0 (Cary, NC). Cohen κ was used to determine the intermethod agreement. κ values greater than 0.75 represent excellent agreement between the two methods. P < 0.05 was taken as statistically significant (i.e., greater than chance alone).


The use of two activPAL devices (one on the thigh and one on the chest) resulted in accurate identification of the four categories (lying down, sitting, standing/light activity, and stepping). There was excellent agreement between the two methods (Table 1).

Confusion matrix showing the classification accuracy of the two-activPAL method.

Lying down was correctly classified 100% of the time, sitting was correctly classified 100% of the time, standing/light activity was correctly classified 90.8% (327/360) of the time, and stepping was correctly classified 100% of the time. The κ statistic was 0.968, indicating a strong level of agreement between the two methods (direct observation and two-accelerometer technique) (P < 0.001). The only misclassifications that occurred were when some of the sweeping 15-s periods were misclassified as stepping.

Figure 1 shows an example of a graph obtained from a participant who wore the two activPAL monitors over a 24-h period. The amount of time spent lying down (8.9 h·d−1), sitting (7.2 h·d−1), standing or doing light intensity activities in the upright position (6.0 h·d−1), and stepping (1.9 h·d−1) is shown. The total daily step count, determined from the activPAL on the thigh, was 10,392 steps per day. (A previous study reported that the activPAL provided valid step counts in adults during continuous walking bouts (7)). The number of breaks per sedentary hour was 46 breaks per 7.2 h of sitting; that is, 6.4 breaks per sedentary hour.

Representative graph of postural allocation and physical activity over 24 h in a single subject (22 yr old). Two activPAL devices (worn on the trunk and thigh) were used to discriminate time spent lying down, sitting, standing/light activity in the upright position, and stepping.


This study reports on a method for measuring sedentary activity through the use of two activPALs, one placed on the thigh and one placed on the chest. The criterion validity of the proposed method was high, and there was excellent agreement between the two-activPAL method’s estimate of time spent in four different categories (i.e., lying down, sitting, standing/light activity in the upright position, and stepping) and that measured by direct observation.

The few errors that occurred were attributable to misclassification of sweeping as “stepping.” We considered sweeping to be a household task that involves both standing and ambulating, and this was supported by the results, which showed that sweeping usually elicited 0 step in 15 s, but occasionally, it elicited 1–12 steps in 15 s. Because of the general movements associated with sweeping and the slight increase in energy expenditure above standing quietly, we chose to have the trained observer classify this activity (sweeping) as “standing/light activity in the upright position,” as opposed to “stepping.” However, when participants moved from one part of the room to another during the sweeping task, they could have taken a series of steps in succession (i.e., walking). Thus, some of the misclassification could have been due to errors in the criterion method rather than the two-activPAL method.

With the two-activPAL method, the monitors can be worn continuously for 7 d or more. If activPAL devices are adhered to the skin using Tegaderm® bandages, the user can even shower while wearing them. Some subjects reported minor discomfort from having the monitors adhered to their skin, but it may be possible to further improve the comfort by having a vest and a leg sleeve made of an elastic breathable fabric (e.g., lycra or spandex) with pockets sewn in.

Measurement of body posture is important in scientific research. It has been reported that obese individuals sit about 2 h·d−1 more than that in normal-weight individuals (12), but this was based on a small sample of individuals. This finding needs to be replicated in a larger group of individuals using low-cost, feasible methods. Currently, the most accurate methods of assessing postural allocation are the body suit designed by Levine et al. (12) and the Intelligent Device for Energy Expenditure and Physical Activity monitor that uses five accelerometers hard-wired to a small computer (30,31). A simpler method that has been used to assess sedentary behavior in epidemiological studies involves the use of a waist-worn ActiGraph accelerometer; one can attempt to discriminate sitting versus standing by using a “threshold” of 100 cpm on the vertical axis, below which the subject is assumed to be sitting (15). However, the accuracy of this method is not as great as that of the activPAL (11). In addition, in most epidemiological studies, the participants are instructed to remove the ActiGraph monitor at bedtime, resulting in variable wear times. In our opinion, the two-activPAL method bridges the gap between expensive, intrusive, and highly accurate methods of assessing postural allocation and the simpler methods such as a waist-worn ActiGraph monitor.

Another reason to assess body posture is that the energy requirements of lying down, sitting, and standing are different. According to an extensive review of the literature reported in the Compendium of Physical Activities (1) the mean energy requirements of “sleeping” (0.90 MET) and “lying quietly watching television” (1.0 MET) are similar and slightly less than the requirement for “lying quietly doing nothing” (1.3 METs). (1 MET is defined as 1.0 kcal·kg−1·h−1, equal to an oxygen consumption of 3.5 mL·kg−1·min−1, and represents the average resting metabolic rate of a healthy young adult.) Surprisingly, “sitting quietly, general” (1.30 METs) and “standing quietly, standing in a line” (1.30 METs) have the same energy cost, but most standing tasks such as “standing, fidgeting” (1.80 METs) and “standing tasks, light effort (e.g., bartending, store clerk, assembling, filing, duplicating, librarian, etc.)” (3.0 METs) are higher. “Light household chores (e.g., sweeping, washing dishes, light cleaning, and folding laundry)” range from 1.60 to 2.90 METs (1). Lying, sitting, and standing/light activity are the most common activities for many individuals.

The two-activPAL method bridges the gap between the single-activPAL method and more expensive and intrusive methods of postural allocation. Participants can wear the devices continuously, reducing the chance that they will take them off at bedtime and forget to put them back on when they awaken. This could eliminate the need for researchers to estimate nonwear time and reduce the burden on participants to complete wear/nonwear and bed/wake logs. For studies of energy balance, it may be important to know how much time an individual spends in each posture. Although the two-activPAL method by itself does not yield accurate estimates of total daily energy expenditure, it could be used in conjunction with other physical activity monitors to provide greater resolution at the low end of the intensity spectrum.

The number of breaks in sitting can be measured with two activPALs, the same way it can be measured with a thigh-worn activPAL (13). Our laboratory study was not designed to validate breaks in sitting because there were few sit-to-stand transitions. However, the number of breaks in sitting per sedentary hour that we observed in an individual who wore the devices for 24 h was within the range of values reported by Lyden et al. (13) who had subjects wear an ActivPAL on the thigh for 10 h. Their study showed that the activPAL was a valid tool for measuring breaks from sedentary time per hour, and using a metric that expressed breaks relative to total sedentary time (i.e., breaks per sedentary hour) was preferable.

The present method has both strengths and limitations. The primary strength is that it enables researchers to discriminate between lying down and sitting, providing an improved ability to detect seated activities. A limitation is that it does not involve measuring the metabolic rate; thus, there are a few sitting activities (e.g., canoeing, kayaking) that could potentially be classified as “sedentary” even though the energy expenditure exceeds 1.5 METs. However, these are not likely to represent a major portion of the sitting activities performed by most adults. A second limitation is that this method does not determine whether a person is sleeping. Technically, sedentary behaviors are those where the individual is sitting or reclining, the energy expenditure is between 1.0 and 1.5 METs and the individual is awake (25). Thus, if an individual were to lie down on a couch and read a book, this would be considered sedentary behavior but those minutes would appear in the “lying down” category rather than in the “sitting” category. Furthermore, if an individual fell asleep while sitting on a chair, it would be categorized as “sitting” even though the person is sleeping. This simply points out that the current method detects sitting per se, which is nearly (but not 100%) synonymous with sedentary behavior.

In summary, the addition of a second activPAL worn on the torso allows researchers to discriminate between sitting and lying down, which is important for determining total sitting time and the number of “breaks” in sitting. The use of two activPAL monitors, placed on the thigh and the torso, had good classification accuracy for four categories (lying down, sitting, standing/light activity in the upright position, and stepping). Thus, the two-activPAL method allows researchers to obtain more detailed information on postural allocation compared with that in the use of a single activPAL on the thigh.

We appreciate the assistance of Cary Springer, University of Tennessee Statistical Consulting Center, who performed the statistical analysis.

No funding was received for this study.

The authors report no conflict of interest. The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Ainsworth B, Haskell W, Herrmann S, et al. [Internet]. The Compendium of Physical Activities Tracking Guide. 2013 May 5. Available from:
2. Bankoski A, Harris T, McClain J, et al. Sedentary activity associated with metabolic syndrome independent of physical activity. Diabetes Care. 2011; 34: 497–503.
3. Dowd K, Harrington D, Donnelly A. Criterion and concurrent validity of the activPAL™ Professional physical activity monitor in adolescent females. PloS One. 2012; 7 (10). Available from. doi:10.1371/journal.pone.0047633.
4. Dunstan D, Barr E, Healy G, et al. Television viewing time and mortality: the Australian Diabetes, Obesity and Lifestyle Study (AusDiab). Circulation. 2010; 121 (3): 384–91.
5. Hamilton MT, Hamilton DG, Zderic TW. Role of low energy expenditure and sitting in obesity, metabolic syndrome, type 2 diabetes, and cardiovascular disease. Diabetes. 2007; 56 (11): 2655–67.
6. Hamilton MT, Hamilton DG, Zderic TW. Exercise physiology versus inactivity physiology: an essential concept for understanding lipoprotein lipase regulation. Exerc Sport Sci Rev. 2004; 32 (4): 161–6.
7. Harrington DM, Welk GJ, Donnelly AE. Validation of MET estimates and step measurement using the ActivPAL physical activity logger. J Sports Sci. 2011; 29 (6): 627–33.
8. Healy GN, Wijndaele K, Dunstan DW, et al. Objectively measured sedentary time, physical activity, and metabolc risk: the Australian Diabetes, Obesity, and Lifestyle Study (AusDiab). Diabetes Care. 2008; 31 (2): 369–71.
9. Katzmarzyk PT, Church TS, Craig CL, Bouchard C. Sitting time and mortality from all causes, cardiovascular disease, and cancer. Med Sci Sports Exerc. 2000; 41 (5): 998–1005.
10. Katzmarzyk P, Lee I. Sedentary behaviour and life expectancy in the USA: a cause-deleted life table analysis. BMJ Open. 2012; 2 (4). Available from. doi:10.1136/bmjopen-2012-000828.
11. Kozey-Keadle S, Libertine A, Lyden K, Staudenmayer J, Freedson PS. Validation of wearable monitors for assessing sedentary behavior. Med Sci Sports Exerc. 2011; 43 (8): 1561–7.
12. Levine JA, Lanningham-Foster LM, McCrady SK, et al. Interindividual variation in posture allocation: possible role in human obesity. Science. 2005; 307 (5709): 584–6.
13. Lyden K, Kozey Keadle SK, Staudenmayer JW, Freedson PS. Validity of two wearable monitors to estimate breaks from sedentary time. Med Sci Sports Exerc. 2012; 44 (11): 2243–52.
14. Matthews CE, George SM, Moore SC, et al. Amount of time spent in sedentary behaviors and cause-specific mortality in US adults. Am J Clin Nutr. 2012; 95 (2): 437–45.
15. Matthews CE, Chen KY, Freedson PS, et al. Amount of time spent in sedentary behaviors in the United States, 2003–2004. Am J Epidemiol. 2008; 167 (7): 875–81.
16. Morris J, Heady J, Raffle P, Roberts C, Parks J. Coronary heart-disease and physical activity of work. Lancet. 1953; 265 (6796): 1111–20 concl.
17. Morris J, Heady J, Raffle P, Roberts C, Parks J. Coronary heart-disease and physical activity of work. Lancet. 1953; 265 (6795): 1053–7 contd.
18. Morris JN, Crawford MD. Coronary heart disease and physical activity of work; evidence of a national necropsy survey. Br Med J. 1958; 2: 1485–96.
19. Morris JN, Heady JA, Raffle PA. Physique of London busmen; epidemiology of uniform size. Lancet. 1956; 271 (6942): 569–70.
20. Owen N, Healy GN, Matthews CE, Dunstan DW. Too much sitting: the population health science of sedentary behavior. Exerc Sport Sci Rev. 2010; 38 (3): 105–13.
21. Paffenbarger RS, Hyde RT, Wing AL, Hsieh CC. Physical activity, all-cause mortality, and longevity of college alumni. N Engl J Med. 1986; 314: 605–13.
22. Paffenbarger RS Jr, Laughlin ME, Gima AS, Black RA. Work activity of longshoremen as related to death from coronary heart disease and stroke. N Engl J Med. 1970; 282 (20): 1109–14.
23. Paffenbarger RS, Wing AL, Hyde RT. Physical activity as an index of heart attack risk in college alumni. Am J Epidemiol. 1978; 108: 161–75.
24. Ramazzini B. De morbis artificum diatriba [diseases of workers]. 1713. Am J Public Health. 2001; 91 (9): 1380–2.
25. Sedentary Behaviour Research Network. Standardised use of the terms “sedentary” and “sedentary behaviours”. Appl Physiol Nutr Metab. 2012; 37: 540–2.
26. Thorp AA, Healy GN, Owen N, et al. Deleterious associations of sitting time and television viewing time with cardiometabolic risk biomarkers: Australian Diabetes, Obesity and Lifestyle (AusDiab) study 2004–2005. Diabetes Care. 2010; 33 (2): 327–34.
27. Trost SG, McIver KL, Pate RR. Conducting accelerometer-based activity assessments in field-based research. Med Sci Sports Exerc. 2005; 37 (11 Suppl): S531–43.
28. Wagner A, Dallongeville J, Haas B, et al. Sedentary behavior, physical activity and dietary patterns are independently associated with the metabolic syndrome. Diabetes Metab. 2012; 38 (5): 428–35.
29. Warren TY, Barry V, Hooker SP, Sui X, Church TS, Blair SN. Sedentary behaviors increase risk of cardiovascular disease mortality in men. Med Sci Sports Exerc. 2010; 42 (5): 879–95.
30. Zhang K, Pi-Sunyer FX, Boozer CN. Improving energy expenditure estimation for physical activity. Med Sci Sports Exerc. 2004; 36 (5): 883–9.
31. Zhang K, Werner P, Sun M, Pi-Sunyer FX, Boozer CN. Measurement of human daily physical activity. Obes Res. 2003; 11 (1): 33–40.


© 2014 American College of Sports Medicine