1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References

The importance of physical activity in weight management is widely documented. Although accelerometers offer an objective measure of activity that provide a valuable tool for intervention research, considerations for processing these data need further development.


This study tests the effects of using different criteria for accelerometry data reduction.

Design and Methods:

Data were obtained from 2,240 overweight and obese individuals with type 2 diabetes mellitus (T2DM) from the Look AHEAD study, with 2,177 baseline accelerometer files used for analysis. Number, duration, and intensity of moderate (≥3 metabolic equivalents (METS)) and vigorous (≥6 METS) activity bouts were compared using various data reduction criteria. Daily wear time was identified as 1,440 min/day minus non-wear time. Comparisons of physical activity patterns for non-wear time (using either 20, 30, or 60 min of continuous zeros), minimal daily wear time (8, 10, and 12 h), number of days with available data (4, 5, and 6 days), weekdays vs. weekends, and 1- or 2-min time interruptions in an activity bout were performed.


In this mostly obese population with T2DM (BMI = 36.4 kg/m2; mean age = 59.0 years), there were minimal differences in physical activity patterns using the different methods of data reduction. Altering criteria led to differences in the number of available data (sample size) meeting specific criteria.


Although our results are likely directly applicable only to obese individuals with T2DM, an understudied population with regards to physical activity, the systematic analysis for data reduction employed can be more generalizable and provide guidance in this area in the absence of standard procedures.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References

Being more physically active is associated with a reduced risk of morbidity and all-cause mortality (1,2), and research shows that increased physical activity is a critical component of successful obesity treatment (3,4). Several large randomized controlled trials in overweight and obese persons with type 2 diabetes mellitus (T2DM), including the Diabetes Prevention Program (5,6) and Look AHEAD (7) demonstrate sustained weight loss with the inclusion of physical activity as part of the intervention. Data from observational and randomized controlled trials also indicate that physical activity is important in reducing cardiovascular events associated with T2DM (8-,9,10,11). Total volume of activity, including duration and frequency are prominent in the recommendations for physical activity in the management of diabetes (12).

A challenge in developing and evaluating physical activity training interventions is obtaining an objective measure of physical activity, particularly those that can assess patterns of physical activity. Historically, large-scale clinical studies have relied on self-report questionnaires for physical activity assessment. These subjective instruments are limited by recall bias and difficulty in developing and obtaining an appropriate population specific instrument. Assessment of physical activity using an accelerometer is advantageous in that it is an objective measure that circumvents recall bias. Numerous studies have demonstrated these devices provide valid and relatively reliable estimates of physical activity patterns and activity energy expenditure in a wide variety of populations (13-,14,15,16). However, as suggested by Masse (17), there is a clear lack of standards in accelerometer data reduction; clarification of methodological issues is required for appropriate data interpretation. Some of these issues include: (i) identifying registered time as an indicator of wear period in a day; (ii) identifying minimal wear requirement for a useable day, a.k.a. wear time; (iii) aggregating days of data; and (iv) defining intensity and duration of bouts for physical activity patterns. To date, one of the largest uses of accelerometers to assess physical activity behavior has been from the 2003-2006 National Health and Nutritional Examination Survey (NHANES), which obtained at least 4 days of accelerometer data from nearly 4,900 participants (18-,19,20). Although NHANES publications described criteria used for determining wear time and times the device is worn, there is minimal justification provided for their approach (18,21) and a lack of data on the implications of using alternate criteria.

The Look AHEAD study, which is a randomized controlled clinical trial investigating the impact of weight loss on the primary and secondary prevention of cardiovascular disease in obese adults with T2DM, included accelerometry to provide an objective assessment of physical activity as a substudy in over 2,200 participants. These accelerometer data provide a unique opportunity to study the impact that different data reduction criteria have on the interpretation of physical activity behavior. We investigated time worn per day, minimal wear time per day, number of days of available data, and use of a short break in an activity bout on physical activity patterns. The aim of this investigation was to examine the methodology for examining accelerometer data reduction for large cohort studies, and not to provide descriptive physical activity data in this cohort. These cross-sectional analyses are presented elsewhere (22). Based on the role of physical activity in obesity management, the current analysis adds to our understanding for implementing and interpreting data from accelerometers.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References

Look AHEAD trial description

Look AHEAD is a multicenter clinical trial with its primary objective to examine, in overweight and obese volunteers with T2DM, the long-term effects of an intensive lifestyle intervention program designed to achieve and maintain weight loss by decreased caloric intake and increased physical activity compared to a control condition involving a program of diabetes support and education on cardiovascular morbidity and mortality. Specific details related to design and procedures for this trial have previously been described (23). A total of 5,145 participants were recruited across 16 clinical sites in the United States. A total of eight sites were selected to be a part of the accelerometer sub-study, which were used for the present analysis, and each site included these data collection procedures in their protocol. The number and location of sites chosen from the Look AHEAD study was determined by the executive committee for Look AHEAD. Because of the cost of these data, it was prohibitive to include participants across all clinic sites in the study.

In total, baseline data from 2,240 participants were examined, with 2,177 files being useable in the analysis procedures. Participants provided written consent approved by the local Institutional Review Boards before participation.

Accelerometry methodology

The RT3 triaxial accelerometer (StayHealthy, Monrovia, CA) was selected for use in the Look AHEAD Trial. The RT3 unit is the size of a pager and is worn on the waist. Participants were instructed to wear the device at the approximate location on the anterior superior iliac spine during the waking hours for a period of 7 consecutive days. Participants were oriented to procedures for wearing the accelerometer and provided a handout that included information on the proper placement of the accelerometer on the body, dates (days) that the accelerometer should be worn, procedures to follow should there be trouble with the operation of the accelerometer, and procedures for returning the accelerometer. If desired, participants could request and were provided with a belt for attaching the unit when wearing certain clothing that did not allow a traditional belt to be worn or did not have a waistband (i.e., certain dress styles).

The RT3 accelerometer was programmed to collect data at 1-min intervals. Before the participant wore the accelerometer, the device was initialized by clinic staff. During the seven day period of data collection, participants were instructed not to alter their typical pattern of physical activity. A diary was provided for the participant to record when they wore the accelerometer. The diary was used as a tool by the clinics to validate and record the use of the accelerometer. Upon return of the accelerometer, an initial visual inspection of the diary was performed to confirm that the accelerometer was worn at least five days within the seven day period. If the diary showed that participants reported wearing the unit for fewer than five days, participants were asked to wear the accelerometer for an additional seven day period. Recent work with activity counts from RT3′s show that they have good intra-unit reliability (24) with a coefficient of variation of <1.81%. They have also been shown to have a coefficient of variation between units ranging from 9.5-34.7% (24) with a intraclass correlation between accelerometers of 0.75 with 95% confidence interval of 0.46-0.95 (25). For the Look AHEAD study, clinics did not perform routine calibration. However, units were serviced regularly by the manufacturer throughout the study.

Training and certification of staff across sites. Staff from each of the 8 clinics participated in centralized training involving the operation of the accelerometer, participant instructions, and data transfer to the Coordinating Center. In addition, staff completed a certification process before data collection for this study, with recertification occurring annually throughout the study.

Central quality control. Accelerometry data files were uploaded to the Look AHEAD data coordinating center (Wake Forest University, Winston-Salem, NC) via the study website, with a tiered approach for data quality control implemented. For a qualitative and validity review, a graph of the minute-by-minute activity was generated at the coordinating center. Partial days at the beginning and the end of the 7-day wearing period were excluded from subsequent analyses. Each individual graph was visually inspected and was reviewed and graded as being “valid,” “invalid,” or “contains no data.” The following criteria were used to determine whether the data file was “valid”:

  • Presence of activity energy expenditure, as depicted by vertical lines, over the course of the wear period;

  • Presence of episodes of activity interspersed with sustained periods of zero activity, indicating the unit was removed for sleeping;

  • Weekly activity energy expenditure was within limits for a varied population (between 200 and 15,000 kcal).

Data files not meeting these criteria were considered “invalid” and were not used in subsequent summaries. Assignment of the “contains no data” grade usually occurred as a result of a malfunction with the unit, problem with the interface procedure between the RT3 unit and computer, or inability to generate a graph from the data file. For files graded as “invalid” or “contains no data” clinics attempted to recollect data on the participant according to standard procedures and resubmit for review again. These criteria provided an overview of the data as well as identifying extreme outliers. These were meant to be inclusive for data processing. This initial qualitative review of all files was performed by one investigator throughout the entire study duration (G.D.M.).

Defining criteria for accelerometer data reduction

Computing daily wear time. Registered time as an indicator of daily wear time, which is the length of time that an accelerometer is worn daily, was determined by subtracting “non-wear time” from total minutes possible in a day (1,440 min). Within the literature different criteria have been used to determine periods of non-wear time throughout the day, with 20, 30, and 60 consecutive minutes without activity (“zero” counts on the accelerometer) used to identify these periods (26,27). The sum of periods of non-wear time was the total daily non-wear time. Daily wear time was then computed by subtracting the total daily non-wear time from 1,440 min.

Defining valid days based on wear time. Based on the published literature, different criteria were applied to the data to define a valid day based on wear time of the accelerometer. These include the following:

  • ≥10 h/day of wear-time of the accelerometer (17)

  • 80% of a standard day, with a standard day often defined as the length of time in which 70% of the study participants wore the monitor (17). Based on data for this study, a standard day was 13.18 h, with 80% being 10.55 h. Because this was similar to the ≥10 h criteria defined above, ∼2 h below and above this value (8 and 12 h) were established as two other criteria that could be used to define a valid day for evaluation in this study.

The analysis for computing daily wear time and defining valid days based on wear time was performed in a step-wise manner. The first criterion was to determine length of periods of non-wear time (20, 30, or 60 min). Once that value was determined, the number of hours required for a valid day was analyzed.

Defining physical activity bouts. Physical activity intensity level was expressed in metabolic equivalents (METS) by dividing the estimated energy expenditure per minute by estimated resting energy expenditure per minute as computed using the proprietary software provided by StayHealthy. Periods of moderate-to-vigorous intensity physical activity were defined as ≥3 METS and vigorous activity was defined as ≥6 METS. The criteria for adults for moderate-to-vigorous activity were recently established by the American College of Sports Medicine and the American Heart Association (28). Activity bouts to describe physical activity patterns were defined as: ≥3 METS for ≥1 min, ≥3 METS for ≥10 min, and ≥6 METS for ≥10 min. Additionally, based on the published cutoffs for activity counts by Rowland et al. (29) in healthy adults, analysis was performed on the activity counts that were reported to coincide with the 3 METs and 6 METs criteria (984.0 and 2,340.8, respectively).

It has been suggested that consideration be made for whether or not a short interruption is allowed during an exercise session (17). This real-life scenario is present when activity intensity goes below a defined criterion for a short period, and then returns to the previously defined level. Thus, an exercise bout may be achieved that otherwise would have been missed. For instance, an individual walks for 8 min at 3 METS; they come to a stoplight and wait 1 min to cross the road. They continue walking for 12 additional minutes. This could count as a single bout of moderate activity of 12 min or if the 1-min interruption is allowed, it would count for a 21 min bout of moderate activity. Similarly, others have examined allowing up to 2 min of interruption and to continue with the bout duration (20). Thus, bouts of activity were defined in three ways; that is, not allowing and allowing for a 1-min and a 2-min break in a bout of moderate or vigorous activity lasting at least 10 min.

Spurious data were defined as any one min that exceeded 100% of the maximal exercise intensity as determined by the maximal graded exercise test that was performed for each Look AHEAD participant (23,30). For cases when the 100% was exceeded, data for that minute were capped at 100% of the maximal MET value and the adjusted value was used in the analyses.

Statistical analysis

All analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC). Three variables were chosen to characterize activity bouts and to compare various methods of data reduction: number of bouts per day meeting the moderate and vigorous intensity levels, duration of bout (in minutes), and intensity of bout (in METS). The variable—“number of bouts per day” was computed as the average daily number of bouts and it was set to be 0 if a participant did not have any activity bouts that met the specific definition. Variables “duration of bout” and “intensity of bout” were defined only when a participant had at least one activity bout of interest and were set to be missing if the participant did not have the type of activity bout.

Results were presented as means ± SD for continuous variables, while median and interquartile range were presented for skewed data. For the definition of a valid day, we have examined nine different criteria: 8, 10, and 12 h daily wear time using 20, 30, and 60 min non-wear time. The resulting sets of summary variables (one for each criterion) for every participant were treated as correlated measurements and were analyzed using mixed effects models (SAS/STAT 9.1 User's Guide). Similarly, mixed effects models were used to examine differences in results using data for the first 4, 5, and 6 full days that met the minimal daily wear time, weekend days (Saturday and Sunday) vs. weekdays, and bouts defined allowing no break vs. allowing for a 1-min break in bouts.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References

Approximately 85.3% (2,240 out of 2,627) of randomized participants at the clinics participating in this substudy had accelerometer data files uploaded to the coordinating center at baseline. The most common reasons for missing files (n = 387) were equipment failure (31.8%, n = 123), participant refusal (22.5%, n = 87), time/scheduling (7.0%, n = 27), and other (38.8%, n = 150). Not all clinics provided every participant the opportunity to participate in the study; this explanation fell within the “other” category. All uploaded files (n = 2,240) were reviewed according to procedures developed for the initial review in data quality control procedures. Of this number, only 95 files (4.2%) were deemed to be invalid. Participant characteristics were assessed between those with valid files vs. invalid files, and although age and gender were similar between the two, race did differ.

Baseline characteristics of participants with valid accelerometer files (n = 2,177) are shown in Table 1. Mean age was 59.0 (SD = 6.8) years, and most were female (57.3%) and white (73.0%). Mean BMI was 36.4 (SD = 6.0) kg/m2 with a maximal MET value of 7.1 (SD = 1.9) as determined by a graded exercise stress test. Over 94% had metabolic syndrome and 86% were hypertensive.

Table 1. Baseline characteristics of accelerometry participants with valid baseline files (n = 2,177)
inline image

Daily wear time

Identification of daily wear time was performed, and using these values, comparisons were made for physical activity patterns between 8, 10, and 12 h of wear-time per day, with non-wear time computed using either 20, 30, or 60 min of continuous zero counts (Table 2). There were no statistically significant differences between the different methods used to define a day's minimal wear time for number, intensity or duration for a bout of activity ≥10 min in duration for moderate (≥3 METs) or vigorous activity (≥6 METs) (Table 2). For our systematic approach, all subsequent analyses for data summary were performed on a single definition of daily wear time. Since our analysis for this cohort showed no differences in the bout criteria identified, we selected 10 h with 30 min of continuous zero counts as the criteria for a useable day.

Table 2. Methods comparison of registered time as an indicator of wear period
inline image

The number of min/day that exceeded at least 3 METs is presented in the bout criteria of ≥3 METs for ≥1 min. This ranged from 26.5 to 28.1 min and signifies that the total time spent above 3 METs was <30 min a day, irrespective of the duration of the bout. This was the only significant effect of the different methods for wear period. In that these analyses were focused primarily on activity bouts of at least 10 min, this result did not factor into our decision of which method of daily wear to use.

As the wear time criteria changed across the nine conditions (Table 2), the number of data files meeting the criteria differed for the moderate intensity ≥10 min exercise bout from 1,346 to 1,433. Similar changes were seen for number of eligible data files for the vigorous exercise bout (398-449). In both cases, this amounted to ∼5-10% drop in number of eligible files as the criteria became more stringent (greater number of h/day required).

Number of wear days

Data were analyzed to determine whether there were differences in physical activity patterns assessing the first 4, 5, or 6 days that met the criteria of ≥10 h of wear time as defined above. Results showed no significant differences between bouts per day, bout duration, or bout intensity when using the first 4, 5, or 6 days of data that met the ≥10 h of wear-time criterion (Table 3); data for activity counts are also shown. For an activity bout meeting ≥984 counts (equivalent to ≥3 METs based on cutoffs by Rowlands et al. (29) for the RT3 accelerometer) for ≥10 min, the average intensity was ∼1,550 counts for assessing the first 4, 5, or 6 days of data collection. For the high intensity activity bout of ≥2,340 counts (≥6 METs), the values were ∼3,100 counts for all three conditions.

Table 3. Comparison of 4, 5, and 6 full days for physical activity patterns
inline image

Changes in eligible data files across the different criteria for day number were apparent for moderate intensity activity with a sample size of 1,099 for using the first 6 full days compared to 1,204 for using the first 4 full days, an ∼10% reduction. Number of data files meeting the vigorous activity criterion ranged from 383 to 353, an ∼8% decrease.

Weekend vs. weekdays

Comparison of data from weekend days vs. weekdays is presented in Table 4. The bout duration for ≥3 METs for ≥10 min was higher on weekends vs. weekdays by about 1 min. There was a significant difference for average bout intensity for vigorous activity with a bout of vigorous activity during the weekend occurring at a higher MET level (7.65 vs. 7.44 for weekends vs. weekdays). However, for each of these significant differences observed, the effect size was small (0.04 for bout duration of moderate intensity exercise and 0.04 for bout intensity for a vigorous exercise bout). Using activity counts, the difference in number of counts per bout was higher for weekdays vs. weekend days achieved statistical significance at P = 0.048. No difference was observed for the more vigorous activity bout.

Table 4. Comparison of weekends vs. weekdays for physical activity patterns
inline image

There were large differences in the number of files that had eligible days during the weekend compared to weekdays. For moderate activity, the sample size was 1,192 for weekdays vs. 797 for weekend days, a 33% difference. Similarly, for vigorous activity, sample size was 625 for weekdays and 361 for weekend days, a 42% discrepancy.

Further analysis was performed to determine the differences in activity patterns using no weekend days, 1 weekend day, and 2 weekend days for the first 4 full days and the first 5 full days of data. Analyses performed incorporating either 1 or 2 weekend days did not change activity patterns compared to having no weekend days in data collected for either the first 4 or 5 full days (data not shown).

No time interruption (no breaks) vs. 1 and 2 min time interruption (breaks)

The number of bouts for ≥10-min and ≥3 METS was greater when a 1- or 2-min interruption was allowed (0.5, 0.6, and 0.6 for no breaks and 1- and 2-min break, respectively) (Table 5). For activity bouts defined as ≥3 METs for ≥10 min, the intensity per bout differed for the three criteria with bouts allowing no breaks having the highest intensity. This effect was also similar for vigorous activity.

Table 5. Comparison of 1-min break, 2-min break, and no breaks in physical activity patterns
inline image

Inclusion of the 1- or 2-min interruption provided a larger sample size for both the moderate and vigorous intensity activity bouts. For moderate intensity, there was nearly a 15% drop in eligible files (1,630 for 2-min interruption vs. 1,402 for no interruption). For vigorous intensity, the decrease in sample size was 16% (525 for 2-min interruption vs. 441 for no interruption).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References

Look AHEAD has the largest accelerometry data set available from a randomized controlled clinical trial of overweight and obese adults with T2DM and will provide ample opportunities to study the unique role that physical activity plays in weight management and subsequent clinical outcomes. Whereas our previous work with the Look AHEAD cohort described the baseline physical activity patterns (22), the aim of the current study was to conduct a systematic analysis for establishing criteria for data reduction and how these outcomes can be interpreted. These results on the specific outcomes are not intended to be generalizable across subsamples of the population, but instead this study highlights the various criteria to be considered when examining accelerometer data. The actual criteria utilized and the results obtained for this population is not likely ideal or the “gold standard” across all populations. The authors acknowledge that the effects or lack of effects of varying data reduction procedures employed in this analysis may only apply to those individuals who have similar characteristics and activity patterns as observed in our sample. For example, the relatively sedentary behaviors of our population shown in the small number of activity bouts at the different intensities may be influencing the findings on how to best analyze the data.

The current data provide an important and innovative contribution to Look AHEAD and to the field of obesity research in general. As the cost of accelerometers has decreased, and their use in large epidemiological studies has risen, it has become necessary to bring attention to issues related to the processing of these data. While the use of an objective measure of physical activity has advantages over self-report, it is still fraught with challenges and concerns (31). Since most accelerometer protocols request the user to wear it only during waking hours, determining the length of time during the day the unit is actually being worn can be a challenge, especially for a large and diverse population. If the activity for any time period is zero, then it is not known if the unit was not being worn or if there was no activity. Studies have used sustained periods of zero counts (20, 30, 60 min) to try to assess this. Thus, obtaining the wear time for each individual to assess if they achieved a daily minimal wear time in order for it to be considered an acceptable day for data analysis is an area of debate (31). Part of this challenge may stem from the difficulty in determining if the zero counts on the accelerometer reflect non-wear time, i.e., the individual removed the unit, or it was due simply due to physical inactivity. However, multisensor activity tracking devices, such as SenseWear Armband, are able to measure skin and/or body temperature along with physical activity (32). This can be used to ease the burden in determining actual wear time. Masse and colleagues computed non-wear time using different criteria (20-60 min of consecutive zero counts), and used these data to compute wear time of the accelerometer (17). Their analysis of data showed that when non-wear time was computed using a greater duration of consecutive zeros (60 min vs. 20 min), the duration of activity bouts computed from wear-time was shorter. In contrast, we found no difference in duration of activity bouts, number of activity bouts per day, or intensity of the activity bouts when non-wear time was computed using either 20, 30, or 60 consecutive minutes of zero counts on the accelerometer (see Table 2). This suggests study cohorts and their activity levels may influence the criteria to choose for data reduction. The cohort in the current work was older and more diseased, as well as less active than that used by Masse and colleagues (17). Considering current findings and previous research in this area, data reduction criteria used in accelerometry assessment warrants continued attention.

Previous reports in the literature have also shown a range in wear time of 1-16 h/day for data to be used for analysis of physical activity (27,33,34). Furthermore, a method that has been proposed is that minimal wear time should be defined as 80% of a standard day, with a standard day being the length of time in which 70% of the study participants wore the monitor, also known as the 80/70 rule (17). Young et al., found in a cohort of over 1,600 obese and overweight adults that 82% of the participants wore their accelerometers for at least 10 h/day (35). For the present study, the 80/70 rule reflects ∼10 h/day, which is consistent with the criteria commonly reported in the adult literature (17). Our study showed no difference in activity patterns when a usable day was defined as 8, 10, or 12 h of wear-time (see Table 2). Moreover, there were negligible differences in the number of subjects defined as meeting these criteria, with only about 30 individuals being dropped as the criteria became more stringent (2,119 vs. 2,150). This suggests that when our participants were instructed to wear the accelerometer for all waking hours, defining usable days as any days that the accelerometer is worn for ≥8, 10, or 12 h appears to provide reliable results with regard to physical activity patterns. However, this result may be due in part to the low level of physical activity in this cohort. One technique that has been used to account for wearing the unit for different durations in a day has been to normalize activity patterns for a set duration, commonly a 12-h day (35). This allows for comparisons of activity for the same time interval; however, it also assumes that each time frame of the day has similar activity patterns. That is, the time the unit is not worn is identical in activity to the time when the unit is worn.

The RT3 is to be worn at the waist attached to a belt or waistband of clothes. However, some devices are gaining popularity because they can be worn on the wrist similar to a watch or bracelet and do not require special clothing. These have been validated and shown to provide estimates of physical activity patterns and energy expenditure (36). Some accelerometers are also waterproof and can be worn 24 h a day without needing to be removed and transferred to other clothes. Taken together, technology has advanced to ease their wearing, lessen burden and improve activity measurements in water activities, thus facilitating long-term recordings.

Allowing a 1- or 2-min interruption within a bout of physical activity increased the number and the average intensity of moderate intensity bouts lasting at least 10 min. This supports findings by Masse et al. who showed a significant decrease in the number of moderate to vigorous bouts of physical activity with no interruption compared to a 1- or 2-min break (17). It is reasonable to consider allowing this interruption since a drop in activity ensues when stopping to take a drink or waiting for a traffic light to change. Since standard accelerometer software is not programmed to make such adjustments, with a large sample size or a high number of days being monitored, data processing programs need to be developed to make it practical and manageable to process accelerometer data files to account for these breaks.

Operationally defining minimal wear time for a day and minimum number of days the unit needs to be worn has important implications for compliance and overall study costs. The burden to the participant must also be considered as wearing the unit longer each day and for more days is demanding and cumbersome. Previous work in healthy adults (age = 45 years) showed that at least 3-4 days of wearing are needed to determine activity patterns (37). Others had observed that reliable results for adults were apparent in as few as 3 days (37); however, up to 12 days of wear time was needed to achieve high reliability in other studies (38). A wider range for number of days has been demonstrated for children, varying from 4 to 9 days (39). Our current findings indicate that increasing the minimum number of wear days from 4, to 5, to 6 did not result in significantly different physical activity patterns in our population (see Table 3). It is not uncommon to observe differences between weekend and weekdays with regard to physical activity patterns (35,37). Since individuals typically have different work and leisure routines between weekend and weekdays, it would be reasonable for patterns and energy expenditure from physical activity to differ among the days. This present study showed longer bout durations (∼1 min longer for each intensity level) for weekend days vs. weekdays for bouts of ≥3 METs for ≥10 min or ≥6 METs for ≥10 min (see Table 4). Furthermore, Matthews et al. reported that physical inactivity, as determined by total daily counts from accelerometer wear, was lower on weekend days compared to weekdays (37), with a difference of ∼30-45 min/day between weekend and weekdays on time spent in physical inactivity.

Altering the criteria for the various parameters involved in data reduction of accelerometer files may not only influence results for physical activity patterns, but the number of eligible files analyzed, i.e., sample size, can differ. Whereas some differences were minimal (∼5%), other criteria showed a more dramatic effect (>15%). Depending on the study aims, this could influence statistical power, or require additional resources for increasing the number of participants to be studied.

In light of our findings and earlier work by others, we recommend that future studies systematically make considerations for data reduction before their analysis. Based on each study's specific aims, deliberations should include defining wear time, minimal daily wear time, number of days required, use of weekend and weekdays, and whether to allow 1- or 2-min interruptions for determining bouts of exercise. Furthermore, the age, health status, activity level, and other factors may influence the data reduction criteria being used. Processing of physical activity data should specify the characteristics of activity bouts, including the length and intensity (e.g., ≥10 min; ≥3 METS) for describing physical activity and their patterns. Acceptance for the specific criteria used in data reduction (e.g., minimum of 4 vs. 5 vs. 6 days of data) should consider the change in sample size and data available for analysis, impact and interpretation of these values, feasibility of achieving the criteria, and participant burden. Stringent requirements may be desired, such as requiring the wear time to be 16 h/day; however, these may present undue burden to the participant with little to no impact on the results. Whereas it is recognized that not all studies that use accelerometers have the resources to systematically evaluate the data as suggested, researchers should be aware of the data limitations, and these need to be addressed in the publication.

It is recognized that the current participants could be considered homogenous in that they are all overweight or obese, are sedentary, and have T2DM. Their low physical activity suggests there is less variability as they have few bouts of moderate and vigorous activity, the intensity of the bouts is low, and the length of their exercise bouts is short as compared to a nonobese group. Although the lower variability may have impacted the results, this does not diminish from the purpose of the paper being to establish a systematic procedure for data reduction with accelerometry data. The findings from the analysis may also be influenced by not having standardized calibration for the units among the various clinic sites. In two previous studies (24,25) that examined the intra- and interunit variability of the RT3s, they found good intraunit reliability for activity counts, but there was concern regarding interunit reliability. Krasnoff et al. observed an inter-unit coefficient of variability range of 9.5-37.4% (24), whereas Reneman et al. reported an intraclass correlation of 0.75 with a 95% confidence interval of 0.46-0.95 (25). Currently, no studies have examined reliability with these devices using energy expenditure or METS outcomes. Although the lack of calibration is a critical issue in the data, it is likely that is not as significant for this work as the purpose of the current analysis was to create a systematic approach for data reduction,

The purpose of this study was to demonstrate a systematic analysis for establishing criteria for data reduction and how these outcomes can be interpreted. This was performed by comparing the effect of using various criteria to define periods of physical activity from accelerometry data in a large sample of obese and overweight subjects with T2DM at baseline prior to randomization into a clinical trial. Reviews of previous work, along with the results from our study demonstrate that interpreting findings can be influenced by how criteria for defining key issues in data reduction of accelerometer data are made. The specific criteria utilized may be dependent on characteristics of the population, including age, gender, and activity level, among others. Given the lack of standards accelerometry data reduction (17), these analyses demonstrate how different analysis criteria can be employed to test the best methods for analyzing and interpreting accelerometer data. In the absence of standardized procedures, data analysis for each study needs to tailored for the group being studied, as well as the specific purpose of the study. Moreover, this may provide an opportunity to standardize the criteria used for accelerometry data to define physical activity across studies and study populations.

These findings are critical to the literature in that they were evolved from a large cohort of overweight men and women in the United States with T2DM, a much understudied population with regards to physical activity behaviors. As the number of researchers studying physical activity in obese populations with T2DM continues to grow, our results will be valuable for their decision making when it comes to including accelerometry in large-scale clinical trials and the processing of these accelerometer data. Finally, the results of this paper will provide guidance and be used as a major reference source for publications of many subsequent investigations from Look AHEAD that involve the accelerometry data.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References

Clinical sites

The Johns Hopkins Medical Institutions: Frederick L. Brancati, MD, MHS1; Jeff Honas, MS2; Lawrence Cheskin, MD3; Jeanne M. Clark, MD, MPH3; Kerry Stewart, EdD3; Richard Rubin, PhD3; Jeanne Charleston, RN; Kathy Horak, RD.

Pennington Biomedical Research Center: George A. Bray, MD1; Kristi Rau2; Allison Strate, RN2; Brandi Armand, LPN2; Frank L. Greenway, MD3; Donna H. Ryan, MD3; Donald Williamson, PhD3; Amy Bachand; Michelle Begnaud; Betsy Berhard; Elizabeth Caderette; Barbara Cerniauskas; David Creel; Diane Crow; Helen Guay; Nancy Kora; Kelly LaFleur; Kim Landry; Missy Lingle; Jennifer Perault; Mandy Shipp, RD; Marisa Smith; Elizabeth Tucker.

The University of Alabama at Birmingham: Cora E. Lewis, MD, MSPH1; Sheikilya Thomas MPH2; Monika Safford, MD3; Vicki DiLillo, PhD; Charlotte Bragg, MS, RD, LD; Amy Dobelstein; Stacey Gilbert, MPH; Stephen Glasser, MD; Sara Hannum, MA; Anne Hubbell, MS; Jennifer Jones, MA; DeLavallade Lee; Ruth Luketic, MA, MBA, MPH; Karen Marshall; L. Christie Oden; Janet Raines, MS; Cathy Roche, RN, BSN; Janet Truman; Nita Webb, MA; Audrey Wrenn, MAEd.

Harvard Center

Massachusetts General Hospital: David M. Nathan, MD1; Heather Turgeon, RN, BS, CDE2; Kristina Schumann, BA2; Enrico Cagliero, MD3; Linda Delahanty, MS, RD3; Kathryn Hayward, MD3;Ellen Anderson, MS, RD3; Laurie Bissett, MS, RD; Richard Ginsburg, PhD; Valerie Goldman, MS, RD; Virginia Harlan, MSW; Charles McKitrick, RN, BSN, CDE; Alan McNamara, BS; Theresa Michel, DPT, DSc CCS; Alexi Poulos, BA; Barbara Steiner, EdM; Joclyn Tosch, BA.

Joslin Diabetes Center: Edward S. Horton, MD1; Sharon D. Jackson, MS, RD, CDE2; Osama Hamdy, MD, PhD3; A. Enrique Caballero, MD3; Sarah Bain, BS; Elizabeth Bovaird, BSN, RN; Ann Goebel-Fabbri, PhD; Lori Lambert, MS, RD; Sarah Ledbury, MEd, RD; Maureen Malloy, BS; Kerry Ovalle, MS, RCEP, CDE.

Beth Israel Deaconess Medical Center: George Blackburn, MD, PhD1; Christos Mantzoros, MD, DSc3; Kristinia Day, RD; Ann McNamara, RN.

University of Colorado Health Sciences Center: James O. Hill, PhD1; Marsha Miller, MS, RD2; JoAnn Phillipp, MS2; Robert Schwartz, MD3; Brent Van Dorsten, PhD3; Judith Regensteiner, PhD3; Salma Benchekroun MS; Ligia Coelho, BS; Paulette Cohrs, RN, BSN; Elizabeth Daeninck, MS, RD; Amy Fields, MPH; Susan Green; April Hamilton, BS, CCRC; Jere Hamilton, BA; Eugene Leshchinskiy; Michael McDermott, MD; Lindsey Munkwitz, BS; Loretta Rome, TRS; Kristin Wallace, MPH; Terra Worley, BA.

Baylor College of Medicine: John P. Foreyt, PhD1; Rebecca S. Reeves, DrPH, RD2; Henry Pownall, PhD3; Ashok Balasubramanyam, MBBS3; Peter Jones, MD3; Michele Burrington, RD; Chu-Huang Chen, MD, PhD; Allyson Clark, RD; Molly Gee, MEd, RD; Sharon Griggs; Michelle Hamilton; Veronica Holley; Jayne Joseph, RD; Patricia Pace, RD: Julieta Palencia, RN; Olga Satterwhite, RD; Jennifer Schmidt; Devin Volding, LMSW; Carolyn White.

University of California at Los Angeles School of Medicine: Mohammed F. Saad, MD1; Siran Ghazarian Sengardi, MD2; Ken C. Chiu, MD3; Medhat Botrous; Michelle Chan, BS; Kati Konersman, MA, RD, CDE; Magpuri Perpetua, RD.

The University of Tennessee Health Science Center

University of Tennessee East: Karen C. Johnson, MD, MPH1; Carolyn Gresham, RN2; Stephanie Connelly, MD, MPH3; Amy Brewer, RD, MS; Mace Coday, PhD; Lisa Jones, RN; Lynne Lichtermann, RN, BSN; Shirley Vosburg, RD, MPH; and J. Lee Taylor, MEd, MBA.

University of Tennessee Downtown: Abbas E. Kitabchi, PhD, MD1; Helen Lambeth, RN, BSN2; Debra Clark, LPN; Andrea Crisler, MT; Gracie Cunningham; Donna Green, RN; Debra Force, MS, RD, LDN; Robert Kores, PhD; Renate Rosenthal PhD; Elizabeth Smith, MS, RD, LDN; and Maria Sun, MS, RD, LDN; and Judith Soberman, MD3.

University of Minnesota: Robert W. Jeffery, PhD1; Carolyn Thorson, CCRP2; John P. Bantle, MD3; J. Bruce Redmon, MD3; Richard S. Crow, MD3; Scott Crow, MD3; Susan K Raatz, PhD, RD3; Kerrin Brelje, MPH, RD; Carolyne Campbell;

Jeanne Carls, MEd; Tara Carmean-Mihm, BA; Emily Finch, MA; Anna Fox, MA;

Elizabeth Hoelscher, MPH, RD, CHES; La Donna James; Vicki A. Maddy, BS, RD; Therese Ockenden, RN; Birgitta I. Rice, MS, RPh CHES; Tricia Skarphol, BS; Ann D. Tucker, BA; Mary Susan Voeller, BA; Cara Walcheck, BS, RD.

St Luke's Roosevelt Hospital Center: Xavier Pi-Sunyer, MD1; Jennifer Patricio, MS2; Stanley Heshka, PhD3; Carmen Pal, MD3; Lynn Allen, MD; Diane Hirsch, RNC, MS, CDE; Mary Anne Holowaty, MS, CN.

University of Pennsylvania: Thomas A. Wadden, PhD1; Barbara J. Maschak-Carey, MSN, CDE2; Stanley Schwartz, MD3; Gary D. Foster, PhD3; Robert I. Berkowitz, MD3; Henry Glick, PhD3; Shiriki K. Kumanyika, PhD, RD, MPH3; Johanna Brock; Helen Chomentowski; Vicki Clark; Canice Crerand, PhD; Renee Davenport; Andrea Diamond, MS, RD; Anthony Fabricatore, PhD; Louise Hesson, MSN; Stephanie Krauthamer-Ewing, MPH; Robert Kuehnel, PhD; Patricia Lipschutz, MSN; Monica Mullen, MS, RD; Leslie Womble, PhD, MS; Nayyar Iqbal, MD.

University of Pittsburgh: David E. Kelley, MD1; Jacqueline Wesche-Thobaben, RN, BSN, CDE2; Lewis Kuller, MD, DrPH3; Andrea Kriska, PhD3; Janet Bonk, RN, MPH; Rebecca Danchenko, BS; Daniel Edmundowicz, MD3; Mary L. Klem, PhD, MLIS3; Monica E. Yamamoto, DrPH, RD, FADA3; Barb Elnyczky, MA; George A. Grove, MS; Pat Harper, MS, RD, LDN; Janet Krulia, RN,BSN,CDE; Juliet Mancino, MS, RD, CDE, LDN; Anne Mathews, MS, RD, LDN; Tracey Y. Murray, BS; Joan R. Ritchea; Jennifer Rush, MPH; Karen Vujevich, RN-BC, MSN, CRNP; Donna Wolf, MS.

The Miriam Hospital/Brown Medical School: Rena R. Wing, PhD1; Renee Bright, MS2; Vincent Pera, MD3; John Jakicic, PhD3; Deborah Tate, PhD3; Amy Gorin, PhD3; Kara Gallagher, PhD3; Amy Bach, PhD; Barbara Bancroft, RN, MS; Anna Bertorelli, MBA, RD; Richard Carey, BS; Tatum Charron, BS; Heather Chenot, MS; Kimberley Chula-Maguire, MS; Pamela Coward, MS, RD; Lisa Cronkite, BS; Julie Currin, MD; Maureen Daly, RN; Caitlin Egan, MS; Erica Ferguson, BS, RD; Linda Foss, MPH; Jennifer Gauvin, BS; Don Kieffer, PhD; Lauren Lessard, BS; Deborah Maier, MS; JP Massaro, BS; Tammy Monk, MS; Rob Nicholson, PhD; Erin Patterson, BS; Suzanne Phelan, PhD; Hollie Raynor, PhD, RD; Douglas Raynor, PhD; Natalie Robinson, MS, RD; Deborah Robles; Jane Tavares, BS.

The University of Texas Health Science Center at San Antonio: Steven M. Haffner, MD1; Maria G. Montez, RN, MSHP, CDE2; Carlos Lorenzo, MD3.

University of Washington/VA Puget Sound Health Care System: Steven Kahn MB, ChB1; Brenda Montgomery, RN, MS, CDE2; Robert Knopp, MD3; Edward Lipkin, MD3; Matthew L. Maciejewski, PhD3; Dace Trence, MD3; Terry Barrett, BS; Joli Bartell, BA; Diane Greenberg, PhD; Anne Murillo, BS; Betty Ann Richmond, MEd; April Thomas, MPH, RD.

Southwestern American Indian Center, Phoenix, Arizona and Shiprock, New Mexico: William C. Knowler, MD, DrPH1; Paula Bolin, RN, MC2; Tina Killean, BS2; Cathy Manus, LPN3; Jonathan Krakoff, MD3; Jeffrey M. Curtis, MD, MPH3; Justin Glass, MD3; Sara Michaels, MD3; Peter H. Bennett, MB, FRCP3; Tina Morgan3; Shandiin Begay, MPH; Bernadita Fallis RN, RHIT, CCS; Jeanette Hermes, MS,RD; Diane F. Hollowbreast; Ruby Johnson; Maria Meacham, BSN, RN, CDE; Julie Nelson, RD; Carol Percy, RN; Patricia Poorthunder; Sandra Sangster; Nancy Scurlock, MSN, ANP-C, CDE; Leigh A. Shovestull, RD, CDE; Janelia Smiley; Katie Toledo, MS, LPC; Christina Tomchee, BA; Darryl Tonemah PhD.

University of Southern California: Anne Peters, MD1; Valerie Ruelas, MSW, LCSW2; Siran Ghazarian Sengardi, MD2; Kathryn Graves, MPH, RD, CDE; Kati Konersman, MA, RD, CDE; Sara Serafin-Dokhan.

Coordinating Center

Wake Forest University: Mark A. Espeland, PhD1; Judy L. Bahnson, BA2; Lynne Wagenknecht, DrPH3; David Reboussin, PhD3; W. Jack Rejeski, PhD3; Alain Bertoni, MD, MPH3; Wei Lang, PhD3; Gary Miller, PhD3; David Lefkowitz, MD3; Patrick S. Reynolds, MD3; Paul Ribisl, PhD3; Mara Vitolins, DrPH3; Michael Booth, MBA2; Kathy M. Dotson, BA2; Amelia Hodges, BS2; Carrie C. Williams, BS2; Jerry M. Barnes, MA; Patricia A. Feeney, MS; Jason Griffin, BS; Lea Harvin, BS; William Herman, MD, MPH; Patricia Hogan, MS; Sarah Jaramillo, MS; Mark King, BS; Kathy Lane, BS; Rebecca Neiberg, MS; Andrea Ruggiero, MS; Christian Speas, BS; Michael P. Walkup, MS; Karen Wall, AAS; Michelle Ward; Delia S. West, PhD; Terri Windham.

Central Resources Centers

DXA Reading Center, University of California at San Francisco Michael Nevitt, PhD1; Susan Ewing, MS; Cynthia Hayashi; Jason Maeda, MPH; Lisa Palermo, MS, MA; Michaela Rahorst; Ann Schwartz, PhD; John Shepherd, PhD.

Central Laboratory, Northwest Lipid Research Laboratories: Santica M. Marcovina, PhD, ScD1; Greg Strylewicz, MS.

ECG Reading Center, EPICARE, Wake Forest University School of Medicine: RonaldJ. Prineas, MD, PhD1; Teresa Alexander; Lisa Billings; Charles Campbell, AAS, BS; Sharon Hall; Susan Hensley; Yabing Li, MD; Zhu-Ming Zhang, MD.

Diet Assessment Center, University of South Carolina, Arnold School of Public Health, Center for Research in Nutrition and Health Disparities: Elizabeth J Mayer-Davis, PhD1; Robert Moran, PhD.

Hall-Foushee Communications, Inc: Richard Foushee, PhD; Nancy J. Hall, MA.

Federal sponsors

National Institute of Diabetes and Digestive and Kidney Diseases: Barbara Harrison, MS; Van S. Hubbard, MD PhD; Susan Z. Yanovski, MD.

National Heart, Lung, and Blood Institute: Lawton S. Cooper, MD, MPH; Jeffrey Cutler, MD, MPH; Eva Obarzanek, PhD, MPH, RD.

Centers for Disease Control and Prevention: Edward W. Gregg, PhD; David F. Williamson, PhD; Ping Zhang, PhD. 1 Principal Investigator; 2Program Coordinator; 3Co-Investigator.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References

This study is supported by the Department of Health and Human Services through the following cooperative agreements from the National Institutes of Health: DK57136, DK57149, DK56990, DK57177, DK57171, DK57151, DK57182, DK57131, DK57002, DK57078, DK57154, DK57178, DK57219, DK57008, DK57135, and DK56992. The following federal agencies have contributed support: National Institute of Diabetes and Digestive and Kidney Diseases; National Heart, Lung, and Blood Institute; National Institute of Nursing Research; National Center on Minority Health and Health Disparities; Office of Research on Women's Health; and the Centers for Disease Control and Prevention. This research was supported in part by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases. Additional support was received from The Johns Hopkins Medical Institutions Bayview General Clinical Research Center (M01RR02719); the Massachusetts General Hospital Mallinckrodt General Clinical Research Center (M01RR01066); the University of Colorado Health Sciences Center General Clinical Research Center (M01RR00051) and Clinical Nutrition Research Unit (P30 DK48520); the University of Tennessee at Memphis General Clinical Research Center (M01RR0021140); the University of Pittsburgh General Clinical Research Center (M01RR000056 44) and NIH grant (DK 046204); and the University of Washington/VA Puget Sound Health Care System Medical Research Service, Department of Veterans Affairs; Frederic C. Bartter General Clinical Research Center (M01RR01346). The following organizations have committed to make major contributions to Look AHEAD: FedEx Corporation; Health Management Resources; LifeScan, Inc., a Johnson & Johnson Company; Optifast of Nestle HealthCare Nutrition, Inc.; Hoffmann-La Roche Inc.; Abbott Nutrition; and Slim-Fast Brand of Unilever North America. Disclosure of funding received for this work from: Department of Health and Human Services through the following cooperative agreements from the National Institutes of Health: DK57136, DK57149, DK56990, DK57177, DK57171, DK57151, DK57182, DK57131, DK57002, DK57078, DK57154, DK57178, DK57219, DK57008, DK57135, and DK56992. All other Look AHEAD staffs are listed alphabetically by site.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. Acknowledgements
  9. References
  • 1
    Franco OH, de Laet C, Peeters A et al. Effects of physical activity on life expectancy with cardiovascular disease. Arch Intern Med 2005; 165: 2355-2360.
  • 2
    Warburton DE, Nicol CW, Bredin SS. Health benefits of physical activity: the evidence. CMAJ 2006; 174: 801-809.
  • 3
    Donnelly JE, Blair SN, Jakicic JM et al.; American College of Sports Medicine. American College of Sports Medicine Position Stand. Appropriate physical activity intervention strategies for weight loss and prevention of weight regain for adults. Med Sci Sports Exerc 2009; 41: 459-471.
  • 4
    Jakicic JM, Otto AD. Treatment and prevention of obesity: what is the role of exercise? Nutr Rev 2006; 64: S57-S61.
  • 5
    Hamman RF, Wing RR, Edelstein SL et al. Effect of weight loss with lifestyle intervention on risk of diabetes. Diabetes Care 2006; 29: 2102-2107.
  • 6
    Knowler WC, Barrett-Connor E, Fowler SE et al.; Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 2002; 346: 393-403.
  • 7
    Wing RR;Look AHEAD Research Group. Long-term effects of a lifestyle intervention on weight and cardiovascular risk factors in individuals with type 2 diabetes mellitus: four-year results of the Look AHEAD trial. Arch Intern Med 2010; 170: 1566-1575.
  • 8
    Eriksson J, Lindström J, Valle T et al. Prevention of Type II diabetes in subjects with impaired glucose tolerance: the Diabetes Prevention Study (DPS) in Finland. Study design and 1-year interim report on the feasibility of the lifestyle intervention programme. Diabetologia 1999; 42: 793-801.
  • 9
    Pan XR, Li GW, Hu YH et al. Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study. Diabetes Care 1997; 20: 537-544.
  • 10
    Thomas DE, Elliott EJ, Naughton GA. Exercise for type 2 diabetes mellitus. Cochrane Database Syst Rev 2006; CD002968.
  • 11
    Seyoum B, Estacio RO, Berhanu P, Schrier RW. Exercise capacity is a predictor of cardiovascular events in patients with type 2 diabetes mellitus. Diab Vasc Dis Res 2006; 3: 197-201.
  • 12
    Weltman NY, Saliba SA, Barrett EJ, Weltman A. The use of exercise in the management of type 1 and type 2 diabetes. Clin Sports Med 2009; 28: 423-439.
  • 13
    Epstein LH, Paluch RA, Coleman KJ, Vito D, Anderson K. Determinants of physical activity in obese children assessed by accelerometer and self-report. Med Sci Sports Exerc 1996; 28: 1157-1164.
  • 14
    Welk GJ, Corbin CB, Dale D. Measurement issues in the assessment of physical activity in children. Res Q Exerc Sport 2000; 71: S59-S73.
  • 15
    Maddison R, Jiang Y, Hoorn SV et al. Estimating energy expenditure with the RT3 triaxial accelerometer. Res Q Exerc Sport 2009; 80: 249-256.
  • 16
    Westerterp KR. Assessment of physical activity: a critical appraisal. Eur J Appl Physiol 2009; 105: 823-828.
  • 17
    Mâsse LC, Fuemmeler BF, Anderson CB et al. Accelerometer data reduction: a comparison of four reduction algorithms on select outcome variables. Med Sci Sports Exerc 2005; 37: S544-S554.
  • 18
    Troiano RP, Berrigan D, Dodd KW et al. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc 2008; 40: 181-188.
  • 19
    Healy GN, Matthews CE, Dunstan DW, Winkler EA, Owen N. Sedentary time and cardio-metabolic biomarkers in US adults: NHANES 2003-06. Eur Heart J 2011; 32: 590-597.
  • 20
    Tudor-Locke C, Brashear MM, Johnson WD, Katzmarzyk PT. Accelerometer profiles of physical activity and inactivity in normal weight, overweight, and obese U.S. men and women. Int J Behav Nutr Phys Act 2010; 7: 60.
  • 21
    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: 875-881.
  • 22
    Jakicic JM, Gregg E, Knowler W et al. Activity patterns of obese adults with type 2 diabetes in the look AHEAD study. Med Sci Sports Exerc 2010; 42: 1995-2005.
  • 23
    Ryan DH, Espeland MA, Foster GD et al.; Look AHEAD Research Group. Look AHEAD (Action for Health in Diabetes): design and methods for a clinical trial of weight loss for the prevention of cardiovascular disease in type 2 diabetes. Control Clin Trials 2003; 24: 610-628.
  • 24
    Krasnoff JB, Kohn MA, Choy FK et al. Interunit and intraunit reliability of the RT3 triaxial accelerometer. J Phys Act Health 2008; 5: 527-538.
  • 25
    Reneman M, Helmus M. Interinstrument reliability of the RT3 accelerometer. Int J Rehabil Res 2010; 33: 178-179.
  • 26
    Cradock AL, Wiecha JL, Peterson KE et al. Youth recall and TriTrac accelerometer estimates of physical activity levels. Med Sci Sports Exerc 2004; 36: 525-532.
  • 27
    Treuth MS, Sherwood NE, Baranowski T et al. Physical activity self-report and accelerometry measures from the Girls health Enrichment Multi-site Studies. Prev Med 2004; 38 Suppl: S43-S49.
  • 28
    Haskell WL, Lee IM, Pate RR et al. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Med Sci Sports Exerc 2007; 39: 1423-1434.
  • 29
    Rowlands AV, Thomas PW, Eston RG, Topping R. Validation of the RT3 triaxial accelerometer for the assessment of physical activity. Med Sci Sports Exerc 2004; 36: 518-524.
  • 30
    Ribisl PM, Lang W, Jaramillo SA et al.; Look AHEAD Research Group. Exercise capacity and cardiovascular/metabolic characteristics of overweight and obese individuals with type 2 diabetes: the Look AHEAD clinical trial. Diabetes Care 2007; 30: 2679-2684.
  • 31
    Ward DS, Evenson KR, Vaughn A, Rodgers AB, Troiano RP. Accelerometer use in physical activity: best practices and research recommendations. Med Sci Sports Exerc 2005; 37: S582-S588.
  • 32
    St-Onge M, Mignault D, Allison DB, Rabasa-Lhoret R. Evaluation of a portable device to measure daily energy expenditure in free-living adults. Am J Clin Nutr 2007; 85: 742-749.
  • 33
    Pate RR, Pfeiffer KA, Trost SG, Ziegler P, Dowda M. Physical activity among children attending preschools. Pediatrics 2004; 114: 1258-1263.
  • 34
    Samuel-Hodge CD, Fernandez LM, Henríquez-Roldán CF, Johnston LF, Keyserling TC. A comparison of self-reported energy intake with total energy expenditure estimated by accelerometer and basal metabolic rate in African-American women with type 2 diabetes. Diabetes Care 2004; 27: 663-669.
  • 35
    Young DR, Jerome GJ, Chen C, Laferriere D, Vollmer WM. Patterns of physical activity among overweight and obese adults. Prev Chronic Dis 2009; 6: A90.
  • 36
    Ekblom O, Nyberg G, Ekblom Bak E, Ekelund U, Marcus C. Validity and Comparability of a Wrist-Worn Accelerometer in Children. J Phys Act Health 2011.
  • 37
    Matthews CE, Ainsworth BE, Thompson RW, Bassett DR Jr. Sources of variance in daily physical activity levels as measured by an accelerometer. Med Sci Sports Exerc 2002; 34: 1376-1381.
  • 38
    Levin S, Jacobs DR Jr, Ainsworth BE, Richardson MT, Leon AS. Intra-individual variation and estimates of usual physical activity. Ann Epidemiol 1999; 9: 481-488.
  • 39
    Trost SG, Pate RR, Freedson PS, Sallis JF, Taylor WC. Using objective physical activity measures with youth: how many days of monitoring are needed? Med Sci Sports Exerc 2000; 32: 426-431.