Accelerometry (ancillary studies 1-4)

Accelerometry (ancillary studies 1-4)

Contact: Marjolein Visser

Background
Accelerometry is an objective method to assess physical activity reflecting a relative measure of change in momentum. In LASA, we used an Actigraph accelerometer in 4 different ancillary accelerometry studies.

1. Accelerometry Study 1 was conducted as part of the Lifestyle Study, a substudy conducted within LASA in 2007. For this study Actigraph model GT1M (Manufacturing Technologies Inc., Fort Walton Beach, FL was used.
2. Accelerometry Study 2 was a LASA ancillary study conducted in 2010 and aimed to develop a questionnaire for assessing sedentary behavior in older persons. In that study model GT3X was used to objectively measure physical (in)activity.
3. Accelerometry Study 3 was conducted as part of the overarching EPOSA project.
4. Accelerometry Study 4 was conducted between 2015-2016 during the regular I-wave. The GT3X Actigraph was used to objectively measure physical activity intensity categories.

Brief descriptions of these studies are provided below.

Actigraph models
The GT1M accelerometer is a uniaxial monitor (vertical plane) that measures 5.1×3.8×1.5 cm, is lightweight (42 g) and powered by a 2430 coin cell lithium battery. The GT3X model is a triaxial accelerometer (Actigraph Inc., Pensacola, FL). The monitors integrate accelerations and decelerations in one, vertical plane (GT1M) or three planes (GT3X) via piezoelectric plates.

Data collection
The accelerometer together with an instruction brochure, which included photographs of how to properly wear the accelerometer was send to the participants by regular mail. The accelerometers were attached to a 3 cm wide, tight elastic belt and was worn superior to the left iliac crest. After two days a phone call was made to ensure that the package was received and the accelerometer was properly worn by the participants. Participants were instructed to wear the accelerometer for a 7-day period during waking hours and only to remove the accelerometer during bathing, showering and swimming. In study 1 the activity was recorded using 1-minute intervals (=epochs), in study 2 using 1-second intervals. Participants also completed a daily log to record the time the accelerometer was put on right after waking and removed just before going to bed. When the accelerometer was not worn for some period during the day, the participants recorded the start and end time of that period as well as the activity performed (e.g. showering or swimming).

STUDY 1

Study sample
Participants who participated in the measurement cycle 2005-2006, and who were <80 years of age, living independently, having a good cognitive status (MMSE score>23), and known to be alive on January 15, 2007 were send an extensive lifestyle questionnaire in February 2007 (N=1,421). Of the 1,058 participants (response 74.5%) who returned a completed questionnaire, 647 participants indicated to be willing to participate in an additional accelerometry study. A random sample of 169 participants was selected to participate in this study. Of those, 153 participants completed the study (between March 12 and July 5, 2007), 12 participants refused (mainly because of acute health problems), and 4 participants could not participate due to being abroad.

Data reduction
Participants’ logs were checked for periods when the actigraph was not worn and matched against actigraph data. The accelerometry data were analyzed in MATLAB (The MathWorks, Natick MA, USA). Data periods with zero counts for 60 min or more were considered invalid. Days with fewer than 10 h per day of valid data were set to missing (data were not considered representative of the total day). A total of 14 participants with fewer than 4 days of valid data were excluded (data were not considered representative of a usual week), as well as one participant who used crutches during the study period because of a recent knee operation. Valid accelerometry data from 138 participants are available for data analysis.

Available data for each measurement day based on the valid time in study 1

Use of simultaneous activity logs
If a participant did not wear the accelerometer during some time during waking hours, the information from the activity log (type of activity) can be used to impute this time. For example, 30 minutes of swimming can be added to the time spent on moderate and vigorous physical activity (≥ 3 METs).

STUDY 2

Study sample
For the study a random sample of 100 men and 100 women who participated in the measurement cycle in 2008-2009 and were aged 65 and older was selected. In April 2010, 130 persons were sent an information letter and were contacted by phone to ask for their participation in a sub study on sedentary activity. Of those, 93 agreed to participate in the study and provided information. Reasons for non-response were death (n=2), refused (n=17), no contact (n=8), too sick or frail (n=5), holiday (n=4), and unknown (n=1). Accelerometry data were obtained from 91 participants.

Data reduction
The data were processed using customized software written in MATLAB R2006a (The MathWorks, Inc., Natick, MA). Data periods of 60 minutes or more with zero counts on all three axes simultaneously were considered as non-wear time and were excluded from the analyses. Data files with fewer than 10 hours per day of wear time were also excluded. The total objective sedentary time per day was assessed using the <100 counts per minute cut point based on the vertical axis.

STUDY 3

Study sample
Between January and November 2012, a random sample of 334 participants from 574 eligible participants from the European Project on OSteoArthritis baseline study (EPOSA; www.eposa.org) were asked to wear an Actigraph triaxial accelerometer (Model GT3X; Actigraph, Pensacola, FL, USA) for an 8-day period. Of all invited individuals, 299 individuals agreed. In total, accelerometry data could be analyzed for 281 respondents.

Data reduction
The accelerometry data were processed using ActiLife Data Analysis software (version 6.10.4) (ActiGraph, Pensacola, FL, USA). Physical activity (PA) was collected using 1-second epochs and were aggregated to 60-seconds epoch for data reduction. Non-wear time was defined by an interval of at least 60 consecutive minutes of zero activity counts on the y-axis, with allowance for 1-2 minutes of counts between 0 and 100 on this axis. Wear time was determined by subtracting non-wear time from 24 hours. Physical activity was measured as the mean time spent on PA in minutes per day. Furthermore, PA was measured as the daily mean time spent in separate PA intensity categories. The accelerometry data file includes the following variables:

STUDY 4

Study Sample
Between October 2015 and December 2016, all 1,770 LASA respondents who participated in the main interview were invited for the accelerometry ancillary study.  A total of 1,412 participants were sent an Actigraph tri-axial accelerometer (Model GT3X; Actigraph, Pensacola, FL, USA) for a 7-day period of which 1,218 worn the accelerometer and sent back the accelerometer.  Reasons of non-participations were: acute health problems (n=151), lost accelerometer (n=24), measurement error/broken meter (n=4), too frail (n=2), too busy (n=2), or for unknown reasons (n=11).

Although the participants were instructed to wear the accelerometer only during waking hours, some participants forgot to or experienced difficulties taking off the accelerometer (n=27 compromising 119 days). For these participants the days were indicated and corrected with the sleep information from the questionnaire to calculate adjusted wear time and sedentary time. Adjusted wear time: total minutes per day 1440 – reported sleep time assuming that a participant directly puts on the accelerometer during wake hours. The adjusted sedentary time is calculated as: adjusted  wear time – low light – light high – moderate – vigorous. For these participants sedentary bouts are missing.

Data reduction
The accelerometry data were analyzed with ActiLife 6.13.3 (Actigraph, Pensacola, USA). Physical activity was collected using 1 second epochs and aggregated to 60 second epochs for data reduction. These 60-second epochs were used to assign count values – a relative measure of change in momentum. These counts were translated into an estimate of physical activity intensity based on the total number of counts per minute. Data periods with zero counts for ≥60 min (indicating non-wear time or a period of undetectable movement) were considered invalid, with allowance for 1-2 min of counts between 0 and 100. Total physical activity was measured as the mean time spent on physical activity in minutes per day. Physical activity intensity categories were based on the following accelerometer cut-points:

• Sedentary: <100 counts per minute (lying, sitting)
• Light: ≥ 100 and <2020 counts per minute
• Light-low: ≥ 100 and <760 counts per minute (e.g. light household, and slow walking)
• Light-high: ≥ 760 and <2020 counts per minute (e.g. walking)
• Moderate: ≥ 2020 and <5999  counts per minute (brisk walking, cycling)
• Vigorous ≥5999 counts per minute (running, active sports)

The intensity categories moderate and vigorous physical activity (MVPA) were also used as one separate category.

Non-wear time was defined by an interval of at least 60 consecutive minutes of zero activity counts on the y-axis, with allowance for 1-2 minutes of counts between 0 and 100 on this axis. Wear time was determined by subtracting non-wear time from 24 hours. Physical activity was measured as the mean time spent on PA in minutes per day. Furthermore, PA was measured as the daily mean time spent in separate PA intensity categories and bouts. Detailed information is available for bouts in different physical activity categories.

All measures are per day based on number of valid days.

Previous use in LASA
Accelerometry data from study 1 have been used to investigate the level of correct perception of adherence to the physical activity recommendation by older persons (Visser et al., 2014). The accelerometer results of study 2 were used to developed a questionnaire for sedentary behavior in older persons (Visser et al., 2013). The accelerometry data from study 3 were used to examine the association of objectively measured physical activity with objective characteristics of the neighborhood built environment in older adults with and without osteoarthritis (Timmermans et al., 2016). The accelerometry study from study 4 has been used to cross-sectionally describe objectively assessed physical activity and sedentary behavior across strata of demographic and lifestyle factors in LASA.

References

1. Freedson PS, Melanson E, Sirard J (1998) Calibration of the Computer Science and Applications Inc. accelerometer. Med Sci Sports Exerc 30:777–781.
2. Matthew CE. Calibration of accelerometer output for adults. Med Sci Sports Exerc 2005;37(11 Suppl):S512-22.
3. Timmermans, E.J., Schaap, L.A., Visser, M., Van der Ploeg, H.P., Wagtendonk, A.J., Van der Pas, S., & Deeg, D.J.H. (2016). The association of the neighbourhood built environment with objectively measured physical activity in older adults with and without lower limb osteoarthritis. BMC Public Health. 16:710.
4. Visser, M., Brychta, R.J., Chen, K.Y., Koster, A. (2014). Self-Reported Adherence to the Physical Activity Recommendation and Determinants of Misperception in Older Adults. Journal of Aging and Physical Activity, 22, 226-234.
5. Visser, M., Koster, A. (2013). Development of a questionnaire to assess sedentary time in older persons – a comparative study using accelerometry. BioMed Central Geriatrics, 13, 80.

Date of last update: December 1, 2017 (H. van Ballegooijen)