Comparison of Wrist Worn to Waist Worn Actigraph Physical Activity Data in Hungarian Adults

The pattern of physical activity in Hungarian adults might provide information on potential obesity prevention strategies. In preparation for a region assessment of physical activity (PA), placement of an activity monitor were studied in leu of the changes that the NHANES protocol has in the location of the ActiGraph monitors from the waist to wrist. Although the cut-off points and interpreted energy levels are expected to be different, if the two locations are related, then placement on the body might still be able to determine patterns of activity and inactivity.

Purpose To test the relationship between accelerometer counts in wrist and waist (hip) worn ActiGraph activity monitors.

Methods Two ActiGraph GT3X+ accelerometers were placed on the nondominant wrist and on the waist over the right hip on a sample of Hungarian adults. Each accelerometer was worn at the same time for three typical work days in the fall 2012. One 24 hour period (second day) was separated out and one minute epoch was downloaded for interpretation. The non-sleeping hours were used to correlate counts between monitors. In addition, minutes in moderate and vigorous physical activity (MVPA) were determined using vector magnitude cut points incorporated in the ActiLife 6.4.3 software from Freedson Adult VM3 2011. Also estimated total PA calories were determined using Freedson VM3 Combination 2011 energy conversions for each location. Pearson product correlations were used to test relationships and descriptive analysis of means were reported on each of the variables tested.

Results Data from 25 subjects ranging in ages from 18-34 years were used. The intra-subject correlations coefficients ranged from 0.55-0.92 with a mean (SD) of 0.75 (0.11). The correlation between minutes MVPA and total PA calories were 0.68 and 0.70 respectively. The mean (SD) difference between wrist and waist for the minutes MVPA were 220.4 (65.0); higher in the wrist; while total PA calories differed on average 1789 (887.4) more calories for the wrist worn device.

Conclusions The strong to moderate relationships between these two locations suggest that patterns of PA would be similar however due to the large differences in actual minutes MVPA and energy outputs, new cut-off points and energy estimation equations for wrist worn data are needed.