Background/Purpose: Few instruments that measure
physical activity (PA) can accurately quantify PA performed at light and
moderate intensities, which is particularly relevant to older adults with OA of
the lower extremities. However, evidence of their ability to capture change
over time in PA is limited. Also, responsiveness has not been compared
across instruments that measure PA. Such investigation would allow for a
well-educated choice of tools to assess changes in PA behavior over time. Our
aim was to determine and compare the responsiveness of the Actigraph (ACT), Sensewear Armband (SWA) and Community Healthy Activities
Model Program for Seniors (CHAMPS) questionnaire in assessing PA from light to
moderate intensities after an exercise intervention in individuals following
total knee arthroplasty (TKA).

Methods: Baseline
and 6 months data on PA from subjects participating in an intervention to
promote PA were analyzed. Internal (distribution based) and external (anchor based)
responsiveness were assessed. Changes in duration in light and moderate intensity
PA from baseline to 6 months and the standardized response mean (SRM) were calculated to assess internal responsiveness, and
were compared across instruments. External responsiveness was assessed by
comparing changes in PA to changes perceived by the subjects using the global
rating of change in PA that was administered at 6 months. In addition, agreement
between instruments on identifying subjects who were less, the same or more
active based on standard error of the measurement was performed using weighted
Kappa.

Results: Thirty
subjects (67±6 years old, 73% female and obese [BMI=30±4 kg/m²])
were included in the analysis. Changes in PA measured by each instrument after the
intervention were small and did not reach statistical
significance (Table). SRMs indicated low degree of
responsiveness (SRM < 0.30) across PA intensity
levels measured by each instrument. Based on the external anchor, all subjects
reported being more active after the intervention, while the distribution of the
changes in PA showed that some of them were less active. Using the measurement
error as a threshold for change in PA, the ACT identified 37% of subjects as
less active, 33% as more active, and 30% who stayed the same. Data from SWA
identified 30%, 37% and 33% respectively. Data from CHAMPS identified 40%, 37%
and 23% respectively. The ACT and SWA agreed on identifying changes beyond
error in moderate PA (Kappa = 0.60) and number of steps (Kappa = 0.63). There
was no agreement between CHAMPS and the activity monitors (Kappa ² 0.22).

Conclusion: Distribution-based
method may not be appropriate to investigate changes in PA after an
intervention since there were no changes on a group basis. However, as
individuals, we observed relevant changes in PA. Using the measurement error as
a threshold for changes in PA may be useful since it allowed for identifying
those who became more or less active after an intervention.