Agreement between the International Physical Activity Questionnaire and Accelerometry in Adults with Orthopaedic Injury

William Veitch, Rachel E. Climie, Belinda J. Gabbe, David W. Dunstan, Neville Owen, Christina L. EkegrenInternational Journal of Environmental Research and Public Health 2020 · DOI

TL;DR

This study investigated how well the International Physical Activity Questionnaire (IPAQ) aligns with objective accelerometer data for assessing physical activity and sitting time in adults recovering from orthopedic injuries. The findings reveal that the IPAQ tends to overestimate physical activity and underestimate sitting time in this population, highlighting the need for caution when relying solely on self-reported data for this group.

What Problem Does This Paper Address?

Assessing physical activity is crucial for understanding recovery and long-term health outcomes, especially for individuals with orthopedic injuries who often experience reduced activity levels. While self-report questionnaires like the IPAQ are widely used due to their ease of administration, their accuracy in specific clinical populations, like those with orthopedic injuries, needs validation. These individuals have unique recovery trajectories and potential limitations that might influence how they perceive and report their activity. This paper addresses the gap in understanding the agreement between the commonly used IPAQ and objective accelerometer-based measures in this specific patient group, aiming to inform clinical practice and research.

Methods

The researchers employed a mixed-methods approach, combining self-report data with objective measurements. Participants wore two types of activity monitors: the ActiGraph wGT3X-BT (a triaxial accelerometer) and the activPAL (an accelerometer/inclinometer). These devices were worn for 10 days, starting two weeks after hospital discharge. Concurrently, participants completed the IPAQ-Short Form for a 7-day period.

To analyze the data, they used:

  • Lin's Concordance Correlation Coefficients (LCCC): This statistical method was used to assess the concordance between the self-reported and device-measured data for walking time, moderate-to-vigorous physical activity (MPA & VPA), total physical activity (METmins), and sitting time. LCCC is preferred over simpler correlation coefficients as it accounts for both precision and accuracy.
  • Bland-Altman Plots: These plots were used to visualize the agreement between the two measurement methods by showing the mean difference and the limits of agreement. This helps to identify systematic biases (over- or underestimation) and the variability of these differences.

Specific data processing steps included:

  • Truncating self-reported activity data to account for expected sleep and limiting bouts to a maximum of three hours per day, as per IPAQ guidelines.
  • Recoding short activity bouts (less than 10 minutes) to zero for both self-reported and device-based data.
  • Adjusting device-based data processing for the specific context of orthopedic injury patients, such as removing the "any one activity that accounts for >95% of waking wear time" condition and lowering the step threshold.

Datasets

The study utilized data collected from 46 adult participants who had sustained isolated upper or lower limb fractures. These participants were recruited from a major trauma center in Victoria, Australia. The data collected included:

  • Demographic information: Age, height, weight, sex, BMI, injury location, and weight-bearing status.
  • Objective activity data: From the ActiGraph wGT3X-BT and activPAL devices over a 10-day period.
  • Self-reported activity data: From the IPAQ-Short Form over a concurrent 7-day period.

No new benchmarks were introduced; the study focused on validating existing measures against each other within a specific clinical population.

Key Insights and Ideas

  • Self-report vs. Objective Measurement Discrepancy: The core insight is that self-reported physical activity (via IPAQ) and objective accelerometer data don't perfectly align, especially in individuals recovering from orthopedic injuries. This highlights the inherent limitations of relying solely on questionnaires for this population.
  • Overestimation of Activity: A significant finding is the consistent tendency for the IPAQ to overestimate how much people are walking and their overall physical activity (measured in METmins). This could be due to social desirability bias, a desire to appear more active, or a mismatch in how participants and devices define "walking."
  • Underestimation of Sitting Time: Conversely, the IPAQ tends to underestimate how much time individuals spend sitting. This is a critical finding, as sedentary behavior is a major concern for long-term health.
  • Injury Type Matters (Slightly): While overall trends were consistent, there were some nuances based on injury location. Those with lower limb injuries tended to overestimate their walking time more than those with upper limb injuries.
  • Low Moderate/Vigorous Activity: The study observed very little moderate or vigorous physical activity among participants, which is expected given their recovery status. This suggests that interventions and assessments should focus more on walking and light-intensity activities.

Key Results

  • Walking Time: The IPAQ overestimated device-derived walking time by an average of 2.34 hours per week.
  • Total Physical Activity (METmins): The IPAQ overestimated total physical activity by an average of 767 METmins per week.
  • Sitting Time: The IPAQ underestimated sitting time by an average of 2.26 hours per day.
  • Concordance:
    • Walking: Fair concordance (ρ = 0.34).
  • Sitting Time: Fair concordance (ρ = 0.38).
  • Total METmins: Moderate concordance (ρ = 0.43).
  • Moderate PA: Fair concordance (ρ = 0.21).
  • Vigorous PA: Slight concordance (ρ = 0.03).

Limitations and Future Work

The authors acknowledge several limitations:

  • Device Validation: The activPAL and ActiGraph accelerometers have not been previously validated specifically in orthopedic injury patients or under non-weight-bearing conditions, although they are widely used.
  • Recall Bias: Participants were asked to recall activity up to 10 days prior, which could introduce recall bias.
  • IPAQ Bout Definition: The IPAQ's requirement for activity bouts to be longer than 10 minutes might miss shorter, relevant activities performed by this patient group.
  • Sample Representativeness: The study was conducted at a major trauma center in an inner metropolitan area, which might lead to a sample with a higher proportion of certain injury types (e.g., cycling accidents) and potentially individuals more interested in physical activity.

Future work could involve:

  • Modifying the IPAQ or using more tailored assessment tools (like the CHAMPS Physical Activity Questionnaire for Older Adults) that are better suited for less active populations.
  • Focusing measurement on domains of physical activity and sedentary behavior most relevant to orthopedic injury patients, such as walking and light-intensity physical activity.
  • Investigating the reasons behind the overestimation of physical activity and underestimation of sitting time, potentially through qualitative methods.
  • Exploring the use of objective activity monitoring in clinical settings where feasible, or developing validated questionnaires that better capture the activity patterns of this population.

Summary updated

Abstract

Orthopaedic injury can lead to decreased physical activity. Valid measures for assessing physical activity are therefore needed in this population. The aim of this study was to determine the agreement and concordance between the International Physical Activity Questionnaire-Short Form (IPAQ) and device-measured physical activity and sitting time in orthopaedic injury patients. Adults with isolated upper or lower limb fracture (<i>n</i> = 46; mean age of 40.5 years) wore two activity monitors (ActiGraph wGT3X-BT and activPAL) for 10 days, from 2 weeks post-discharge. The IPAQ was also completed for a concurrent 7-day period. Lin's concordance correlation coefficients and Bland-Altman plots were calculated to compare walking/stepping time, total METmins, and sitting time. The IPAQ overestimated device-derived walking time (mean difference = 2.34 ± 7.33 h/week) and total METmins (mean difference = 767 ± 1659 METmins/week) and underestimated sitting time (mean difference = -2.26 ± 3.87 h/day). There was fair concordance between IPAQ-reported and device-measured walking (ρ = 0.34) and sitting time (ρ = 0.38) and moderate concordance between IPAQ-reported and device-measured METmins (ρ = 0.43). In patients with orthopaedic injury, the IPAQ overestimates physical activity and underestimates sitting time. Higher agreement was observed in the forms of activity (walking, total PA and sitting) commonly performed by this patient group.
BibTeX
@article{veitch2020agreement,
  author = {William Veitch and Rachel E. Climie and Belinda J. Gabbe and David W. Dunstan and Neville Owen and Christina L. Ekegren},
  title = {{Agreement between the International Physical Activity Questionnaire and Accelerometry in Adults with Orthopaedic Injury}},
  journal = {International Journal of Environmental Research and Public Health},
  year = {2020},
  doi = {10.3390/ijerph17176139},
}
Figure