Imputing Accelerometer Nonwear Time When Assessing Moderate to Vigorous Physical Activity in Children

Area of Science:

• Pediatric physical activity assessment
• Biometric data analysis
• Public health research

Background:

• Accelerometer-based physical activity measurement is limited by nonwear time.
• Nonwear time is typically excluded from analysis, potentially biasing results.
• Novel imputation methods using participant data can address nonwear time.

Purpose of the Study:

• To evaluate the impact of imputing accelerometer nonwear time on moderate to vigorous physical activity (MVPA) estimates in children.
• To determine if nonwear time imputation affects the association between MVPA and health outcomes (BMI, blood pressure).

Main Methods:

• Accelerometer data collected for 7 days from 332 children (10-13 years).
• MVPA estimated using two datasets: one with nonwear time deleted, another with nonwear time imputed using sociodemographic, health, and time data.
• Associations between MVPA, BMI, and blood pressure analyzed in both datasets.

Main Results:

• Nonwear time constituted 7% of waking hours.
• Average MVPA estimates were similar between nonimputed (56.8 min/d) and imputed (58.4 min/d) datasets.
• The strength of associations between MVPA and health outcomes did not differ between the two datasets.

Conclusions:

• Imputing nonwear time does not substantially bias MVPA estimates or their health associations in this pediatric cohort.
• Traditional deletion of nonwear time is acceptable for MVPA estimation when accelerometer wear-time compliance is high.