You are here

Article Text

Article menu

Download PDFPDF

Free Communications
084 Monitoring workload to evaluate injury risk: the impact of missing data
Free
  1. Lauren Benson1,
  2. Carlyn Stilling1,
  3. Oluwatoyosi Owoeye1,2,
  4. Carolyn Emery4,5,1,6,3
  1. 1Sport Injury Prevention Research Centre, Faculty of Kinesiology, University of Calgary, Calgary, Canada
  2. 2Department of Physical Therapy and Athletic Training, Doisy College of Health Sciences, Saint Louis University, St. Louis, USA
  3. 3Alberta Children’s Hospital Research Institute, University of Calgary, Calgary, Canada
  4. 4McCaig Bone and Joint Institute, Cumming School of Medicine, University of Calgary, Calgary, Canada
  5. 5Department of Community Health Sciences, Cumming School of Medicine, University of Calgary, Calgary, Canada
  6. 6Department of Pediatrics, Cumming School of Medicine, University of Calgary, Calgary, Canada

Abstract

Background The acute: chronic workload ratio (ACWR) is used to monitor workload, with both low and high ACWR associated with injury risk. Ignoring or imputing missing data points may influence ACWR calculations.

Objective To examine the effect of ignoring versus imputing missing data on ACWR.

Design Cohort, longitudinal.

Setting Youth basketball.

Participants Fifty (25F, 25M; 16.5 years; 66.2 kg; 173.5 cm) basketball players on four high school teams.

Assessment of Risk Factors Participants wore a jump counter (VERT® Classic) to record external workload during practices and games throughout the 17-week season.

Main Outcome Measurements Two datasets were created: missing data were ignored, and missing data were imputed using a machine learning algorithm based on typical jump counts for the individual, team and sex. The distribution of ACWR was compared between datasets using a two-sample Kolmogorov-Smirnov test. Pearson correlations were used to assess how the ACWR for the ignored and imputed datasets relate to the difference between the percent of missing acute and chronic data.

Results The distribution of ACWR was significantly different between the ignored and imputed datasets (D=0.164, p<0.001). The ignored dataset had 40% more cases of ACWR<0.5 and 97% more cases of ACWR>2.0 than the imputed dataset. There was a significant moderate association between ACWR and the difference between the percent of missing acute and chronic data for the ignored dataset (rho=0.617, p<0.001). When more acute than chronic data are missing, ACWR is low; when more chronic than acute data are missing, ACWR is high. There was no relationship between missing data and ACWR for imputed data (rho=0.061, p=0.147).

Conclusions When missing data are ignored, ACWR is dependent on the quantity of missing acute and chronic data. Additionally, ignoring rather than imputing missing data is likely to result in more extreme ACWR, which could influence evaluation of the relationship between workload and injury risk.

https://doi.org/10.1136/bjsports-2021-IOC.80

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.