ROC generated thresholds for field-assessed aerobic fitness related to body size and cardiometabolic risk in schoolchildren

PLoS One. 2012;7(9):e45755. doi: 10.1371/journal.pone.0045755. Epub 2012 Sep 21.


Objectives: 1. to investigate whether 20 m multi-stage shuttle run performance (20mSRT), an indirect measure of aerobic fitness, could discriminate between healthy and overweight status in 9-10.9 yr old schoolchildren using Receiver Operating Characteristic (ROC) analysis; 2. Investigate if cardiometabolic risk differed by aerobic fitness group by applying the ROC cut point to a second, cross-sectional cohort.

Design: Analysis of cross-sectional data.

Participants: 16,619 9-10.9 year old participants from SportsLinx project and 300 11-13.9 year old participants from the Welsh Schools Health and Fitness Study.

Outcome measures: SportsLinx; 20mSRT, body mass index (BMI), waist circumference, subscapular and superilliac skinfold thicknesses. Welsh Schools Health and Fitness Study; 20mSRT performance, waist circumference, and clustered cardiometabolic risk.

Analyses: Three ROC curve analyses were completed, each using 20mSRT performance with ROC curve 1 related to BMI, curve 2 was related to waist circumference and 3 was related to skinfolds (estimated % body fat). These were repeated for both girls and boys. The mean of the three aerobic fitness thresholds was retained for analysis. The thresholds were subsequently applied to clustered cardiometabolic risk data from the Welsh Schools study to assess whether risk differed by aerobic fitness group.

Results: The diagnostic accuracy of the ROC generated thresholds was higher than would be expected by chance (all models AUC >0.7). The mean thresholds were 33 and 25 shuttles for boys and girls respectively. Participants classified as 'fit' had significantly lower cardiometabolic risk scores in comparison to those classed as unfit (p<0.001).

Conclusion: The use of the ROC generated cut points by health professionals, teachers and coaches may provide the opportunity to apply population level 'risk identification and stratification' processes and plan for "at-risk" children to be referred onto intervention services.

MeSH terms

  • Adolescent
  • Area Under Curve
  • Body Mass Index
  • Cardiovascular Diseases / etiology
  • Cardiovascular Diseases / prevention & control
  • Child
  • Cluster Analysis
  • Cross-Sectional Studies
  • Exercise Test*
  • Exercise Tolerance
  • Female
  • Humans
  • Male
  • Metabolic Diseases / etiology
  • Metabolic Diseases / prevention & control
  • Overweight / complications
  • Overweight / physiopathology
  • Physical Fitness*
  • ROC Curve*
  • Risk Assessment

Grants and funding

The SportsLinx study was funded by Liverpool John Moores University, Liverpool City Council, and Liverpool Primary Care Trust. The Welsh Schools Health and Fitness study was funded by the National Institute for Social Care and Health Research within the Welsh Assembly Government. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.