Validity of stable isotope data in doping control: perspectives and proposals

Area of Science:

• Analytical Chemistry
• Biochemistry
• Forensic Science

Background:

• Carbon isotope ratio (¹³C/¹²C) analysis of endogenous steroids is used for doping detection.
• Variances in measurements arise from analytical uncertainty and biological scatter.
• Statistical distributions for steroid users are less understood than for controls.

Purpose of the Study:

• To present Bayesian approaches and mixture models for valid discrimination of steroid users.
• To address the challenge of overlapping distributions between users and non-users.
• To propose improved methods for evaluating stable isotope data in anti-doping.

Main Methods:

• Application of Bayesian statistical approaches.
• Deconvolution and characterization of routine data using mixture models.
• Analysis of ¹³C/¹²C ratios in endogenous urinary steroids.

Main Results:

• Mixture models effectively deconvolved data into components representing users and non-users.
• Significant overlap exists between the distributions of steroid users and non-users.
• Detection validity depends on analytical uncertainty and prior probability of doping.

Conclusions:

• Elaborate statistical treatment, including Bayesian methods, is necessary for accurate steroid user discrimination.
• Prior suspicion of doping and low analytical uncertainty enhance detection validity.
• Reporting conditional probability of steroid administration is recommended over measurement uncertainty.