Assessing uncertainty in forensic alcohol calculations: The case for using percentile ranges

Area of Science:

• Forensic Science
• Toxicology
• Statistical Analysis

Background:

• Forensic alcohol calculations estimate alcohol consumption using biological samples or BAC over time.
• Uncertainty in calculations stems from parameters like volume of distribution (Vd) and elimination rate (β).
• Existing guidelines suggest using percentile ranges, but a clear framework is lacking.

Purpose of the Study:

• To establish a transparent and defensible framework for utilizing percentile ranges in forensic alcohol calculations.
• To demonstrate the necessity and appropriate selection of percentile ranges using simulated data and statistical principles.

Main Methods:

• Utilized simulated data and statistical principles to analyze the impact of percentile ranges.
• Employed bootstrapping to estimate percentiles with a two-tailed 95% confidence interval.
• Differentiated percentile range recommendations based on case type (civil vs. criminal) and data availability.

Main Results:

• Recommended specific percentile ranges for civil (25th-75th) and criminal (0.5th-99.5th or 2.5th-97.5th) cases.
• Showed that using case-specific percentile ranges improves the reflection of underlying uncertainty.
• Demonstrated that percentile ranges should be determined from an appropriate population.

Conclusions:

• The selection of percentile ranges in forensic alcohol calculations should be tailored to the case type and data volume.
• This approach offers a more accurate representation of uncertainty compared to relying on population averages.
• The proposed framework enhances the transparency and defensibility of forensic alcohol calculations.