Does a Strong Blood–Breath Alcohol Correlation Make the Results Interchangeable?

No. Blood and breath alcohol can correlate strongly while giving different numerical results under a fixed conversion convention. In a reanalysis of controlled drinking data from 100 volunteers, venous blood alcohol and breath alcohol had a correlation of 0.948. Yet individual blood-to-breath ratios ranged from approximately 2,125:1 to 2,765:1.

Jones and Cowan’s 2020 paper addresses a specific DUI interpretation problem: correlation describes how measurements vary together across a group; it does not prove that a fixed ratio reconstructs each person’s venous blood alcohol exactly.

How the paired measurements were obtained

The participants were 85 men and 15 women, all physically fit employees of the Texas Department of Public Safety. They consumed a controlled whiskey dose after a sandwich. Serial breath readings were used to identify two consecutive decreases before the paired comparison.

For that comparison, a modified Intoxilyzer 8000 measured breath alcohol, breath temperature and exhalation information. Venous blood was collected immediately afterward and analyzed by dual-column headspace gas chromatography, producing four analytical results per blood specimen from which a mean was calculated.

The paper statistically re-evaluated previously collected drinking-study data. It was not a representative roadside survey and did not measure actual crash outcomes.

Specify the specimen and units before comparing

Blood alcohol was expressed in grams per 100 millilitres of venous whole blood. Breath alcohol was expressed in grams per 210 litres of breath. These denominators describe different specimens and quantities. Similar printed numbers are not literally identical concentrations in the same fluid.

The study’s mean blood-to-breath ratio was approximately 2,382:1, with a standard deviation of 119. No participant had a ratio below 2,100:1 under these post-absorptive testing conditions. Consequently, the breath values expressed in g/210 L were numerically lower than the corresponding venous values expressed in g/100 mL.

That finding does not mean the breath instrument failed to measure breath accurately. Analytical measurement of breath ethanol and biological inference about venous blood are separate issues. Serum/plasma, urine and biomarkers were not substitutes for the paired whole-blood specimens in this analysis.

Why correlation is not individual agreement

A high correlation means that participants with higher blood results tended to have higher breath results. It does not eliminate differences in their ratios or provide an exact personal conversion. The observed group range also is not a guaranteed interval covering every future driver.

The mean ratio cannot simply replace 2,100:1 in an individual case without acknowledging prediction uncertainty and the conditions of testing. A repeat analytical result can be precise while a biological conversion remains uncertain.

Temperature and exhalation were associated with the ratio

Higher breath and body temperatures were associated with lower blood-to-breath ratios, as were longer exhalations. The multivariable model explained approximately 43% of ratio variation. Underlying BAC, breath temperature, body temperature and exhalation duration remained significant predictors in that model.

These associations do not supply a validated personal temperature correction. They also do not establish that one breathing pattern caused the entire difference between a particular driver’s blood and breath results. The laboratory instrument and instructed prolonged exhalation are part of the conditions under which the findings were obtained.

There were no significant mean ratio differences between men and women or among the study’s three reported racial groups. Small and uneven subgroup sizes limit that null finding: it is not proof of identical ratios for every subgroup or individual.

The post-absorptive restriction matters

The analysis deliberately selected measurements after serial breath readings declined. It therefore cannot demonstrate the same ratio distribution during active absorption. Arterial and venous ethanol need not track identically during absorption, and breath is related differently to those blood sources.

The study collected one paired venous specimen per person for this analysis. It did not establish a stable lifetime ratio or a complete within-person ratio curve. Food, dose, health, instrument and sampling differences can limit extrapolation beyond the experiment.

Using the study in a DUI evidence review

First identify what the instrument actually measured and in what units. Then distinguish a directly reported breath result from a calculated venous-blood estimate and assess timing, protocol and uncertainty. The scientific question of conversion is separate from which specimen a particular legal rule governs; this paper does not establish current legal requirements.

The importance of collection source is also illustrated by the fingertip-versus-venous ethanol comparison. “Blood alcohol” needs a defined specimen, and a correlation cannot erase that distinction. Neither this ratio study nor a converted number independently measures behavioral impairment or crash causation.

Primary research source

Jones AW, Cowan JM. Reflections on variability in the blood–breath ratio of ethanol and its importance when evidential breath-alcohol instruments are used in law enforcement. Forensic Sciences Research. 2020;5(4):300–308. https://doi.org/10.1080/20961790.2020.1780720.