How a blood alcohol number gets made
Your blood goes into a gray top tube containing sodium fluoride, a preservative that stops fermentation, and potassium oxalate, an anticoagulant. The tube gets inverted to mix, labeled, sealed, and transported. At the lab a portion is placed in a sealed vial, heated, and the vapor above the liquid is drawn off and injected into a gas chromatograph. Compounds separate as they move through the column and hit a detector at different times. The detector produces a chromatogram: peaks over time. The area under the ethanol peak, compared against calibrators of known concentration, produces your number.
Every sentence in that paragraph is a place where something can go wrong.
What I request, on every blood case
- The chromatograms for your sample and for every calibrator and control in the batch
- The full batch sequence, so I can see what ran immediately before your vial
- Calibration and calibration verification records
- Blank runs, and what they showed
- Instrument maintenance and repair history
- The analyst's training file, proficiency test results, and any corrective action history
- The lab's standard operating procedure and its uncertainty of measurement budget
- Chain of custody, tube lot numbers and expiration dates, and storage temperature logs
Some of that is produced routinely. Some of it has to be fought for. All of it is discoverable.
The things that actually move a number
Fermentation. A tube with a failed or insufficient preservative can grow organisms that produce ethanol after collection. The sample gets drunker while it sits.
Serum versus whole blood. Hospital labs usually test serum, which runs roughly twelve to twenty percent higher than whole blood. Converting between them requires the right factor, and getting it wrong inflates the result.
Co-elution. Two compounds leaving the column at nearly the same moment get reported as one, and the area under the peak goes up. You can see it in the peak shape. You cannot see it on the report.
Carryover. A high sample can leave residue affecting the next run. The control is a blank between samples. Whether one was run, and what it showed, is in the batch record.
Manual reintegration. When software draws the baseline poorly, an analyst can redraw it by hand. That is legitimate practice and it is also a human judgment that changes your number, and it is supposed to be documented.
Breath tests
The Intoxilyzer does not measure blood. It measures breath and converts, using an assumed blood to breath ratio of 2100 to 1. Real people vary around that number, and the variance runs in both directions. Add mouth alcohol from reflux, a breathing pattern that changes the sample, a missed fifteen minute observation period, and reference sample solutions that drift, and you have a device that is checkable in a dozen ways.
Maintenance records, the operator's certification, and the observation period video are all obtainable. The observation period is on video far more often than officers expect.