“Good” Science For Gas Chromatographic Blood Alcohol Analysis

First of all, we need a cast of characters for this story.

A Laboratory

The central calibration and testing laboratory in Texas is the Texas Department of Public Safety Forensic Toxicology Laboratory [DPS].

The Rules

All laboratories have their own Standard Operating Procedures. These are usually in the form of a manual that describes in great detail the analytical process for blood alcohol. This manual always should be followed! As my friend Tom Pittman says: “The SOPs are the scientist’s bible, and if you don’t believe in the Bible you can’t be a Christian.”

The Machinery

There are a number of companies that make gas chromatographs. Agilent, Perkin Elmer, and Shimadzu are the companies most used in Texas. DPS labs use Perkin Elmer gas chromatographs. This is what they look like:

Gas Chromatograph

The chromatograph is on the left, and it is attached by a black tube and some software to another machine on the right called a turbomatrix. On the turbomatrix is a circular carrel with all the round slots where the testing vials are placed – hopefully in the same order listed in the software program.

The Result

There will be a printer somewhere that prints out the chromatograms. They look like this:

The software program (Total Chrome, if you are interested) will print one of these for every vial that was placed in that carrel. This chromatogram is actually the raw data file from which we can calculate the ratio of the ethanol peak area to the n-propanol peak area. The first peak is ethanol, second n-propanol. That ratio helps to calculate an amount: .08, right?

The Problem

When I think of “ethanol” I’m thinking “margarita.” I don’t think about n-propanol much except when I talk about this stuff. N-propanol is the “internal standard.” They mix it up the same way every time, and they put the same amount of it in all the testing vials. They also put the same amount of blood in each vial for every test, but the thing is, they don’t know if or how much of the margarita, I mean ethanol, is in that blood. That’s why we are here. If there is ethanol in that blood, it will make a peak at that location on that chromatogram every time it is detected in someone’s vial: EXCEPT the peak is going to be bigger or smaller depending on the number of margaritas, right?

Okay, we ran the batch. The needle came out of that turbomatrix, pierced the septum on the vial, took out the headspace gas, and sent it over to the chromatograph through that black tube to be analyzed. However, in that batch there were also other vials that had ethanol in them with known concentrations. They are called “controls.” They are made by companies who manufacture them and seal them into glass vials. The controls are certified to be an exact amount of an ethanol concentration. The companies certify that these controls comply with the requirements of the National Institute of Standards and Technology: www.nist.gov. Why? We must know that the machine is measuring the ethanol concentration accurately.

If our controls are measured accurately in the batch of tests with your blood sample, then we will know that we are measuring your margaritas correctly. Right? Not so fast, professor. Let’s look at the laboratory data, starting with the in-house review of the work by the analyst. It may look something like this:

Hmmm… The comments by the batch analyst’s reviewer were:

“HSGCC (Run A) had communication error starting at vial 44, so vials 44-60 were recapped and the sequence was restarted on 6/20/18 for vials 44-60.”

The comment refers to software communication error between the chromatograph and the turbomatrix that handles the vials and sends the gas sample to the chromatograph for testing. The vials in this batch, as is usual, each had a septum that was pierced by a needle used by the turbomatrix and through which gasses within the vial are in turn sent through that black tube connecting the two machines to the chromatograph for analysis.

Okay! Remember that moment before you were going to get that shot in your whatever, and the nurse took that long needle and inserted it through the top of the vial that held the penicillin? There was a rubber stopper inside a metal frame on top of her vial. That rubber thing is a “septum.” Our analyst put the blood and the n-propanol in the glass vial. He put the septum on top. He took the metal cap and placed it over the septum, and used a “crimper” to seal the deal.

The turbomatrix and the chromatograph communicate during the analysis via Total Chrome software. When the carrel has been loaded with all of the vials to be tested, once begun, the testing process proceeds automatically.

During the automated testing process in this batch, software communication between the turbomatrix and the chromatograph ceased to function at vial 44 even though the mechanical testing proceeded to conclusion of all 60 vials. Therefore, vials 44 – 60 each had the septum pierced by the turbomatrix needle, and a portion of gas was extracted from each of the vials. However, no information was derived from vials 44-60 in the chromatograph that could be used to create chromatograms for those specimens.

Laboratory Quality Action Plan

Once the error was recognized by the analysts, they should have looked to see if the SOP’s have a procedure to correct the problem. If it was a problem that could affect the batch as a whole, the failsafe correction would be to redo the batch. Pipette everything again, and put all the vials in again for re-testing. Sure, it’s a lot of trouble, and yes, the lab has a LOT of work to do, but it’s a good solution. It will be scientifically reliable for everyone in the batch as long as the analyst follows the SOPs again.

However, maybe the SOPs did not cover the particular problem involved. Then what happens? In this case the solution was to uncap vials 44-60; use the same remaining mixture; recap them with a new, un-pierced septum; and restart the analyses using the same software program, beginning at vial 44 the next morning.

Defense Analysis

The failsafe, of course, is the more reliable scientific resolution to the problem. Run all sixty samples, including calibrators and controls, and run the tests correctly to conclusion using the same calibration and controls for each vial in the batch. Why?

  1. When vials 44-60 were opened and recapped some additional portion of the gasses escaped from the vials, including the n-propanol. It appears that the analysts assumed when using this procedure that the same amount and proportion of n-propanol and ethanol escaped from the vials. “Assume” is not really a measurement.
  2. The analyst calculated that the difference of the amounts of n-propanol between the two sets of vials was less than 10%. So he reported all results in one batch sheet. Was that reliable scientifically?
  3. Even if the n-propanol amount in the set of tests from 44-60 differed from the first 43 tests by less than 10%, vials 44-60 were still less accurate than the first 43 tests.
  4. Why 10%? That percentage is the acceptable “range of error” adopted by the Society of Forensic Toxicologists and the American Academy of Forensic Scientists for blood/alcohol testing.

You and your friends are dining at a good restaurant. The head chef comes to your table after the meal. (You ordered his specialty!) He asks how was the food? You say, “Acceptable.” Is that an indication of quality to him? What quality?

You are a defendant in a criminal case, and the analyst was able keep his error in your case within a range that was “acceptable.”Do you think that is scientific reliability? You might say, “If I was a juror on the case I would probably have to accept the analysis because it was scientific.” Well, was it really scientifically reliable?

Different people in the same batch obtained more accurate readings than others because of a lab analyst’s opinion on how the error could be corrected. There is no mention of the lab’s Standard Operating Procedures. How would you feel if you were the defendant in a trial in which the rights you were provided were merely acceptable based upon someone’s unguided opinion? Was this solution “good” science?


Leave a Comment