Breath Test Machines: Less Reliable Than You Think

Chris Perri, next to his very own breathalyzer.

In DWI investigations, breath test results are very common evidence. Although I advise clients to refuse to provide a sample of their breath, the case isn’t over just because the machine says that the sample is over .08 BAC.

At the outset, it’s important to realize that BAC means blood alcohol concentration, not breath alcohol concentration. Breath test science relies on the assumption that it can obtain a reliable breath alcohol concentration that mirrors a particular person’s blood alcohol concentration. This assumption is fraught with difficulties, as I’ll explain below.

Let’s start with a very simplified description of the mechanics of the breath-test machine (also known as an intoxilyzer or breathalyzer). An arrested person blows into the machine, which takes this breath sample and shoots it through a tube that’s then injected with infrared light. Because alcohol particles block infrared light, the machine detects the amount of alcohol in a subject’s breath by determining how much of the infrared light has been blocked. It then takes this number, makes some calculations, and reports a breath alcohol concentration.

The calculation of this breath alcohol concentration is problematic because there’s a lot less alcohol in the breath than in the blood. For example, in an average person, the number of grams of alcohol in 1 part of the blood is equivalent to the number of grams of alcohol in 2100 parts of breath. This 1:2100 ratio is known as a “partition rate.” So, while a person’s blood alcohol concentration is defined as the number of grams of alcohol in 100 milliliters of blood, that same person’s breath alcohol concentration is defined as the number of grams of alcohol in 210 liters of breath. (Note: 100 milliliters x 2100 = 210 liters). In other words, if you have .08 grams of alcohol in 100 milliliters of your blood, then it’s assumed that you have .08 grams of alcohol in 210 liters of your breath.

While it’s not hard for police to obtain 100 milliliters of your blood, it’s impossible for them to obtain 210 liters of your breath (think about a 1 liter bottle and imagine filling up 210 of those bottles with your breath). For this reason, the breath test machine must multiply any amount of alcohol that it detects by a very large number. As an example, if you provide the machine with one liter of your breath, then the machine multiplies the amount of alcohol it detects by 210 in order to determine the number of grams of alcohol per 210 liters of your breath. This calculation is then reported as your BAC.

Consequently, any error by the machine in determining the amount of alcohol in a given sample would be exacerbated when it multiplies that incorrect number by two-hundred-fold. Such errors can occur when the machine interprets non-alcoholic particles in the breath as alcohol. For example, the machine cannot distinguish acetone (a common substance in the breath of diabetics) from alcohol. Or, imagine the complication of a stray particle of liquid alcohol entering the machine in the form of spit. Any error in the initial measurement of alcohol will render the entire breath test unreliable.

Photograph courtesy of Oregon Dept. of Transportation

For my next point, let’s give the machine the benefit of the doubt and assume that it can accurately measure the quantity of alcohol in a person’s breath. Even then, the machine makes a critical and troublesome assumption: that the subject’s partition rate is 1:2100 (recall from above that this means that the amount of alcohol in one part of blood is equal to the amount of alcohol in 2100 parts of breath). This assumption is not true across the population, as studies show that partition rates of normal people vary from 1:1100 to 1:3000.

As an illustration, let’s take a hypothetical subject arrested for DWI and call her Sue. She has a partition rate of 1:1100, which means that the number of grams of alcohol in 1 part of Sue’s blood is the equivalent of the number of grams of alcohol in 1100 parts of her breath. On this particular evening, Sue has consumed enough alcohol that her BAC is .06 grams of alcohol per 100 milliliters of blood. Since her partition rate is 1:1100, there would also be .06 grams of alcohol in 110 liters of her breath (note: 100 milliliters multiplied by 1100 equals 110 liters).However, the breath machine is going to overestimate the BAC by nearly a factor of two. Here’s why:

When Sue takes a breath test, the machine is programmed to incorrectly assume that her partition rate is 1:2100. As a result, it will determine the number of grams of alcohol in 210 liters of her breath. Since Sue’s correct BAC is .06 grams of alcohol per 110 liters of breath (due to her partition rate of 1:1100), she has .11 grams of alcohol per 210 liters of her breath. Due to the machine’s inaccurate assumption that Sue’s partition rate is 1:2100 instead of 1:1100, the breath-test machine will report a BAC of .11, which is nearly twice as high as her actual BAC of .06. This inaccuracy results because the machine is improperly measuring the number of grams of alcohol per 210 liters of Sue’s breath, as opposed to the number of grams of alcohol per 110 liters of her breath. Thus, the machine’s assumption that everyone has a partition rate of 1:2100 creates a critical error by reporting that Sue is intoxicated even though she’s actually below the legal limit of .08 BAC.

Finally, it’s important to remember that it’s only illegal to be intoxicated while driving. It’s not a crime to be intoxicated 10 minutes, 30 minutes, or an hour after operating a motor vehicle. But these chemical tests usually occur over an hour after driving, and the prosecution has the burden of proving how that BAC measurement relates to the time of driving. Unless law enforcement knows a lot of information about a particular person (e.g., number of drinks consumed, when the drinks were consumed, the type of alcohol consumed, amount of food consumed, when food was consumed, etc.), it’s impossible to conduct a reliable retrograde extrapolation, which is the science of determining a past BAC level based on a known BAC level. In other words, if the breath test machine reports that a person has a .11 BAC over an hour after driving, we don’t know whether the person’s BAC at the time of driving was below, above, or the same as the level reported by the machine.

While breath tests are an important tool for law enforcement in that they give a ballpark figure about an arrested person’s intoxication level, it’s a common misconception that a breath test machine provides an exact measurement of a person’s BAC at the time of driving. If you or a loved one have been arrested for DWI and submitted to a breath test that reported a BAC over .08, don’t despair. Call an experienced criminal defense attorney to fight the machine’s potentially inaccurate result.