I have commented repeatedly in the past about the inaccuracy and unreliability of breath-testing devices used in DUI investigations. This is due to a wide range of factors: inherent design defects (see, for example, my previous post "Why Breathalyzers Don’t Measure Alcohol"); ineffective calibration and maintenance of the machines; improper administration of the test; radio frequncy interference; and, most importantly, physiological variability in humans.
The main problem with breath machines is that they are designed to assume all human beings are the same (see "Convicting the ‘Average’ DUI Suspect"). In fact, we are all very different from one another in ways that are critical to such testing — and we are ourselves physiologically different from one moment to the next. Each of us, for example, is inherently different in our partition ratio — the ratio of alcohol in our breath compared to alcohol in our blood — and this ratio differs within ourselves from hour to hour (see "Breathalyzers — and Why They Don’t Work").This is critical, as the breathalyzer will automatically compute the amount of alcohol in the blood based upon the measured alcohol on the breath — using a uniform ratio that (falsely) assumes we are all the same.
Another human variable is the existence of such conditions as diabetes (see "Diabetes and the Counterfeit DUI"), acid reflux ("GERD, Acid Reflux and False Breathalyzer Results")…. and anemia. A person suffering from anemia has a low red blood cell count, perhaps half as much as would be normal. Put simply, when there are fewer red blood cells, the body will increase the amount of plasma to fill the void. Red and white blood cells are solid; plasma is liquid. Alcohol is attracted to liquid in the body, not muscle, bone, or other solids. It follows that the higher the ratio of liquid to solids in the blood (called the hematocrit), the higher the amount of alcohol in the blood — and the higher will be the reading on the breathalyzer. The male-female average hematocrit is 45% (men average 47%, women 42%), but the range varies for men from 42 to 52%, and for woman from 37 to 47%. The machine, of course, assumes that all suspects have a hematocrit of 45%.
The effect of an individual’s hematocrit on breath analysis can be mathematically computed. The partition ratio of alcohol in blood to alcohol in breath uniformly used in breath testing is 2100:1. If the suspect’s hematocrit is, say, 54%, the breath test result could be computed by multiplying it by 45/54. Assuming a breath test result of .11%, for example, the true blood-alcohol concentration could be determined by the formula .11 x 45/54 = .09%. In other words, a person with a true BAC of .09% but a hematocrit of 54% would test on an otherwise "accurate" machine as .11%. Just because he/she is anemic — or simply varies from the statistical norm.