Vol 2 Iss 4 Art 4 - Diabetes 2 - Counterpoint - The Journal of Science & the Law

Diabetes & Breath Alcohol Testing: Part 2

Issues with breath testing devices

Counterpoint Volume 2: Issue 4 - Article 4 (February 2018)

An article in the Foundational Skills I-4 module

Jan Semenoff, BA, EMA
Forensic Criminalist

In part one of this series, we discussed Diabetes Mellitus, a chronic metabolic disease that reduces or eliminates the body’s natural production of insulin in the pancreas. In this article, we will look at the effects that diabetes may have on a breath alcohol test. You may wish to review the metabolic processes of diabetes (and fasting diets) before continuing with this article.

The effects of diabetes

In general, the observable signs & symptoms of Hyperglycemia (Diabetic Ketoacidosis) are similar to that of an intoxicated or impaired person. Levels of consciousness may be altered, and there may be a musty, alcohol odor on the breath. The ability of as person to follow directions may be affected, as impaired cognitive function is typical.

Insulin shock symptoms (Hypoglycemia) also display lowered levels of consciousness or confusion, heightened emotional states or violence, and a smell of an unusual odor on the breath.

The effects of fasting

When subject to fasting conditions, the body will begin to metabolize stored fats. Ketone levels will then begin to rise in the blood and urine. This creates a situation where the person becomes hypoglycemic from a cause other than too much insulin. In general, the signs and symptoms of hypoglycemia display the same symptoms as a diabetic in insulin shock: lowered levels of consciousness or confusion, heightened emotional states or violence, and a smell of an unusual odor on the breath.

First Responders are taught that the signs and symptoms of each, although initiated by very different root causes, may be difficult to differentiate. Indeed, a standard way of teaching how to identify diabetic disorders in first aid courses is to describe the individuals as having a “drunken appearance.” The first aid strategy when one cannot differentiate the symptoms is to give sugar as a fallback position, as it will immediately assist the hypoglycemic patient, and will not appreciably harm the hyperglycemic person. Symptomatic diagnosis, even by a trained medical assessment, is virtually impossible. Blood glucose or blood ketone levels are the only accurate way to assess which diabetic condition is present, if any. Ketone level measurement is emerging as the more accurate predictor of early-onset diabetic assessment.

Diabetic metabolism

The initial situation for a diabetic or people in a hypoglycemic state are rising levels of ketones (b-hydroxybutyrate in the blood that is first metabolized to form Acetoacetic Acid, then finally Acetone which can be metabolized further into Isopropanol). The infrared signature of b-hydroxybutyrate and isopropanol is similar to that of ethanol in the range read by earlier infrared breath alcohol testing devices. There is quite a significant level of b-hydroxybutyrate in the blood before any acetone is produced. It is only the later stage of acetone metabolization that is detected by breath alcohol devices. See Figure 1:

Figure 1 - The metabolism of diabetic ketones during Diabetic Hypoglycemia

This overlap of diabetic ketones with ethanol in various u (micron) ranges is indicated in the black rectangles in Figure 2. You should note that different breath testing devices use different micron ranges of the ethanol molecule to read for the presence and concentration of ethanol.

Figure 2 - The infrared overlap of Ethanol (Red) with the Diabetic Ketones and Isopropanol. Note the areas read by various Infrared breath testing devices in the black rectangles.

The fuel cell component of a breath testing device is also not immune to the effects of the ketones bodies on the reading obtained. In 2007, internationally noted toxicologist Dr. A.W. Jones (with S. Rossner) reported an instance of a false-positive breath test in a fuel-cell interlock device from a person on a ketogenic diet who was an absolute abstainer from alcohol.

It is believed that the increased levels of ketones, including acetone that is known to be biotransformed into isopropanol by the action of liver alcohol dehydrogenase leads to this false-positive effect on the fuel cell device. They concluded that the side effects of the ketogenic diet warrant caution and further evaluation by authorities during breath alcohol testing.

Ketoacidosis can be smelled on a person’s breath and is commonly dismissed as alcohol consumption. The levels of exhaled ketones, including acetone, rise appreciably as the person continues to metabolize fats and sugars. It is the exhaled acetone that is designed to be detected by the acetone detectors in modern breath alcohol testing instruments. However, the level at which each detector system is set to trigger vary from jurisdiction to jurisdiction.

Acetone in and of itself is not very toxic to humans. For this reason, the levels of the detector are often set quite high. Many jurisdictions use a concentration of 1 milliliter of acetone in 100 milliliters of water [1] as the threshold level. This may be higher than the level experienced by most uncontrolled diabetics experiencing an episode of severe diabetic ketoacidosis and may therefore have little use in an evidentiary breath alcohol instrument. Also remember that acetone is only produced by the diabetic at the latter stages of metabolism. It has also been demonstrated that the acetone is not a total waste product, being then converted into isopropanol by the diabetic. This is also an important step to consider.

[1] Past conversations and testimony from various state forensic criminalists indicate they use a 1-milliliter volume (This is typically measured in the lab as 20 drops of acetone).

The Partition ratio of acetone
The blood to air partition ratio for isopropanol has an accepted value of 1372:1. The partition ratio for ethanol has been legislatively accepted at 2100:1. Therefore, any trace levels of isopropanol found within the body would have an exaggerated effect on the readings obtained on a device measuring at 2100:1. It has also been reported that blood levels of isopropanol have a tendency towards very low elimination rates in the body (Jones, 1996).

Figure 3 - The Intoxilyzer 8000

Newer instruments such as the Intoxilyzer 8000 advertise, and this has been supported in various state training manuals, the notion that the 9.5-micron range is better suited in reading ethanol levels. Florida’s state training manual states, as an example, that the 3.3 – 3.8 µ range (the range used in the older Intoxilyzer 5000EN to determine the presence and concentration of ethanol) is better suited to determine the presence of Interferents. The overlap of acetone, and its metabolite isopropanol, mimic that of ethanol in the 3.3 – 3.8 µ range. And, in experiments that I've been able to perform, it appears that the 8000 is more reliable at determining the presence of interfering substances than the older model 5000.

It has been identified by various researchers that certain hydrocarbon compounds, in concert with low levels of ethanol cause inflated readings on the Intoxilyzer 5000 that are often not detected by its interferents-detect algorithm (Hak, 1995, Jones et al, 1996, Caldwell & Kim, 1997, Bell et al, 1992 and Memari, 1999). Infrared breath testing devices attempt to mitigate this tendency through the inclusion of additional filter points (sometimes at 3.36 µ and 3.52 µ, as with the Intoxilyzer 5000EN). However, these points will only assist in the detection of toluene and acetaldehyde. Toluene was not present. Acetaldehyde is a naturally occurring by-product of ethanol consumption and is expected to be produced by the body.

Figure 4 - The Intoxilyzer 5000 Series

Infrared or fuel cell breath testing devices are designed to identify the presence and concentration of ethanol. To earlier divices, all alcohols and many hydrocarbons produce an absorption pattern from the methyl group of the molecules between 3.2-3.5μ range. To a certain extent, all alcohols, and many other hydrocarbons, therefore appear as ethanol to these infrared instruments. However, the difference in the individual infrared signatures of these hydrocarbons as compared to ethanol, as interpreted by the breath testing device, is supposed to “flag” the sample as containing an interfering compound. It has been my observation that this feature does not always function as designed and can routinely report a falsely elevated BrAC reading.

I have identified the tendency of earlier devices such as the Intoxilyzer 5000EN to over-report the true BAC of a sample of ethanol in the presence of interferents such as isopropanol, methanol, D-limonene, Dimethyl Sulfone, Castor oil, Adipic acid and Methylsulfonylmethane. With a mixture of both ethanol and another infra-red hydrocarbon, the interferent detection algorithm fails.

I have observed that, when confronted with a variety of potential interferents, the Intoxilyzer 5000 and 5000EN will report exaggerated BAC readings. The Intoxilyzer 5000, even the enhanced EN version, is just not sophisticated enough to discern the overlapping infrared signatures and separate them from ethanol. I performed a series of simulations on an Intoxilyzer 5000 66-series and had a “true” ethanol level of 0.035 grams in a simulator elevated to an average of 0.072 grams/dL with the inclusion of less than 1.0 ml of isopropanol and less than 1.0 ml of acetone (the expected metabolite of isopropanol). However, the unit only reported an interferent in 7 out of 15 samples. When the same 0.035 grams ethanol solution vapor was introduced into an Intoxilyzer 5000EN 68-series, the average reported BAC was inflated to .117 grams/dL, yet the interferent detect circuitry only reported the interferent on 8 out of 15 occasions.

What is more alarming is the tendency of the units to report the BAC as a subtracted, or corrected, value on the occasions when the units did discover the interferent. In the simulations described above, during those times that the Intoxilyzer 5000 or 5000EN did determine the presence of the interferent isopropanol, they reported the inflated BAC value as being a corrected value, and apparently the product of subtraction of the false interferent value. However, these BAC values reported were still over-represented, often more than threefold.

See Table 1 for these test results:

Table 1 - The effect of acetone and isopropanol on the Intoxilyzer 5000 series

It should be noted that newer evidentiary breath test devices like the Intoxilyzer 9000 or Datamaster DMT are very good at identifying interferent substances like acetone, or other ketones, and may not provide a false positive reading with a diabetic. See the article on the Intoxilyzer 9000, and its specificity towards ethanol. An overview of that article is provided below.

Fuel cell devices are NOT immune to the diabetic ketoacidosis effect:

An Experiment - Methodology & Results

Using a GUTH Model 34C simulator, I created an Alcohol Standard Solution using 500 mL of distilled water and added about 0.5 mL of 40% ethanol. After allowing the solution to come to equilibrium, I tested the resulting vapors which produced on a recently calibrated Intoximeter FST. My result was as expected and was reported at 0.024 g/dL. This became my baseline solution.

To this solution, I added 0.5 mL of lab grade Isopropanol. Again, after allowing the mixture to come to equilibrium, the vapor was introduced into the same Intoximeter FST. The reading generated by a combination of both the Isopropanol and Ethanol was 0.071 g/dL.

To this solution, I added 0.5 mL of lab grade Acetone. The same equilibrium wait time was completed, and the vapor introduced directly into the Intoximeter FST. The reading generated by the mixture of all three chemicals (Acetone, Isopropanol and Ethanol) was 0.077 g/dL.

For interest’s sake, I increased the amount of Isopropanol by another 0.5 mL, for a total of 1.0 mL Isopropanol, 0.5 mL Acetone and 0.5 mL Ethanol. The usual equilibrium time was allowed, and the vapor also introduced into the Intoximeter FST. The reading generated was 0.124 g/dL.

This is exactly .100 g/dL HIGHER as a false positive reading.

Table 2 might better summarize my findings:

Practice Tip:
Many people don't know that they are pre-diabetic, or emerging as diabetic patients. I can not tell you how many times I've heard a story along the lines of, "My client worked all day without taking even a lunch break. He stopped for a single glass of wine and some appetizers at a fund-raiser on the way home. How could he have blown a .17 reading on a single glass of wine?" We get the client medically assessed, an low and behold, he's an emerging diabetic, uncontrolled at the time of the event.

Response to diabetic ketoacidosis by the Intoxilyzer 9000:

I had an opportunity to inspect and review an Intoxilyzer 9000, and its response to interferent chemical. Using the same methodology as for the fuel cell device testing described above, I obtained a series of readings on the Intoxilyzer 9000 as follows:

In short, the Intoxilyzer 9000 correctly identified the presence of an infrared absorbing substance OTHER than ethanol. Further study is needed to determine the 9000’s response to other potentially interfering chemicals.

NOTE: The 9000 often displayed an incorrect, falsely elevated reading on the display screen as a preliminary result. However, the final display result, and that printed on the test card, was INTERFERENT. I am concerned that the operator may interpret the preliminary results as a correct representation of the Breath Alcohol Concentration (BrAC) of the test subject, and try and report these readings to the court. As silly as this sounds - reporting a preliminary result that is ultimately disregarded by the instrument itself - does occur. I've testified about these sorts of issues on more than one occasion.

For complete details, see the Counterpoint article in Volume 2, Issue 2, Article 4:

Response to diabetic ketoacidosis by the DataMaster DMT:

I had an opportunity to replicate my experiments above on the DataMaster DMT. In all cases, the DMT correctly identified the interference chemical and aborted the breath test process. In no cases were the preliminary results inflated or printed on the final test card. Even with extremely low levels of contamination, the DMT aborted the testing process and identified the presence of an interfering substance.

Final thoughts:

I am left to reasonably conclude, as have other researchers, that a combination of infrared absorbing substances in the test chamber with levels of ethanol may falsely over-report the true BAC level and may do so without triggering the interferent detector algorithm. The reported incidence of an interferent may vary among jurisdictions depending upon the threshold levels set in the acetone detect or subtract algorithm. IT IS ALSO DEPENDENT ON THE DEVICE USED. Typically, older devices are more susceptible to false-positive BrAC readings due to the effects of interfering substances.

As such, persons routinely displaying symptoms caused by uncontrolled blood ketone or blood glucose levels are poor candidates for breath alcohol testing. Person who are fasting, and emerging, newly diagnosed or uncontrolled diabetics will have even greater instabilities due to the nature of their uncontrolled and shifting blood glucose levels and corresponding blood ketone levels and are even worse candidates for accurate and reliable breath alcohol test results.

Return to the Foundational Skills I-4 Table of Contents

Or, go to the next article in the Foundational Skills I course.

Send me your questions or comments:

Comments and questions will be posted here with their responses:

For further study:

Bailey, D., Detection of Isopropanol in Acetonemic Patients Not Exposed to Isopropanol, Clinical Toxicology, 28(4), 1990, Pages 459-466.
Dubowski, K., Absorption, Distribution and Elimination of Alcohol: Highway Safety Aspects, Journal of Studies on Alcohol, Supplement No. 10, July 1985.
Dubowski, K.M., Acceptable Practices for Evidential Breath-Alcohol Testing, Center for Studies of Law in Action, Borkenstein Course Materials, Indiana University, May 2008.
Dubowski, K.M., The Technology of Breath-Alcohol Analysis, U.S. Department of Health and Human Services, Prepared for The National Institute on Alcohol Abuse and Alcoholism, 1991.
Friel, P.N., Bear, J.S. and Logan, B.K., Variability of Ethanol Absorption and Breath Concentrations During a Large-Scale Alcohol Administration Study, Alcoholism: Clinical and Experimental Research, Volume 19, Number 4, August 1995, Pages 1055-1060.
Jones, A.W., Variability of the Blood: Breath Ratio in vivo., Journal Studies of Alcohol 39, 1978, pages 1931-39.
Jones, A.W. and Andersson, L., Influence of Age, Gender and Blood-Alcohol Concentration on Disappearance Rate of Alcohol from Blood in Drinking Drivers, Journal of Forensic Science 1996; 41(6), pages 922-926.
Jones, A.W., Jonsson, K.A. and Neri, A., Peak Blood-Ethanol Concentration and the Time of Its Occurrence After Rapid Drinking on an Empty Stomach, Journal of Forensic Sciences, JFSCA, Vol. 36, No. 2, March 1991, pages 376-385.
Jones, A.W., Interfering Substances Identified in the Breath of Drinking Drivers with Intoxilyzer 5000S, Journal of Analytical Toxicology, Vol. 20, November/December 1996, Pages 523-527.
Jones, A.W. and Andersson, L., Comparison of Ethanol Concentrations in Venous Blood and End-Expired Breath During a Controlled Drinking Study, Forensic Science International 132 (2003), pages 18-25.
Jones, A.W., Andersson, L., Biotransformation of Acetone to Isopropanol Observed in a Motorist Involved in a Sobriety Check, Journal of Forensic Sciences, JFSVA, Vol. 40, No. 4 July 1995, Pages 686-687
Jones, A.W. & Summer, R., Detection of Isopropyl Alcohol in a Patient with Diabetic Ketoacidosis, The Journal of Emergency Medicine, Vol. 19, No. 2, 2000, Pages 165-168.
Jones, A.W. & Rossner, S., False-Positive Breath Alcohol Test After a Ketogenic Diet, International Journal of Obesity, (2007) 31, Pages 559-561.
Laasko, O., Pennanem, T., et al, Effect of Eight Solvents on Ethanol Analysis by Draeger 7110 Evidential Breath Analyzer, Journal of Forensic Science, Sept 2004, Vol 49, No. 5.
Norfold, G. & Quartly, C., Volatile Substances and their Potential to Interfere with Breath Alcohol Reading Instruments, Journal of Clinical Forensic Medicine (1997) 4, Pages 21-23.
Puttanna, A., and Padinjakara, R.N.K., Diabetic Ketoacidosis in Type 2 Diabetes Mellitus, Practical Diabetes, Vol. 31, No. 4, February 2014.
Simpson, G., Accuracy and Precision of Breath Alcohol Measurements for Subjects in the Absorptive State, Clinical Chemistry Volume 33, 1987, pages 73-75.