The Breathalyzer: Problems & Defenses in DUI cases

A breathalyzer (breath analyser) is a device for estimating blood alcohol content (BAC) from a breath sample. "Breathalyzer" is the brand name of a series of models made by one manufacturer of these instruments (originally Smith and Wesson, later it was sold to National Draeger), but has become a genericized trademark for all such instruments. In Canada, a preliminary non-evidentiary screening device can be approved by Parliament as an approved screening device and an evidentiary breath instrument can be similarly designated as an approved instrument. The U.S. National Highway Traffic Safety Administration maintains a "Conforming Products List" of breath alcohol devices approved for evidentiary use, [1] as well as for preliminary screening use.[2]

Origins
A 1927 paper produced by Emil Bogen[3] who collected air in a football and then tested this air for traces of alcohol discovered that the alcohol content of 2 litres of expired air was a little greater than that of 1cc of urine. However, research into the possibilities of using breath to test for alcohol in a person's body dates as far back as 1874 when Anstie made the observation that small amounts of alcohol were excreted in breath[4].

The first practical roadside breath testing device intended for use by the police was the drunkometer. The drunkometer was developed by Professor Harger in 1938. The drunkometer collected a motorist's breath sample directly into a balloon inside the machine. The breath sample was then pumped through an acidified potassium permangate solution. If there was alcohol in the breath sample, the solution changed colour. The greater the colour change, the more alcohol there was present in the breath.
The drunkometer was quite cumbersome and was approximately the size of a shoe box. It was more reminiscent of a portable laboratory.

In late 1927, in a case in Marlborough, England, a Dr. Gorsky, Police Surgeon, asked a suspect to inflate a football bladder with his breath. Since the 2 liters of the man's breath contained 1.5 ml of ethanol, Dr. Gorsky testified before the court that the defendant was "50% drunk".[5] Though technologies for detecting alcohol vary, it's widely accepted that Dr. Robert Borkenstein (1912–2002), a captain with the Indiana State Police and later a professor at Indiana University at Bloomington, is regarded as the first to create a device that measures a subject's blood alcohol level based on a breath sample. In 1954, Borkenstein invented his breathalyzer, which used chemical oxidation and photometry to determine alcohol concentration. Subsequent breathalyzers have converted primarily to infrared spectroscopy. The invention of the breathalyzer provided law enforcement with a non-invasive test providing immediate results to determine an individual's BAC at the time of testing. It does not, however, determine an individual's level of intoxication, as this varies by a subject's individual alcohol tolerance. Also, the BAC test result itself can vary between individuals consuming identical amounts of alcohol due to gender, weight, and genetic pre-disposition.

Breath analyzers do not directly measure blood alcohol content or concentration, which requires the analysis of a blood sample. Instead, they estimate BAC indirectly by measuring the amount of alcohol in one's breath. Two form factors are most prevalent. Desktop analyzers generally utilize infrared spectrophotometer technology, electrochemical fuel cell technology, or a combination of the two. Hand-held field testing devices, are generally based on electrochemical fuel cell analysis, and depending upon jurisdiction may be used by officers in the field as a form of "field sobriety test" commonly called PBT (preliminary breath test) or PAS (preliminary alcohol screening), or as evidential devices in POA (point of arrest) testing.

Consumer use
There are a number of models of breath alcohol analyzers that are intended for the consumer market. These hand-held devices are less expensive and can be much smaller than the devices used by law enforcement, and are less accurate, but can still give a useful indication of the user’s BAC. Almost all of these devices use less expensive alcohol sensors (frequently called "Taguchi cell" based sensors), which are not as stable as fuel cell sensors or infrared devices, and are more prone to false positives. Breath alcohol analyzers sold to consumers in the United States are required to be certified by the Food and Drug Administration[6], while those used by law enforcement must be approved by the Department of Transportation's National Highway Traffic Safety Administration.[7]

Manufacturers of over the counter consumer breathalyzers must submit a FDA 510(k) Premarket Clearance to demonstrate that the device to be marketed is at least as safe and effective, that is, substantially equivalent, to a legally marketed device (21 CFR 807.92(a)(3)) that is not subject to Premarket Approval (PMA). Submitters must compare their device to one or more similar legally marketed devices and make and support their substantial equivalency claims.[8]
Breath test evidence

A breathalyzer in action
The breath alcohol reading is used in criminal prosecutions in two ways. Unless the suspect refuses to submit to chemical testing, he will be charged with a violation of the illegal per se law: that is, it is a misdemeanor throughout the United States to drive a vehicle with a BAC of .08% or higher (.02% in most states for drivers under 21). One exception is the State of Wisconsin, where a first time drunk driving offense is normally a civil ordinance violation.[9] The breathalyzer reading will be offered as evidence of that crime, although the issue is what the BAC was at the time of driving rather than at the time of the test. Some jurisdictions now allow the use of breathalyzer test results without regard as to how much time passed between operation of the vehicle and the time the test was administered. The suspect will also be charged with driving under the influence of alcohol (sometimes referred to as driving or operating while intoxicated). While BAC tests are not necessary to prove a defendant was under the influence, laws in most states require the jury to presume that he was under the influence if his BAC was over .08% when driving. This is a rebuttable presumption, however: the jury can disregard the test if they find it unreliable or if other evidence establishes a reasonable doubt.

If a defendant refused to take a breathalyzer test, most states allow evidence of that fact to be introduced; in many states, the jury is instructed that they can draw a permissible inference of "consciousness of guilt." Many states also operate under "implied consent," meaning that anyone issued a driver's license in the state agrees to submit to a test of his or her breath, blood, or urine when requested by a law enforcement officer. Failure to submit to such a test may result in automatic suspension of his or her driver's license even if not convicted of drunk driving. Failure to submit to such a test may also serve to enhance the penalties for a drunk driving conviction. In drunk driving cases in Massachusetts and Delaware, if the defendant refuses the breathalyzer there can be no mention of the test during the trial.[citation needed]

Instruments such as the Intoxilyzer 5000 are known as "evidentiary breath tests" and generally produce court-admissible results.[citation needed] Other instruments, such as the SD-2 by CMI or the Alcosensor III by Intoximeters, are known as "preliminary breath tests", and their results, while valuable to an officer attempting to establish probable cause for a drunk driving arrest, are generally not admissible in court. Some states do not permit data or "readings" from hand-held PBTs to be presented as evidence in court. They are generally admissible, if at all, only to show the presence of alcohol or as a pass-fail field sobriety test to help determine probable cause to arrest. South Dakota does not permit data from any type or size of breath tester but relies entirely on blood tests to ensure accuracy.

Common sources of error
Breath testers can be very sensitive to temperature, for example, and will give false readings if not adjusted or recalibrated to account for ambient or surrounding air temperatures. The temperature of the subject is also very important.[citation needed]

Breathing pattern can also significantly affect breath test results. One study found that the BAC readings of subjects decreased 11 to 14% after running up one flight of stairs and 22–25% after doing so twice. Another study found a 15% decrease in BAC readings after vigorous exercise or hyperventilation. Hyperventilation for 20 seconds has been shown to lower the reading by approximately 32%. On the other hand, holding your breath for 30 seconds can increase the breath test result by about 28%.[citation needed]
Some breath analysis machines assume a hematocrit (cell volume of blood) of 47%. However, hematocrit values range from 42 to 52% in men and from 37 to 47% in women. A person with a lower hematocrit will have a falsely high BAC reading.

Failure of law enforcement officers to use the devices properly or of administrators to have the machines properly maintained and re-calibrated as required are particularly common sources of error.[citation needed] However, most states have very strict guidelines regarding officer training and instrument maintenance and calibration.
Research indicates that breath tests can vary at least 15% from actual blood alcohol concentration. An estimated 23% of individuals tested will have a BAC reading higher than their true BAC. Police in Victoria, Australia use breathalyzers that give a recognized 20 percent tolerance on readings. Noel Ashby, former Victoria Police Assistant Commissioner (Traffic & Transport), claims that this tolerance is to allow for different body types.[10]

Calibration
Most handheld breathalyzers use a silicon oxide sensor to determine the blood alcohol concentration. Without proper software calibration, the accuracy of these sensors degrades over time and with repeated use. The calibration process aims to focus the sensor's ability to detect an accurate reading. New advances in breathalyzer design allow some models to self-calibrate or easily replace the sensor module without the need to send the unit to a calibration lab.[citation needed]

Non-specific analysis
One major problem with older breathalyzers is non-specificity: the machines not only identify the ethyl alcohol (or ethanol) found in alcoholic beverages, but also other substances similar in molecular structure or reactivity.

The oldest breathalyzer models pass breath through a solution of potassium dichromate, which oxidizes ethanol into acetic acid, changing color in the process. A monochromatic light beam is passed through this sample, and a detector records the change in intensity and, hence, the change in color, which is used to calculate the percent alcohol in the breath. However, since potassium dichromate is a strong oxidizer, numerous alcohol groups can be oxidized by it, producing false positives.

Infrared-based breathalyzers project an infrared beam of radiation through the captured breath in the sample chamber and detect the absorbance of the compound as a function of the wavelength of the beam, producing an absorbance spectrum that can be used to identify the compound, as the absorbance is due to the harmonic vibration and stretching of specific bonds in the molecule at specific wavelengths (see infrared spectroscopy). The characteristic bond of alcohols in infrared is the O-H bond, which gives a strong absorbance at a short wavelength. The more light is absorbed by compounds containing the alcohol group, the less reaches the detector on the other side—and the higher the reading. Other groups, most notably aromatic rings and carboxylic acids can give similar absorbance readings.[11]

Interfering compounds
Some natural and volatile interfering compounds do exist, however. For example, the National Highway Traffic Safety Administration (NHTSA) has found that dieters and diabetics may have acetone levels hundreds and even thousand of times higher than those in others. Acetone is one of the many substances that can be falsely identified as ethyl alcohol by some breath machines. However, new machines like the Draeger Breathalyzer use technology that filters out substances like acetone.

A study in Spain showed that metered-dose inhalers (MDIs) used in asthma treatment are also a cause of false positives in breath machines.

Substances in the environment can also lead to false BAC readings. For example, methyl tert-butyl ether (MTBE), a common gasoline additive, has been alleged anecdotally to cause false positives in persons exposed to it. Tests have shown this to be true for older machines; however, newer machines detect this interference and compensate for it.[12] Any number of other products found in the environment or workplace can also cause erroneous BAC results. These include compounds found in lacquer, paint remover, celluloid, gasoline, and cleaning fluids, especially ethers, alcohols, and other volatile compounds.

Homeostatic variables
Breathalyzers assume that the subject being tested has a 2100-to-1 "partition ratio"[13] in converting alcohol measured in the breath to estimates of alcohol in the blood. If the instrument estimates the BAC, then it measures weight of alcohol to volume of breath, so it will effectively measure grams of alcohol per 2100 ml of breath given. This measure is in direct proportion to the amount of grams of alcohol to every 100 ml of blood. Therefore, there is a 2100 to 1 ratio of alcohol in blood to alcohol in breath. However, this assumed "partition ratio" varies from 1300:1 to 3100:1 or wider among individuals and within a given individual over time. Assuming a true (and legal) blood-alcohol concentration of .07%, for example, a person with a partition ratio of 1500:1 would have a breath test reading of .10%—over the legal limit.

Most individuals do, in fact, have a 2100-to-1 partition ratio in accordance with William Henry's Law (1803), which states that when the water solution of a volatile compound is brought into equilibrium with air, there is a fixed ratio between the concentration of the compound in air and its concentration in water. This ratio is constant at a given temperature. The human body is 37 degrees Celsius on average. Breath leaves the mouth at a temperature of 34 degrees Celsius. Alcohol in the body obeys Henry's Law as it is a volatile compound and diffuses in body water. To ensure that variables such as fever and hypothermia could not be pointed out to influence the results in a way that was harmful to the accused, the instrument is calibrated at a ratio of 2100:1, underestimating by 9 percent. In order for a person running a fever to significantly overestimate, he would have to have a fever that would likely see the subject be in the hospital rather than driving in the first place. Studies suggest that about 1.8% of the population have a partition ratio below 2100. Thus, a machine using a 2100-to-1 ratio could actually under-report. As much as 14% of the population has a partition ratio above 2100, thus causing the machine to overestimate the BAC.

Further, the assumption that the test subject's partition ratio will be average—that there will be 2100 parts in the blood for every part in the breath—means that accurate analysis of a given individual's blood alcohol by measuring breath alcohol is difficult, as the ratio varies considerably.

Variance in how much one breathes out can also give false readings, usually low. [14] This is due to biological variance in breath alcohol concentration as a function of the volume of air in the lungs, an example of a factor which interferes with the liquid-gas equilibrium assumed by the devices. The presence of volatile components is another example of this; mixtures of volatile compounds can be more volatile than their components, which can create artificially high levels of ethanol (or other) vapors relative to the normal biological blood/breath alcohol equilibrium.

Mouth alcohol
One of the most common causes of falsely high breathalyzer readings is the existence of mouth alcohol. In analyzing a subject's breath sample, the breathalyzer's internal computer is making the assumption that the alcohol in the breath sample came from alveolar air—that is, air exhaled from deep within the lungs. However, alcohol may have come from the mouth, throat or stomach for a number of reasons. To help guard against mouth-alcohol contamination, certified breath test operators are trained to carefully observe a test subject for at least 15-20 minutes before administering the test.

The problem with mouth alcohol being analyzed by the breathalyzer is that it was not absorbed through the stomach and intestines and passed through the blood to the lungs. In other words, the machine's computer is mistakenly applying the "partition ratio" (see above) and multiplying the result. Consequently, a very tiny amount of alcohol from the mouth, throat or stomach can have a significant impact on the breath alcohol reading.

Other than recent drinking, the most common source of mouth alcohol is from belching or burping. This causes the liquids and/or gases from the stomach—including any alcohol—to rise up into the soft tissue of the esophagus and oral cavity, where it will stay until it has dissipated. The American Medical Association concludes in its Manual for Chemical Tests for Intoxication (1959): "True reactions with alcohol in expired breath from sources other than the alveolar air (eructation, regurgitation, vomiting) will, of course, vitiate the breath alcohol results." For this reason, police officers are supposed to keep a DUI suspect under observation for at least 15 minutes prior to administering a breath test. Instruments such as the Intoxilyzer 5000 also feature a "slope" parameter. This parameter detects any decrease in alcohol concentration of .006 g per 210L of breath in 6/10th's of a second, a condition indicative of residual mouth alcohol, and will result in an "invalid sample" warning to the operator, notifying the operator of the presence of the residual mouth alcohol. PBT's, however, feature no such safeguard.

Acid reflux, or gastroesophageal reflux disease, can greatly exacerbate the mouth alcohol problem. The stomach is normally separated from the throat by a valve, but when this valve becomes herniated, there is nothing to stop the liquid contents in the stomach from rising and permeating the esophagus and mouth. The contents—including any alcohol—are then later exhaled into the breathalyzer.[15]

Mouth alcohol can also be created in other ways. Dentures, for example, will trap alcohol. Periodental disease can also create pockets in the gums which will contain the alcohol for longer periods. Also known to produce false results due to residual alcohol in the mouth, is passionate kissing with a intoxicated person. And recent use of mouthwash or breath freshener—possibly to disguise the smell of alcohol when being pulled over by police—contain fairly high levels of alcohol.

Testing during absorptive phase
One of the most common sources of error in breath alcohol analysis is simply testing the subject too early—while his or her body is still absorbing the alcohol. Absorption of alcohol continues for anywhere from 45 minutes to two hours after drinking or even longer. Peak absorption normally occurs within an hour; this can range from as little as 15 minutes to as much as two-and-a-half hours.[citation needed]

During this absorptive phase, the distribution of alcohol throughout the body is not uniform; uniformity of distribution—called equilibrium—will not occur until absorption is complete. In other words, some parts of the body will have a higher blood alcohol content (BAC) than others. One aspect of this non-uniformity is that the BAC in arterial blood will be higher than in venous blood (laws generally require blood samples to be venous). During peak absorption arterial BAC can be as much as 60 percent higher than venous.

Retrograde extrapolation
The breathalyzer test is usually administered at a police station, commonly an hour or so after the arrest. Although this gives the BAC at the time of testing, it does not by itself answer the question of what it was at the time of driving. The prosecution typically provides evidence of this in the form of retrograde extrapolation. Usually presented in the form of an expert opinion, this involves projecting the BAC backwards in time—that is, estimating the probable BAC at the time of driving by applying mathematical formula, commonly the Widmark factor. This process, however, has been the subject of considerable criticism.
Photovoltaic assay
The photovoltaic assay, used only in the dated Intoximeter 3000, is a form of breath testing rarely encountered today. The process works by using photocells to analyze the color change of a redox (oxidation-reduction) reaction. A breath sample is bubbled through an aqueous solution of sulfuric acid, potassium dichromate, and silver nitrate. The silver nitrate acts as a catalyst, allowing the alcohol to be oxidized at an appreciable rate. The requisite acidic condition needed for the reaction might also be provided by the sulfuric acid. In solution, ethanol reacts with the potassium dichromate, reducing the dichromate ion to the chromium (III) ion. This reduction results in a change of the solution's colour from red-orange to green. The reacted solution is compared to a vial of nonreacted solution by a photocell, which creates an electric current proportional to the degree of the colour change; this current moves the needle that indicates BAC.
Like other methods, breath testing devices using chemical analysis are somewhat prone to false readings. Compounds which have compositions similar to ethanol, for example, could also act as reducing agents, creating the necessary color change to indicate increased BAC.

Myths
A common myth is that breath testers can be "fooled" (that is, made to generate estimates making one's blood alcohol content appear lower) by using certain substances. An episode of the Discovery Channel's MythBusters tested substances usually recommended in this practice—including breath mints, mouthwash, and onion—and found them to be ineffective. Adding an odor to mask the smell of alcohol might fool a person, but does not change the actual alcohol concentration in the body or on the breath. Interestingly, substances that might actually reduce the BAC reading were not tested on the show. These include a bag of activated charcoal concealed in the mouth (to absorb alcohol vapor), an oxidizing gas (such as N2O, Cl2, O3, etc.) which would fool a fuel cell type detector, or an organic interferent to fool an infra-red absorption detector. The infra-red absorption detector is especially vulnerable to countermeasures, since it only makes measurements at particular discrete wavelengths rather than producing a continuous absorption spectrum as a laboratory instrument would do.

Products That Interfere With Testing
On the other hand, products such as mouthwash or breath spray can "fool" breath machines by significantly raising test results. Listerine, for example, contains 27% alcohol; because the breath machine will assume the alcohol is coming from alcohol in the blood diffusing into the lung rather than directly from the mouth, it will apply a "partition ratio" of 2100:1 in computing blood alcohol concentration—resulting in a false high test reading. To counter this, officers are not supposed to administer a PBT for 15 minutes after the subject eats, vomits, or puts anything in their mouth. In addition, most instruments require that the individual be tested twice at least two minutes apart. Mouthwash or other mouth alcohol will have dissipated after two minutes and cause the second reading to disagree with the first, requiring a retest. (Also see the discussion of the "slope parameter" of the Intoxilyzer 5000 in the "Mouth Alcohol" section above.)

This was clearly illustrated in a study conducted with Listerine mouthwash on a breath machine and reported in an article entitled "Field Sobriety Testing: Intoxilyzers and Listerine Antiseptic," published in the July 1985 issue of The Police Chief (p. 70). Seven individuals were tested at a police station, with readings of .00%. Each then rinsed his mouth with 20 milliliters of Listerine mouthwash for 30 seconds in accordance with directions on the label. All seven were then tested on the machine at intervals of one, three, five and ten minutes. The results indicated an average reading of .043 blood-alcohol concentration, indicating a level that, if accurate, approaches lethal proportions. After three minutes, the average level was still .020, despite the absence of any alcohol in the system. Even after five minutes, the average level was .011.

In another study, reported in 8(22) Drinking/Driving Law Letter 1, a scientist tested the effects of Binaca breath spray on an Intoxilyzer 5000. He performed 23 tests with subjects who sprayed their throats, and obtained readings as high as .81 — far beyond lethal levels. The scientist also noted that the effects of the spray did not fall below detectable levels until after 18 minutes.

References
1. ^ http://a257.g.akamaitech.net/7/257/2422/01jan20061800/edocket.access.gpo.gov/2006/pdf/E6-10258.pdf
2. ^ http://www.dot.gov/ost/dapc/testingpubs/20070131_CPL_ASD.pdf
3. ^ The Diagnosis of Drunkenness
4. ^ Professor Robert F.Borkenstein - An Appreciation of his Life and Work
5. ^ Mitchell, C. Ainsworth (March/April 1932). "Science and the Detective". The American Journal of Police Science (Northwestern University) 3 (2): 169–182. doi:10.2307/1147200. http://www.jstor.org/stable/1147200. Retrieved on 4 April 2008.
6. ^ FDA > CDRH > Product Classification Database Search
7. ^ Retrieve Pages
8. ^ http://www.fda.gov/CDRH/DEVADVICE/314.html
9. ^ WI DOT
10. ^ Jane Holroyd, Breathalyser's 20 per cent tolerance defended, Sydney Morning Herald, 16 May 2006
11. ^ Organic Chemistry Resources Worldwide
12. ^ Elsevier Article Locator
13. ^ Forensic-Evidence.com: Biological Evidence/ Breath Tests for Blood Alcohol Determination: Partition Ratio
14. ^ Quantitative measurements of the alcohol concentra...[Acta Physiol Scand. 1982] - PubMed Result
15. ^ Kechagias, et al., "Reliability of Breath-Alcohol Analysis in Individuals with Gastroesophageal Reflux Disease", 44(4) Journal of Forensic Sciences 814 (1999).