Best Practice in Breath Alcohol Testing
Part 2 - Operational considerations
Counterpoint Volume 2: Issue 2 - Article 1 (July 2017)
An article for participants in the myCAMprogram
Jan Semenoff, BA, EMA
Forensic Criminalist
Wait, deprivation & observation periods:
By whatever name they are referred to in your jurisdiction, a wait, deprivation, or observation period is a space of time before a breath sample is obtained where the test subject is closely and continuously observed to ensure the following:
- They do not burp, belch or regurgitate stomach contents, including air from the stomach, that could contaminate the mouth and oral pathway
- They do not consume any product, including water
- They have no foreign objects in their mouth that could contaminate the breath test process
|
It is important to recognize that many of the acts, practices and conditions surrounding breath testing were necessary, from a historical perspective, due to limitations of the breath testing technology of the day. The early breathalyzer (the original evidentiary breath testing device from 1954) had absolutely no ability to identify contamination from fresh mouth alcohol, recent regurgitation, or a burp. Any contamination in the mouth was exhaled and combined with the breath sample. The chemical reaction measuring the BrAC level couldn’t differentiate from the source of the ethanol. As a result, operators were trained to observe the test subject “closely and continuously” to make certain that this type of contamination didn’t affect the test results.
This limitation in technology is coupled in the practice of obtaining two samples that were within a certain value of one another – the “.02 agreement” … More on this in a bit.
|
Figure 1 - The original Breathalyzer, here a Model 900A, circa 1965. These devices had no ability to detect the presence of mouth alcohol contamination.
|
Figure 2 - Loose chewing tobacco, often in place DURING alcohol consumption, can become contaminated with ethanol that can create a false-positive on a breath test. The chewing tobacco should be removed before the deprivation, wait or observation period begins, and if necessary, the mouth should be rinsed with fresh water to cleanse to contaminate from the mouth prior to the observation period.
|
First, the operator should check the mouth of the test subject prior to beginning the wait period. There is no sense in performing a twenty-minute wait period with a test subject with contaminated chewing tobacco or gum in their mouths – the contaminate will remain and not naturally dissipate.
Next, the operator must closely and continuously observe the test subject for the entire deprivation or observation period. Now, I’m not suggesting that they need to sit and stare at the test subject intently – that would creep people out – but they at least need to be in the same room. Coming and going, performing a bunch of administrative tasks and paperwork, or otherwise having their own divided attention issues with regards to their arrest – in essence losing continuity on their evidence (the test subject’s breath), will lead to a potential source of contamination. |
The Residual Alcohol Detector:
Breath test devices and their operators rely upon the so-called “slope detector”, or Residual Alcohol Detection Systems to identify the presence of a burp or belch, or contamination form this source, including contamination from medical issues such as GERD. The issue of reliability of this technology deserves, at the very least, an entire article, or more, on its very own. In short, they are not reliable enough to warrant this degree of trust from the operator.
Breath test devices and their operators rely upon the so-called “slope detector”, or Residual Alcohol Detection Systems to identify the presence of a burp or belch, or contamination form this source, including contamination from medical issues such as GERD. The issue of reliability of this technology deserves, at the very least, an entire article, or more, on its very own. In short, they are not reliable enough to warrant this degree of trust from the operator.
|
If the operator notices a belch or a burp, however slight, the deprivation period must be started over again. If they leave the room, losing continuity of their evidence, they need to start the observation period over again from the start.
This is a framing issue as much as anything else – the officers simply lost continuity of their evidence. The chain of custody of a critical piece of evidence - your client's lack of mouth contamination, was lost. |
|
Practice Tip: |
Medical history:
|
This is a dual-edged sword. Operators often ask a series of questions regarding medical history along the lines of “current medications”, “recent visits to the doctor or dentist”, and “history of Diabetes”. It seems they ask these questions more to get a negative response (that they can use later as evidence) than an actual understanding of how these issues may affect the breath test results. Do you want your clients to answer these questions?
This is again a historical issue. The old Breathalyzer 900 was greatly affected by ketone levels in the test subject, or ketoacidosis on their breath. Some modern instruments are still affected, others not. I refer you to the Part 1 article, “Diabetes and Breath Alcohol Testing”, and the upcoming Part 2 article in Volume 2, Issue 2 on how Diabetes affects breath test devices. I would also add a series of questions about potential exposure to occupational chemicals, if I were creating breath sampling policy. This is again a topic for another article. |
Breath sampling criteria:
The breath samples received have to be suitable for analysis. This means free of potential contaminates, as discussed in both the sections on wait & deprivation periods, and medical conditions, above. It also means that the breath sample results (BrAC) should accurately reflect a true Blood Alcohol Concentration (BAC).
For a more complete discussion on this issue, I refer you to the articles, “Breath Sampling Criteria: Determining the Suitability of a Breath Sample” and “Establishing Reliability: Standard Acts, Practices & Conditions” in Counterpoint.
It also raises the issue of whether or not a sample of “deep lung air” is truly representative of the test subject's BAC…
For a more complete discussion on this issue, I refer you to the articles, “Breath Sampling Criteria: Determining the Suitability of a Breath Sample” and “Establishing Reliability: Standard Acts, Practices & Conditions” in Counterpoint.
It also raises the issue of whether or not a sample of “deep lung air” is truly representative of the test subject's BAC…
Agreement between samples
|
Let’s pretend that YOU are a newly qualified breath test operator. You have just obtained a breath sample from your test subject. It appears as if the subject had no problem providing the sample. It appears as if the breath test device is in correct working order – it operated as per usual with no error messages or problems. It appears as if the numerical result is not unusual (that is to say, you didn’t get a result of .02 with a person who appears gutter-puking, stinking drunk, and conversely, you didn’t get a reading of .350 with a person who looks completely sober, in two extreme examples).
|
You are simply left with a BrAC of reading X. How do you know it is correct? How do you know it is representative of the true BAC level? In scientific terms, how can you claim the breath reading is accurate, precise and reliable? I refer you to the article, “Accuracy, Precision & Reliability” so you can properly differentiate those three terms.
Short answer – you can’t. It could be accurate and reliable, or maybe totally inaccurate and unreliable. You just don’t know which. You need a second breath sample reading…
So, being a prudent breath test operator, you wait the required period of time, still keeping the person under direct and continuous observation (you can’t risk contamination for the next test, as well). A second sample is received, again with no apparent issues of reliability. In terms of sample agreement, you have only two possibilities:
- The second sample agrees within a certain range with the first sample, or
- The second sample is outside the acceptable range of agreement with the first sample.
Sample agreement:
In most jurisdictions, the breath test results have to agree within a certain margin – typically within 0.02 grams (20 milligrams) of one another. So, a set of readings of .09 and .10 (.01 grams or 10 milligrams apart) would be considered in agreement. If the two readings were .09 and .11 (now .02 grams or 20 milligrams apart), they would still be considered in agreement. Some jurisdictions require +/- 10% of one another. Some require a different degree of agreement at different reported BrAC levels.
The situation gets a bit more complicated when we speak about truncating the readings. To truncate is to round down to the next lowest integer. For example, a reading of 0.089 grams would be truncated, or rounded down, to 0.080 grams, or perhaps simply called 0.08 grams. Think of truncating as simply lopping off the last digit.
But, if you are truncating, a reading of .129 and a reading of .100 would be in agreement, because the .129 is truncated to .12, and the .100 is truncated to .10 – with .12 and .10 being within .02 of one another – yet we know that they are actually .029 apart, which is certainly greater than .020 apart… See the problem?
In most jurisdictions, the breath test results have to agree within a certain margin – typically within 0.02 grams (20 milligrams) of one another. So, a set of readings of .09 and .10 (.01 grams or 10 milligrams apart) would be considered in agreement. If the two readings were .09 and .11 (now .02 grams or 20 milligrams apart), they would still be considered in agreement. Some jurisdictions require +/- 10% of one another. Some require a different degree of agreement at different reported BrAC levels.
The situation gets a bit more complicated when we speak about truncating the readings. To truncate is to round down to the next lowest integer. For example, a reading of 0.089 grams would be truncated, or rounded down, to 0.080 grams, or perhaps simply called 0.08 grams. Think of truncating as simply lopping off the last digit.
But, if you are truncating, a reading of .129 and a reading of .100 would be in agreement, because the .129 is truncated to .12, and the .100 is truncated to .10 – with .12 and .10 being within .02 of one another – yet we know that they are actually .029 apart, which is certainly greater than .020 apart… See the problem?
Now, here is where things get a bit tricky. By getting the second sample, and having it basically agree with the first reading (it is within the acceptable range of agreement), you have increased the overall reliability of both sample readings obtained, but you still can’t claim that they are totally reliable, and that the reading is an absolute indicator of the true BAC level. All you can really claim is that the samples agree within the desired parameters.
And if the first and second samples DO NOT agree within the acceptable parameters? Then the operator is required in most areas to wait the specified period of time between samples, and perform a third test. Then, the results of sample 1 can agree with sample 3, or sample 2 can agree with sample 3. We simply assume one of the samples is incorrect, either too high, or two low. Regardless of the agreement or disagreement of the results, we are left with many questions:
- What issue does truncation play, beyond what we’ve just begun to discuss here?
- Does agreement mean the breath test results are correct? Are they accurate? Are they reliable?
- Does agreement mean that the deprivation and observation period were carried out correctly?
- Can I have agreement between two unreliable breath test results? Does agreement imply reliability?
- What role does the slope detector (Residual Alcohol Detection System) play in agreement?
- What role does the preliminary digital display of results play in agreement? Can agreement be manipulated?
- Can two readings agree if the test subject has GERD? Is occupationally exposed to a volatile hydrocarbon?
- What roles do wait and deprivation periods play in agreement?
- If the time between the two tests is only 5 minutes, or between two agreeing tests 10 minutes, does that mean they are accurate and reliable?
- If the two tests are in agreement, does that necessarily imply that the observation and deprivation period was done correctly?
- Can two unsuitable samples agree with one another?
The issue of agreement raises so many points of contention and concern, all I can say is that we will need to address them in another complete article.
The importance of Calibrations, & Calibration Checks
Why do you need to perform a Calibration Check in a breath test sequence? What purpose does this serve?
|
Here is a bizarre hypothetical:
You just fulfilled your life-long dream and bought a candy factory with the intention of producing fancy designer marshmallows. You need a machine that weighs out 1.000 kilograms of mini-marshmallows for each bag sold, each that weighs exactly one gram. It would be ludicrous to pay employees to count 1000 mini-marshmallows into each bag. You wouldn’t make any sort of profit. So, you hire an engineering firm to build a mini-marshmallow counting machine to place exactly 1000, or one kilogram worth of mini-marshmallows in each bag for sale. How can you verify: |
|
A) That you are not cheating your customers by delivering only 900 mini-marshmallows (-10%), or
B) You are not losing money by delivering 1100 mini-marshmallows that weigh 1.10 kg (+10%)?
Now, you might say that the error rate is only +/- 10% of our target value of one kilogram, but multiplied by a thousand bags an hour, the error rate would add up. The obvious answer is to do one of two things:
So, you decide to weigh the bags of mini-marshmallows as they come down the line.
B) You are not losing money by delivering 1100 mini-marshmallows that weigh 1.10 kg (+10%)?
Now, you might say that the error rate is only +/- 10% of our target value of one kilogram, but multiplied by a thousand bags an hour, the error rate would add up. The obvious answer is to do one of two things:
- Count the marshmallows in randomly selected bags to ensure there are 1000 mini-marshmallows in each bag, or
- Weigh the marshmallows as they are deposited in each bag to ensure you are delivering exactly 1.000 kilograms of product.
So, you decide to weigh the bags of mini-marshmallows as they come down the line.
The Calibration Check:
We need to check the calibration of our weighing system, and we would do this by occasionally stopping the production line, and placing a 1.000-kilogram weight, known, traceable, and standardized, on the mini-marshmallow dispenser. The weigh scale reads exactly 1.000 kilograms. We now know that:
a) We have 1.000 kilograms of mini-marshmallows in each bag, and
b) Based on their weight, we have 1000 mini-marshmallows in each bag.
Our calibration check has verified the operation of our mini-marshmallow dispensing device. The weight we used was traceable, so we can prove the weight of the marshmallows, if required.
a) We have 1.000 kilograms of mini-marshmallows in each bag, and
b) Based on their weight, we have 1000 mini-marshmallows in each bag.
Our calibration check has verified the operation of our mini-marshmallow dispensing device. The weight we used was traceable, so we can prove the weight of the marshmallows, if required.
Calibration:
|
The next week, we checked the calibration of our dispenser/weigh scale again. This time our 1.000-kilogram weight standard was measured at only .950 kilograms... but it is incorrectly saying 1.000 kilograms. In essence, the marshmallow counting machine must be ADDING 50 marshmallows to each bag to come up with an incorrect weight of 1.000 kilograms. So, in fact, we are selling 1.050 kg or 1050 marshmallows per bag to make up for the discrepancy in the weight scale.
We don’t want to lose 50 mini-marshmallows, or 50 grams of marshmallows, with each bag sold. The machine is reading outside of our allowable tolerance (we don't really care about +/- 10 mini- marshmallows). So, we correct the weigh scale’s value back to 1.000 kilograms. We have corrected, or calibrated the scale. Think of calibrations as a verb – an action to be performed. To put this analogy to rest, know that +/- 10% in a breath test device (the typical calibration accuracy required in most jurisdictions) is equivalent to +/- 100 marshmallows (900 – 1100) per bag. |
Easy enough to do with a mini-marshmallow counting or weighing device. We can count random bags of marshmallows, or we can weigh each bag as it comes down the conveyor. Breath test results are a little more difficult. You can’t count the molecules of ethanol in the sample. You could capture the sample and analyze it independently, but this is time consuming, expensive, and too much of a hassle (in fact, replicate-capture samples used to be performed, but are now a thing of the past).
The calibration check performed by a breath test is just like placing a known and traceable 1.000-kilogram weight on the scale. Instead, we use a sample of either wet or dry vapor, laden with a known amount of alcohol to check the calibration of the device. If we get a reading that is within our acceptable tolerance, we can say with a degree of certainty that the measurement system of the breath test device is accurate – it is hitting the target value.
See “The Calibration of Breath Alcohol Testing Devices” Counterpoint Volume 2, Issue 1, Article 3 for more on calibrations and calibration checks. I promise, no more mini-marshmallows…
The calibration check performed by a breath test is just like placing a known and traceable 1.000-kilogram weight on the scale. Instead, we use a sample of either wet or dry vapor, laden with a known amount of alcohol to check the calibration of the device. If we get a reading that is within our acceptable tolerance, we can say with a degree of certainty that the measurement system of the breath test device is accurate – it is hitting the target value.
See “The Calibration of Breath Alcohol Testing Devices” Counterpoint Volume 2, Issue 1, Article 3 for more on calibrations and calibration checks. I promise, no more mini-marshmallows…
The necessity of routine & annual maintenance:
|
Ok, I lied… a year later, your candy factory starts producing the occasional, random green marshmallow, and not because they are the minty flavoured ones sold on St. Patrick’s Day... Yuck...
The Food & Drug Inspector determines that the vat used to melt the confectioners’ sugar, cornstarch and gelatin in your factory has a bad growth of E-coli, and orders a recall. Maintenance records kept as normal business records by your firm indicate that the vat has never been cleaned, never been disinfected, in fact… you get the picture. |
Breath test devices need annual maintenance. Fuel cells drift. Sample chambers become contaminated with mildew and mold from moist exhaled human breath containing bacteria being stored within and heated to about 47-50 degrees Celsius in a moist environment. Think of the potential impact of mold and green spots on the inside of the test chamber, and the breath test results they could produce. Are the readings reliable under these conditions? Are the test results obtained under standard operating conditions? Are the operating conditions sub-standard? Gaskets rupture. O-rings lose their seal. Heating blankets drift in temperature… Again, you get the picture.
The problem I am beginning to see is that various jurisdictions are starting to cut costs by eliminating annual maintenance and calibration procedures. Some units are going year after year without any sort of cleaning, maintenance or routine calibration. They are run until they fail, and only then are they repaired, recalibrated and returned to service. In the meantime, they produce numerical results. Now, you might hear that this is not a problem, because the calibration checks kept working within acceptable tolerance until failure occurred.
The problem I am beginning to see is that various jurisdictions are starting to cut costs by eliminating annual maintenance and calibration procedures. Some units are going year after year without any sort of cleaning, maintenance or routine calibration. They are run until they fail, and only then are they repaired, recalibrated and returned to service. In the meantime, they produce numerical results. Now, you might hear that this is not a problem, because the calibration checks kept working within acceptable tolerance until failure occurred.
|
Editors Note: The problem became worse during the COVID-19 pandemic when a number of state and government agencies simply stopped doing any routine maintenance and calibration, citing concerns for officer safety from contamination. Some jurisdictions took a very long time to get back to routine or annual maintenance and calibration systems. |
The problem is that the units have a tendency to drift. In review of calibration records, I rarely, if ever see them drift negatively. The units have a tendency, for reasons I can’t adequately explain, to drift with positive bias. That is, the readings get higher and higher, until they go out of range. However, this raises the possibility of test subjects who actually had a true BAC of .075 getting a reported reading of .085 occurs. I’ve probably written 80-100 calibration drift expert opinion letters after reviewing calibration and maintenance records, and attended court on similar matters about 40 times over the last ten years.
I once had a prosecutor argue in a question put to me, “Well, it’s not like your pickup truck just stops running just because you haven’t changed the oil in 30,000 miles?” |
That is the issue of routine maintenance and annual calibration, and in essence, the BrAC results delivered are the equivalent of the device’s “gas mileage.” You guessed it, maintenance and calibration issues will be a whole other article. Stay tuned…
Final thoughts:
As I discuss in the Counterpoint article “Accuracy, Precision & Reliability” the concept of reliability is a system’s concept. Reliability occurs when the system, as a whole, works together to produce accurate and precise results. So, best-practice in breath alcohol testing means that all of the issues discussed here meet minimum performance standards of acts, practices and conditions under which the breath tests are conducted:
- The room air is free of contaminates
- The Air Blanks are true zero air blanks, not floating
- There are no sources of radio frequency interference in the room or near the device
- The power supply is clean, not shared by other devices, and not subject to brown-out
- A proper wait, deprivation or observation was conducted, uninterrupted. If there is any interruption of this pre-test quality control check, the required time should start again. The operator should be in a close and continuous position to observe any burp, belches or regurgitation. If any of these are suspected, again, the observation period should begin again.
- The breath samples received were suitable for analysis. They were not contaminated, or potentially contaminated, as described above. They were of the same duration (you can't compare the BrAC results of a 5-second sample with the results of a 15-second sample provided a few minutes later).
- The samples closely agreed with one another, regardless of the rather wide and arbitrary range of 0.02 grams as required in most jurisdictions.
- A Calibration Check was performed, with the results in range with all tests performed.
- Annual Calibrations were performed, as a part of a comprehensive annual maintenance protocol
Send me your questions or comments:
Comments and questions will be posted here with their responses:
Comments and questions will be posted here with their responses:
For further study:
- 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., Alcohol Determination in the Clinical Laboratory, American Journal of Clinical Pathologists, Volume 74, No. 5, Pages 747-750, 1980.
- Dubowski, K.M. and Caplan, Y.H., Alcohol Testing in the Workplace, Chapter 19, from Medicolegal Aspects of Alcohol, 3rd Edition, edited by Garriott, J.C., Lawyers & Judges Publishing Co, 1996.
- Dubowski, K.M. and Caplan, Y.H., Alcohol Testing in the Workplace, Chapter 20, from Medicolegal Aspects of Alcohol, 4th Edition, edited by Garriott, J.C., Lawyers & Judges Publishing Co, 2003.
- Dubowski, K.M., Quality Assurance in Breath-Alcohol Analysis, Journal of Analytical Toxicology, Vol. 18, Oct 1994.
- Gullberg, R. G., Breath Alcohol Measurement Variability Associated with Different Instrumentation and Protocols, Forensic Science International 131 (2003) 30-35.
- Jones, A. W., Concerning Accuracy and Precision of Breath-Alcohol Measurements, Clinical Chemistry, 33/10, 1701-1706 (1987).
