Forcon: Forensic Consulting

Forensic Science



Breath Testing




Historical Legal Issues

Current Issues



Email Forcon

Approved breath-test instruments

Accuracy and precision considerations

All breath testing equipment, whether approved for roadside screening or evidential purposes, is subject to measurement error or uncertainty. The main sources of this uncertainty are analytical variability and biological considerations. As a result, breath test devices are no different than any other quantitative measuring system used by scientists, engineers, technologists and technicians.

The uncertainty is deemed acceptable when it is reproducible and small with respect to the measured result and unacceptable when it is large and not reproducible. The acceptability of an error level is most often determined by those who are practitioners in a given field, after giving consideration to all factors that may influence an analytical result.

Precision of measurement

Several elements form the basis for accepting or rejecting the uncertainty level produced by a certain method or instrument. These are things like the criteria used to select samples, how instruments are designed, how data is manipulated and reported and the purpose of the analysis and the significance of the intended results.

For a given measurement, there are two elements of concern. The first is the precision of a measurement which reflects the reproducibility (variability) of the method of analysis. It is usually determined by the repeated analysis of a standard substance, such as a standard alcohol solution, in the case of breath testing.

Accuracy of measurement

The second consideration is accuracy which denotes the proximity of a measurement result to the true value. In terms of breath testing, 10 breath readings in succession, each yielding a result of 100 mg%, would be considered to be highly precise. They would also be highly accurate if the true blood alcohol level (direct measurement of alcohol in blood) were also 100 mg%.

On the other hand, 10 breath readings in succession yielding results of 100 mg% would be considered highly precise, but highly inaccurate if the true blood alcohol level at the time of measurement had been 150 mg% or 50 mg%. It follows therefore that methods can be highly precise and highly inaccurate at the same time.

Breathalyzer models 900 and 900A

It is accepted in the forensic scientific community that the precision of an approved instrument is ± 10 mg%.For the Breathalyzer models 900 and 900A, this precision is reflected in various ways. Firstly, the BAC scale is graduated in increments of 10 mg%. Readings to the nearest 1 mg% must be interpolated by the operator. Secondly, a tolerance of ± 10 mg% is allowed for the blank test, as well as the standard alcohol test. Values obtained outside this range require the operator to start over.

A precision tolerance of ± 10 mg% also applies to the Intoxilyzer 5000C. This is affirmed on the manufacturer's computer print-out card which states, "calibration checks must be 0.09 g% to 0.110 g%." Finally, the practice of truncation further supports the precision considerations outlined above.

Truncation of breath readings

Truncation is the practice of rounding the breath test values down to the nearest 10 mg%. Truncation was not adopted to simply lower a given result or to provide "some benefit of doubt" to an accused person. It simply reflects a well-established scientific principle.

A precision tolerance of ± 10 mg% is deemed to be necessary for a variety of reasons:

  1. The instrument used to measure a breath sample has a series of tolerances related to how it captures a breath sample, measures the signal produced by the alcohol contained in the human breath, manipulates the data and reports the results;
  2. The standard alcohol solution used for the calibration check is independently formulated with a certain tolerance;
  3. The simulator/equilibrator used is subject to certain temperature tolerances and manner of use, unique to an operator;
  4. Instrumental service and preventative maintenance would also play a role with respect to precision.

No measuring system is absolute in terms of the values reported because of the inherent analytical variability. Even a measurement with a ruler is subject to precision considerations.

Accuracy depends upon how well a measuring device agrees with the true blood alcohol level (determined by actual analysis of a blood sample). This aspect depends upon several biological variables.

Quality of breath sample

In the respiratory system, air moves back and forth through the same set of tubes, hence there is continuous mixing of inspired and expired air. Air passes from the nose to the trachea to two tubes (one for each lung) called bronchi.

The bronchi divide into several subdivisions, which end in a large number of tubules. At the end of the tubules are the alveoli. Each lung contains several million alveoli. Surrounding the alveoli is a fine network of capillary blood vessels.

Deep lung-air required

Exchange of gases, including alcohol, between the blood and the air in the lung occurs in the alveoli. As a result, in order to obtain a good correlation between blood and breath alcohol levels, deep lung-air must be analyzed by the breath tester.

The vital capacity (volume of air that can be expelled without collapsing the lung) of a test person may range from 7 L for a trained athlete to less than 1 L for persons with severely impaired lung function. If a high minimum volume is chosen, as the operational parameter for a breath testing device, there is a danger that many subjects will not be able to provide a valid breath sample. If the minimum volume chosen is too low then the correlation between breath alcohol values and blood alcohol levels will be poor. In this instance, breath predicted blood alcohol levels will be lower than the true blood alcohol level.

The more modern instruments have software algorithms that monitor the breath sample as it is provided. Sampling characteristics such as breath pressure, time, flow rate, exhalation volumes, rate of change in alcohol concentration, as well as certain other features may be monitored to ensure compliance with those specifications deemed by a given manufacturer to be crucial to the suitability of a given sample. Different manufacturers may use different ways to comply with the "deep lung air" requirement which is so necessary to proper breath testing procedure.

Variability in breath-test results

The breathing technique employed just prior to breath testing can also cause variability in breath test results. Hyperventilation immediately prior to sampling has been shown to cause a decrease in the breath alcohol concentration of up to 20%. On the other hand, it has also been reported that a subject who holds his or her breath for a short period of time before exhalation can increase the alcohol content in exhaled air by up to 15%. As a result, it is clear that the quality of breath samples may cause significant differences between two readings taken on a single instrument, as well as readings taken from two different approved instruments.

Breath temperature

Breath testing for alcohol is based upon the principle that alcohol is excreted unchanged in human breath, because of equilibrium conditions between lung capillaries and pulmonary alveoli. At constant temperatures the amount of alcohol in the breath is proportional to the amount of alcohol in the blood, obeying the scientific principle called Henry's Law.

Obviously, the constant temperature refers to a body temperature of 37°C. As a result, if the body temperature fluctuates, even slightly, the relationship between breath and blood alcohol content will be altered.

End-expiratory breath temperatures

The temperature of expired air is always below 37°C because the air loses its heat as it passes through and out of the respiratory system. It has been shown that end-expiratory breath temperatures can vary 32.4-35.7°C.

Breath testing instruments are based upon a breath temperature of 34°C. A lower body temperature would result in less alcohol in a given volume of breath and a higher temperature would result in more alcohol.

It has been shown in a study that a breath testing instrument will overestimate the true BAC by about 8.6% per °C increase in body temperature (hyperthermia). In a separate study, it was demonstrated that a breath testing device will underestimate the true BAC by about 6.9% per °C decrease in body temperature (hypothermia).

Time difference between the two tests

In criminal matters, two breath samples are taken a minimum of 15 minutes apart. As should be obvious from the previous comments, alcohol is continually being absorbed, distributed, metabolized and excreted. As a result, it is expected that blood alcohol levels will change slightly over a 15-30 minute interval.

One can not conclude on the basis of two readings if a subject's blood alcohol content was increasing or decreasing. Several readings in succession are necessary to make that determination.

Intoxilyzer 5000C vs. Breathalyzer

Given the implementation of newer more modern breath testing devices, such as the Intoxilyzer 5000C, it is not uncommon to hear evidence in criminal proceedings that the instruments are more accurate than the Breathalyzer. This is not correct.

Comparison testing of these devices has demonstrated accuracy and precision parameters that are very similar. Indeed, the Breathalyzer was used as a reference instrument when the Intoxilyzer 5000C was first evaluated as a candidate for "approved instrument" status.

Readings can differ by 29 mg%

It is not uncommon to observe a difference between two truncated readings of ± 20 mg%. The Alcohol Test Committee of the Canadian Society of Forensic Science (Principle Scientific Advisor to the Federal Department of Justice with respect to Criminal Code matters) has stipulated that this difference is acceptable on scientific grounds when prosecuting alcohol-related driving offenses.

In fact, two actual readings can differ by as much as 29 mg% to allow for a truncated difference of ± 20 mg%. The practice of allowing a difference of ± 20 mg% lends further support to the inherent variability in breath samples.

Timing of BAC important

In some drinking scenarios, the BAC at the time of driving is lower than the readings obtained later. This phenomenon, referred to as the "rising curve," warrants consideration when low readings are involved.

Since breath tests are conducted up to two hours after the time of an alleged offense, a given subject's BAC will likely have changed in the interim. The BAC at the time of the alleged offense could have been higher or lower than the reading obtained later.

BAC can increase by 20 mg% or more after the time of an alleged offense. As a result, caution should be exercised with respect to the significance of a BAC taken up to two (2) hours after an alleged offense, when the readings obtained are only slightly in excess of the legal limit. A difference of 20 mg% would represent the non-absorption of one drink of alcohol or less for a significant number of persons within a given population.

Accuracy crucial

It is clear that accuracy and precision considerations become very important at BAC levels that exceed the legal limit only slightly. These concerns are evident in the standards developed by the Alcohol Test Committee of the Canadian Society of Forensic Science and reinforced by the practice whereby roadside screening devices are calibrated to fail at 100 mg% or higher.

The inherent variability in breath samples when combined with the passage of time considerations should result in extreme caution when relying on BACs in the vicinity of 80-100 mg% for criminal purposes. Since a greater portion of this variability can be attributed to biological factors rather than analytical aspects, it is not the measuring device that is the limiting factor but rather the subject providing the sample.

Click here
                          for a printable version
Copyright © 2004 G. Kupferschmidt Consulting Services Ltd.