Factors contributing to the variability observed in duplicate forensic breath alcohol measurement.

Several factors contribute to the variability observed among repeated measurements of breath alcohol concentration. Identifying these factors and the magnitude of their contribution is the focus of this study. Large breath alcohol data sets consisting of duplicate test results from drivers arrested for driving while intoxicated were obtained from four jurisdictions: Sweden, Alabama, New Jersey and Washington State. The absolute difference between duplicate results were fitted to a multivariate linear regression model which included the following predictor variables: mean breath alcohol concentration, absolute exhalation time difference between repeated measurements, absolute exhalation volume difference, gender and age. In all data sets considered here, the breath alcohol concentration was the most statistically and practically significant predictor of absolute difference between the duplicate results. The next two most important predictors to enter models for all jurisdictions were exhalation volume difference and exhalation time difference. The maximum multivariate R² for any jurisdiction, however, was only 0.24, suggesting that other factors not considered here may be of importance. Two predictors over which the subject would have some influence included exhalation time and volume. When these were set at values expected to have maximum impact, the effect on duplicate test differences was very small, 0.008 g/210 L or less in all jurisdictions, indicating that subject manipulation of exhalation time and volume would have at most a very small systematic effect on estimated breath alcohol concentration. This study presents multivariate models useful for identifying the impact of five variables that may influence breath test variability.

New Zealand's breath and blood alcohol testing programs: further data analysis and forensic implications.

Paired blood and breath alcohol concentrations (BAC, in g/dL, and BrAC, in g/210 L), were determined for 11,837 drivers apprehended by the New Zealand Police. For each driver, duplicate BAC measurements were made using headspace gas chromatography and duplicate BrAC measurements were made with either Intoxilyzer 5000, Seres 679T or Seres 679ENZ Ethylometre infrared analysers. The variability of differences between duplicate results is described in detail, as well as the variability of differences between the paired BrAC and BAC results. The mean delay between breath and blood sampling was 0.73 h, ranging from 0.17 to 3.1 8h. BAC values at the time of breath testing were estimated by adjusting BAC results using an assumed blood alcohol clearance rate. The paired BrAC and time-adjusted BAC results were analysed with the aim of estimating the proportion of false-positive BrAC results, using the time-adjusted BAC results as references. When BAC results were not time-adjusted, the false-positive rate (BrAC>BAC) was 31.3% but after time-adjustment using 0.019 g/dL/h as the blood alcohol clearance rate, the false-positive rate was only 2.8%. However, harmful false-positives (defined as cases where BrAC>0.1 g/210L, while BAC< or =0.1g/dL) occurred at a rate of only 0.14%. When the lower of duplicate breath test results were used as the evidential results instead of the means, the harmful false-positive rate dropped to 0.04%.

Common Legal Challenges and Responses in Forensic Breath Alcohol Determination.

No other single criminal/civil offense connects the forensic science community together with the legal community more than driving while under the influence of alcohol (DUI). Statutory changes along with political pressure have caused DUI to be a very serious and expensive offense. The result has been an enormous increase in DUI defense challenges in recent years with the primary focus on breath alcohol measurement. This paper addresses several of the more common DUI defense challenges regarding breath alcohol analysis, along with suggested responses. Common DUI defense challenges include: uncertainty in breath alcohol results near statutory limits; interfering substances; storage of alcohol simulator solutions; simulator thermometer uncertainties; instrument repair history; Widmark computations; instrument software; program records; pre-exhalation observation time period; and concurrence between the officer's report and breath alcohol results. Most of these issues are addressed with the prudent construction of administrative rules and employing a forensically sound breath test protocol.

Breath alcohol analysis in one subject with gastroesophageal reflux disease.

A large number of people suffer from the heartburn symptoms associated with gastroesophageal reflux disease (GERD). Relatively little has been published on its potential for biasing a breath alcohol measurement. The present case describes an individual (white male, aged 23) who experimentally consumed 1.0 g/kg of an alcohol beverage and subsequently provided breath and blood samples for analysis. Breath expirograms were also collected following several different preexhalation breathing maneuvers. Shortly after the end of drinking the mean of replicate breath alcohol results exceeded that of the corresponding venous blood alcohol. A later paired comparison (during the postabsorptive phase) showed the blood alcohol to exceed the breath. None of the expirograms provided evidence that "mouth alcohol" due to gastroesophageal reflux had biased any test results. People with GERD can provide biased-free end-expiratory breath alcohol results where sound forensic practice is followed, which includes: 15-min. preexhalation observation, duplicate testing, instrumental detection systems, and trained alert operators who ask appropriate questions and watch for associated signs.

Statistical Considerations for Readers and Authors of the Forensic Science Literature.

The use of numerical information and statistical methods is increasing sharply among forensic scientists publishing research and presenting analytical results for peer and judicial review. The importance of their work should not be obscured by the tendency for complex statistical analyses to confuse and mislead. Some considerations that would assist in our statistical communication include: sound statistical thinking, clear numerical notation, details of data collection, sample size, analytical details, inferential methods and assumptions, confidence intervals, linear regression assumptions, graphs, etc. Sound statistical thinking should guide the research and communication rather than technical details. The informed judgment of the researcher should be clear, apart from the statistical methodology. Better descriptive methods that emphasize the differences in data rather than complex inferential techniques may alone greatly improve our statistical communication. Since numerical information can easily confuse and mislead those unprepared, careful attention to a few basic principles should enhance the effectiveness of statistical communication in the forensic science literature.

Enforcement of Zero Tolerance in the State of Washington - Evidence from Breath-Test Records.

Several studies have demonstrated the effectiveness of zero tolerance (ZT) laws that make it illegal for drivers younger than 21 years to have any measurable alcohol in their bodies when operating a vehicle. All 50 states in the United States now have such laws. However, there is some question as to the extent they are being enforced. Because it has a centralized file of all breath tests performed by police in that state, Washington provided a special opportunity to evaluate the extent of ZT enforcement. Analyses were performed on arrests and alcohol-related fatal crashes of drivers younger than 21 years using data from the Washington State Patrol Breath Test Section and from the National Highway Traffic Safety Administration Fatality Analysis Reporting System, respectively. The results indicate that low blood alcohol concentration citations increased and crashes were reduced for drivers younger than 21 years, but crashes for drivers older than 21 years also declined. This suggests that the decline in underage drinking drivers in fatal crashes could not be unequivocally attributed to the ZT law.

Methodology and Quality Assurance in Forensic Breath Alcohol Analysis.

Breath alcohol analysis has become widely established in the forensic science and legal communities. The increasingly serious consequences of a drunk driving conviction, however, requires that further attention be focused on improving quality assurance. Although computerized instrumentation with advanced technology has enhanced forensic interpretation and confidence, other important areas of measurement protocol and program details must receive equal attention. Measurement results are the product of a process and not simply an instrument. Confidence in results can occur only after showing the entire program is "fit-for-purpose". Forensic quality assurance results from a balanced consideration for instrumentation, protocol, administrative rules, record keeping, interpretation, communication, etc. The following review will discuss current breath alcohol instrumentation along with several program features important to forensic quality control. Statistical methods are also available to assist in the quantitative interpretation of results to ensure statistical control and fitness-for-purpose. Increasing public and political attention on drunk driving requires the highest possible standards for quality control applied from a total program perspective.

Reproducibility of within-subject breath alcohol analysis.

Random samples from normal distributions are an important assumption for many statistical methods. The present study evaluates this assumption with regard to quantitative breath alcohol analyses. Eight individuals (six male and two female) consumed alcoholic beverages and subsequently provided replicate (n ranging from 22 to 69) breath samples to an infrared breath alcohol instrument within short time intervals. The serially collected data were treated with several descriptive and inferential methods. Descriptive results among the eight individuals included: mean 0.0420-0.1175 g/210 L, SD 0.0008-0.0045 g/210 L and CV: 1.9%-4.7%. Statistical tests for normality showed seven of the distributions to be reasonably normal (p > or = 0.25) and the other marginal (p = 0.051). A test for runs about the median showed random results (p > or = 0.10) for four individuals and non-random (p < or = 0.01) for the other four. The results suggest an individual's breath alcohol measurement, when appropriately collected and analysed, should be considered a random sample from a normal within-subject distribution. The existing variability in breath alcohol analysis, due largely to biological and sampling considerations, is acceptably minimized to warrant forensic application.

Lack of effect of tongue piercing on an evidential breath alcohol test.

Defendants in several driving under the influence cases have asserted that the presence in the mouth of a metal stud through a hole pierced in the tongue invalidates the breath alcohol test because of the prohibition against foreign substances in the mouth, and because of the potential for the jewelry to retain alcohol and interfere with the breath test. Rates of mouth alcohol elimination were evaluated in two subjects with pierced tongues and in two control subjects. No differences in the mouth alcohol elimination patterns were observed. The 15 min alcohol deprivation period prior to the test ensures no effect from residual mouth alcohol. For the purposes of breath alcohol testing, oral jewelry should be treated in the same manner as dental work, and may be left in place during the test without affecting its outcome.

Lack of predictable site-dependent differences and time-dependent changes in postmortem concentrations of cocaine, benzoylecgonine, and cocaethylene in humans.

This study evaluated the stability of cocaine, benzoylecgonine, and cocaethylene in postmortem fluids in cases of cocaine-related death. Femoral and ventricular blood and cisternal cerebrospinal fluid were collected soon after death and again at the time of autopsy. In addition, iliac blood was collected at autopsy. There were no consistent patterns of site-specific differences for any of the analytes, and the central compartment showed both higher and lower concentrations than the peripheral. There was no consistent pattern of direction or magnitude of change in the concentrations with respect to time for any of the analytes. This is consistent with anecdotal reports from other workers and is believed to be a result of competing processes of tissue release and chemical and enzymatic degradation of the analytes. Postmortem cocaine and metabolite concentrations in blood are not necessarily reflective of the perimortem concentrations and should not be the primary consideration in determining the cause of death in suspected cocaine-related deaths.

Simple versus sophisticated models of breath alcohol exhalation profiles.

For medicolegal purposes, breath alcohol content is typically determined from an end-expiratory sample. Measurements obtained by this method necessarily underestimate the alveolar breath alcohol content, and therefore underestimate the blood alcohol content. We suggest and analyse an improved paradigm which uses the entire time-series of breath alcohol measurements during exhalation, not simply the last recorded value. We present two mathematical models for the exhaling lung, and discuss the implications of each for more accurate and therefore more reliable breath alcohol measurement.

Repeatability of replicate breath alcohol measurements collected in short time intervals.

The effect of short time interval sampling between replicate breath alcohol samples has been investigated. The results from 10 samples, which were collected approximately one minute apart from eight individuals and approximately 20 seconds apart from one individual, were evaluated by simple linear regression. The regression coefficient (slope) and its standard error were evaluated for the presence of any trend in alcohol depletion. Other statistical analyses were also included in this assessment. All nine subjects had linear regression coefficients for the end-expiratory results that were not significantly different from zero (P > 0.05). In view of the respiratory physiology, there does not appear to be any measurable depletion of breath alcohol concentration due to sampling intervals as short as one minute.

Guidelines for estimating the amount of alcohol consumed from a single measurement of blood alcohol concentration: re-evaluation of Widmark's equation.

This article deals with the pharmacokinetics of ethanol and the reliability of estimating the amount of alcohol ingested from a single measurement of a person's blood alcohol concentration (BAC). Blood alcohol curves were plotted for 108 male subjects after they drank various doses of ethanol (0.51-0.85 g/kg body weight). The rate of disappearance of ethanol from the blood (beta-slope) and the apparent volume of distribution of ethanol (Widmark's rho factor, rho) were calculated for each subject; the mean beta-slope was 13.3 mg/dl/h (SD = 2.0), and the mean rho factor was 0.689 l/kg (SD = 0.061). The value of beta increased slightly with increasing dose of alcohol (P < 0.05). The blood alcohol parameters beta and rho were negatively correlated (r = -0.135). The BACs measured at 2 h and 5 h post-drinking were used to estimate the amount of alcohol each subject had consumed according to the method proposed by Widmark [1]. The mean differences (estimated-actual) and the +/- 95% limits of agreement were -0.72 g (+/- 12), and 2.2 (+/- 15), for the 2 h and 5 h BAC values, respectively. A method based on error propagation was used to derive the 95% limits of uncertainty in the amount of alcohol ingested. On the basis of a single measurement of BAC, we could estimate the amount of alcohol ingested within +/- 20%.

Employing simulated data to illustrate an important cause of the 'steepling' effect in breath alcohol analysis.

The 'steepling' effect (large excursions in analytical data over time) is a debated issue in forensic breath alcohol analysis with various explanations being postulated. Simulated breath alcohol data was generated according to a hypothetical kinetic model where single random samples as well as means of duplicate random samples were plotted with respect to time at 0.2 hour intervals. In addition, the simulated data was compared when both two or more digit treatment was employed. Results showed the occurrence of significant noise or 'steepling' when single, two-digit breath alcohol samples were employed as compared to a four-digit mean computed from three-digit duplicates. The magnitude of variability was quantified by means of nonlinear regression resulting in the residual sum of squares (RSS) = 0.00202 for the single analysis and RSS = 0.00053 for the mean of duplicates. The method of data collection and treatment appears to contribute significantly to the 'steepling' phenomenon. Intuitively, replicate analyses reduce variability and allow for more accurate kinetic modelling employing breath alcohol analysis.

Isopropanol interference with breath alcohol analysis: a case report.

The presence of interfering substances, particularly acetone, has historically been a concern in the forensic measurement of ethanol in human breath. Although modern infrared instruments employ methods for distinguishing between ethanol and acetone, false-positive interferant results can arise from instrumental or procedural problems. The case described gives the analytical results of an individual arrested for driving while intoxicated and subsequently providing breath samples in two different BAC Verifier Datamaster infrared breath alcohol instruments. The instruments recorded ethanol results ranging from 0.09 to 0.17 g/210 L with corresponding interferant results of 0.02 to 0.06 g/210 L over approximately three hours. Breath and venous blood specimens collected later were analyzed by gas chromatography and revealed in the blood: isopropanol 0.023 g/100 mL, acetone 0.057 g/100 mL and ethanol 0.076g/100 mL. Qualitative analysis of the breath sample by GCMS also showed the presence of all three compounds. This individual had apparently consumed both ethanol and isopropanol with acetone resulting from the metabolism of isopropanol. An important observation is that the breath test instruments detected the interfering substances on each breath sample and yet they did not show tendencies to report false interferences when compared with statewide interferant data.

The frequency of apparent acetone in a group of breath alcohol data: statistical treatment and forensic implications.

The potential presence of interfering substances (specifically acetone) is a concern in the forensic reporting of evidential breath alcohol analysis. As a result, manufacturers have designed instruments to monitor its occurrence through various hardware and software features. This paper is a retrospective study where 35,945 duplicate breath samples from BAC Verifier DataMaster instruments are evaluated for the frequency of 'interferant' values > or = 0.010 g per 210 l ethanol equivalent. A total of 264 (0.74%) of the duplicate samples had an interferant value on the first sample only, while 235 (0.66%) had interferant values on the second samples only. A total of 77 (0.21%) of the duplicate samples had in interferant values on both breath samples and only in these cases could the presence of measurable acetone even be considered. The occurrence of interferant results appeared also to be instrument-dependent with 55.7% of the interferant values on the first breath sample occurring on nine (13%) of the instruments displaying such results. The occurrences of interferant values on the first breath sample did not conform to the Poisson distribution (P < 0.0001) for the instrument with the largest number of occurrences, while there was conformance for other instruments evaluated. Finally, approximately 23 cases (0.064%) remained where the presence of acetone is a possible consideration. Several issues are presented that the forensic scientist should consider when attempting to explain an apparent interferant result in an individual case. It should be remembered that measurement results need to be interpreted in their context, and data analysis concerning an instrument's performance should be considered.

Evaluating the variability of duplicate breath alcohol analyses as a function of subject age.

The influence of the person's age upon reproducibility in duplicate breath alcohol analyses is investigated. A total of n = 30,324 duplicate results (with both > or = 0.01 g/210L) were selected and divided into eight age groups from 10-19, 20-29, and up through 80+. Two duplicate agreement criteria, +/- 10% of the mean and +/- 0.02 g/210L, were evaluated according to age. A X2 trend analysis was employed and resulted in: +/- 10% of mean criteria, P < 0.001 and +/- 0.02 g/210L criteria, P > 0.05. The proportions of duplicates not conforming to the two agreement standards along with 95% confidence intervals were: +/- 10% of mean, 0.026 (0.024 to 0.028) and +/- 0.02 g/210L, 0.052 (0.050 to 0.054). Depending on the agreement criteria selected there will be proportional differences, but neither appears to be importantly influenced by the subject's age. The breath sampling criteria employed in the instrument studied does not appear to inhibit acceptable agreement, even in elderly subjects where the risk of respiratory disfunction increases.

The application of control charts in breath alcohol measurement systems.

Measurement provides numerical information, usually to assist in some decision process. Quality control is fundamental to the measurement process if the results are to provide confidence to the decision maker. The degree of quality control required depends on the context and purpose of the measurements. Quality control is particularly important in the forensic measurement of breath alcohol in light of the significant consequences involved. Control charts are an important and widely used tool in quality control for both measuring and manufacturing processes. They help evaluate measurement variability and provide a visual assessment of the system's state of statistical control. Control charts can be developed and applied in a variety of different ways. Several examples are illustrated that apply control charts to breath alcohol measuring systems. Application of these methods should result in improved process monitoring in addition to improved confidence for forensic purposes.