The Dubious Practice of Retrograde Extrapolation - DRAFT              6/2/2009

The “Dubious Practice”[1] of Retrograde Extrapolation in DWI cases


Retrograde extrapolation is a fundamentally flawed practice – “the inadequacies of retrograde extrapolation extend beyond mere technical inaccuracies to problems which are inherent in the basic premises and calculations of this technique.”  Smith v. Tuscaloosa, 601 S. 2d 1136, 1140 (Ala. Ct. App. 1992), quoting E. Abbott, ‘One for the Road’ – The Reliability of Retrograde Extrapolation and the Implications for Vermont Statutes, 16 Vt. L. Rev. 395, 397 (1991).

In particular, retrograde extrapolation in DWI cases is based on two false assumptions:

False Assumption 1: All people metabolize alcohol at a consistent rate at all times.[2]

In fact, there is significant variation in the rate of alcohol metabolism between individuals, and equally significant variation in the rate of alcohol metabolism in the same individual at different times.  See A. W. Jones, Biochemical and Physiological Research on the Disposition and Fate of Ethanol in the Body, in Garriott's Medicolegal Aspects of Alcohol 47, 93 (James Garriott, 5th 2008) [hereinafter Jones, Disposition and Fate of Ethanol].

Studies have shown significant variation in alcohol metabolism rate within the same individuals when rates were measured on 10 separate occasions.  For example, alcohol metabolism rates varied between .011 and .015 in one subject, and between .012 and .017 in another.  Id., citing a study by Schonheyder 1942.

Generalizing from an average to your client’s alcohol metabolism rate is doubly dubious: the State's expert can’t tell you whether your client exhibits an “average” alcohol metabolism rate, or whether he is one of the thousands of people who display more unusual rates.  And even if the State's expert could measure your client’s alcohol metabolism rate on one occasion, research shows that he could not reliably predict your client’s alcohol metabolism rate on any other occasion! 

The State’s experts will testify that the “average” rate of alcohol metabolism is approximately .0165 per hour.  See Transcript of Record (Excerpts) at 74, State v. Thrasher, Nos. 01-CRS 877-80 (Currituck County NC Super. Ct. April 7, 2003).  This means that if a person blows a .070 one hour after being pulled over, the State will ask the court to find that your client had a .0865 blood alcohol content (BAC) at the time he was pulled over.

And remember, the higher the alcohol metabolism rate, the higher the judge will find your client’s BAC at the time of driving.  Defense counsel’s job, therefore, is to educate the judge on the broad range of alcohol metabolism rates between and within individuals, and to make sure our clients get the benefit of the doubt when it comes to alcohol metabolism rates. 

If reputable studies have found alcohol metabolism rates in the general population that would exonerate a Defendant, reasonable doubt dictates that the Defendant should get the benefit of that uncertainty.  “In criminal cases it is important to give the accused any benefit of the doubt when scientific and technical evidence is proffered.” Jones, Disposition and Fate of Ethanol, supra, at 87.  In short, we should fight to have the lowest experimentally demonstrated alcohol metabolism rate possible applied to our clients.  This is only fair, since the State will not be able to offer any evidence that our client did not exhibit such an alcohol metabolism rate.

Paul Glover has testified that a .0165 alcohol metabolism rate is “conservative”, because “most” DWI cases involve Defendants with a significant history of alcohol consumption.  See Thrasher, Tr. of Record at 74-5.  (As a general rule, regular heavy drinkers metabolize alcohol faster than non-drinkers.)  But this is hardly a “conservative” assumption, since Mr. Glover likely has no idea whatever of your client’s history with alcohol.  He doesn’t know your client’s actual alcohol metabolism rate, and he doesn’t know if the incident at trial was your client’s very first experience with alcohol. 

  1. Beware of averages!


The State's experts will try to convict your client on the assumption that he exhibited an “average” alcohol metabolism rate on the day of his arrest.

Needless to say, your client is an individual and not an “average” person.  And averages can seriously distort the reality of scientific data: 

  1. If you have $100 and I have no money, on average each of us has $50.  Only one of us can buy dinner. 
  2. The average human being living today is Chinese and female.  Of all the people who have ever lived, the average person is dead.  Is your client Chinese and female?  (Let’s hope no dead clients are appearing in court.) 
  3. The average person driving down the road is not DWI – should that be enough to exonerate your client?
  4. Imagine being forced to wear “average” sized shoes all day – how would your feet feel?  How would “average” sized garments fit you?   
  5. How would you feel if you were forced to attend the most popular (or “average”) church, or subscribe to the most popular or “average” political party?
  6. If you were having an operation and doctors gave you a transfusion of the most common or “average” blood type, would it help you or kill you?

Greek mythology tells the story of Procrustes, an infamous landlord who insisted his guests’ bodies be altered to fit his only bed.  If the guest was too short, Procrustes would break and stretch the guest’s body on the rack.  If the guest was too tall, Procrustes would chop off his feet.  Retrograde extrapolation enacts a similar kind of violence on Defendants’ due process rights. 

Much of the practice of retrograde extrapolation is based on the work of a Swedish psychologist named Widmark in the 1930s.  Interestingly, Widmark’s original study only tested 30 individuals.  Of those 30 subjects, only three actually exhibited “average” alcohol metabolism rates.  Lawrence E. Taylor, Drunk Driving Defense 369 (6th ed. 2005) [hereinafter Taylor, Drunk Driving Defense].  Despite this wide degree of variation among individuals, State experts still try to convince judges and juries that all people eliminate alcohol at the same rate.

Your client is entitled to a tailored, individualized determination of guilt or innocence – that’s due process.  Justice is not a clown costume – it must be made to measure exactly, otherwise reasonable doubt applies.   

For this reason “average” alcohol metabolism rates are irrelevant and immaterial in a trial setting.  It’s irrelevant because the State has no way of proving whether your client’s actual alcohol metabolism rate is below average, average or above average.  Without more evidence, the State can’t show any relevance at all to your client’s particular situation.

  1. Real life variations in alcohol metabolism rates

Alcohol metabolism rates vary dramatically between people, with “the vast majority” people falling in a range between .010 and .025 g/dl alcohol metabolism per hour.  Jones, Disposition and Fate of Ethanol, supra, at 93.  Rates as high as .06 per hour have also been shown.  Thrasher, Tr. of Record at 27.

In addition, research has demonstrated significant variations in alcohol metabolism in the same individuals at different times.  Jones, Disposition and Fate of Ethanol, supra, at 93.  Some of the factors include 1) what kind of alcohol is consumed; 2) how fast the alcohol is consumed; 3) how much the subject had to eat before or during alcohol consumption; 4) what kind of food the subject ate before or during alcohol consumption; 5) the subject’s recent history of alcohol consumption.  Id. 

  1. The neglected value of statistical analysis

Practicing scientists have long understood the potential distorting effect of averages in summarizing experimental results (see examples above).  For this reason, scientists and mathematicians developed methods of statistical analysis to determine the reliability of averages and other values in summarizing results.

One of the core concepts of statistics is the standard deviation, which measures the degree of variation in a population around the mean or average value.  The smaller the standard deviation, the more closely the data points cluster around the mean.  In example one above, the standard deviation would be extremely high, showing that the average value is not a reliable indicator of the distribution of money.  But if you have $55.00 and I have 45.00$ the average would be still be $50.00, while the standard deviation would be much smaller. In that case the average would be a more useful guide to reality.

Statistical analysis allows scientists to confirm that experimental findings are real, and helps protect consumers (like judges) from being misled by experimenter’s wishful thinking or bias.

We need to challenge the State's experts to discuss their statistical analysis, if any, of research results they present in court.  Without such analysis, presentation of averages conceals more than it reveals in terms of our clients’ actual alcohol metabolism rates. 

But remember that discussing statistics does NOT mean that average alcohol metabolism rates are relevant.  We must still demand individualized determination of guilt or innocence, and even robust average values are insufficient to satisfy proof beyond a reasonable doubt.

  1. Publish or perish – the value of peer review

Reputable scientists publish their findings in peer-reviewed journals.  Then other scientists can review their experimental model, methods, data and statistical analysis.  Other scientists can also attempt to replicate the results using the same methods in order to confirm that the authors’ conclusions are correct.

Failure to publish “scientific findings” in a peer-reviewed journal is not proof that those purported findings are false or unreliable, but mainstream scientists are very reluctant to rely on findings that have not been quality-checked by the publication process.

A real scientist is not afraid to have his results checked by real experts.  Less reputable scientists, on the other hand, might try to sell these damaged goods to less knowledgeable consumers like prosecutors, juries and judges.  Would you buy a used car from a dealer who refused to let your mechanic check it out?

False Assumption 2: All people tested will have reached their peak BAC at some point before the time of BAC testing.


The state will usually have only one data point (the Intoximeter blow) purporting to show your client’s BAC, and hence cannot definitively demonstrate whether his BAC was going up or down at the time of measurement.  A primer on alcohol toxicology for prosecutors nicely highlights the potential for confusion: 

Be aware that any given alcohol concentration other than the peak AC will be achieved at least twice during a drinking session, once on the way up and again on the way down.  An alcohol concentration from a single sample cannot by itself be used to predict whether a subject is absorptive, post-absorptive or at a peak AC.  The measured alcohol concentration only provides a snapshot of the AC at the time of collection.

American Prosecutors Research Institute, Alcohol Toxicology for Prosecutors 10 (2003), National District Attorneys Association, at (visited May 25, 2009).

Unless the State can show whether your client’s BAC was going up or going down at the time of testing, the Intoximeter value is useless for retrograde extrapolation.  As a result, the prosecutors’ manual contains a warning that the retrograde extrapolation “provides a good estimate [of BAC at time of driving] only if there was no unabsorbed alcohol from recent drinking at the time of offense, and/or there was no alcohol consumed after arrest and before the test.”  APRI, Toxicology for Prosecutors, supra, at 21.   

Studies show that even “two blood samples taken 30-60 min apart is not sufficient to allow making an unequivocal statement of whether the person was in the post-peak phase at the time of driving.”  Jones, Disposition and Fate of Ethanol, supra, at 95-6.  In order to prove whether your client’s BAC was going up or down at the time of driving, the state would need a series of reliable BAC measurements over time.  APRI, Toxicology for Prosecutors, supra, at 25.   

Remember also that numerical portable breath test (PBT) results are explicitly inadmissible pursuant to statute (N.C.G.S. § 20-16.3) and case law (State v. Bartlett, 130 N.C. App. 79, 82 (1998)).  N.C.G.S. § 20-16.3 refers to PBTs as “alcohol screening tests” because they are designed to screen for the presence or absence of alcohol, rather than to determine a particular BAC.  Certainly such tests are unreliable and inadmissible to prove a particular alcohol metabolism rate in conjunction with an Intoximeter reading.  It may be useful to adopt the term “screening test” when discussing PBTs to remind the court of the devices’ limited purpose. 

The state’s experts often rely on an assumption that essentially all alcohol is absorbed into the bloodstream within 17 minutes after the last drink is taken, and that BAC begins to go down approximately 42 minutes after the last sip of alcohol.  Thrasher, Tr. of Record at 21.  Based on this assumption, the expert will tell the court that your client’s BAC must have been on its way down when he was tested an hour or so later.

In fact, research has shown a very wide variation in time required to reach maximum BAC.  Factors include individual genetics, body position, food eaten before or during alcohol consumption, type of alcohol consumed, how fast the alcohol is consumed, and the individual’s drinking history.  Jones, Disposition and Fate of Ethanol, supra, at 89-103.  See also Taylor, Drunk Driving Defense, supra, at 362. 

The mainstream range of time to full absorption of alcohol is 30 to 90 minutes.  Thrasher, Tr. of Record at 21.  But in extreme cases, it can take three to four hours to reach max BAC, and Mr. Glover has admitted this.  Thrasher, Tr. of Record at 35, 53.  “Extreme cases are still possible because some individuals might have had a delayed gastric emptying and thus not reached peak BAC at the time of sampling.”  Jones, Disposition and Fate of Ethanol, supra, at 88.  This is particularly important in retrograde extrapolation cases, since retrograde extrapolation can grossly overestimate a defendant’s BAC if his BAC was still rising at the time he was tested.  (See graph.) 

The False Foundation of State v. Catoe

State's witnesses routinely testify that the practice of retrograde extrapolation was endorsed by the NC Court of Appeals in State v. Catoe, 78 N.C. App. 167 (1985).  They will even claim that Catoe stipulated a particular alcohol metabolism rate to be used in future retrograde extrapolation cases.  Thrasher, Tr. of Record at 14. 

In fact, Catoe contains so such holding.  To begin with, the Catoe court based its decision on a limited record.  The court notes that the Defendant failed to object to the scientific reliability of retrograde extrapolation at trial, “which hinders our consideration as a full record was not developed.”  Catoe, 78 N.C. App. at 169.

In addition, the Catoe court analyzed the trial court’s decision under an abuse of discretion standard, a standard very deferential to the analysis and judgment of the court below.  Howerton, 358 NC at 458.  Such a holding does not foreclose other trial courts from reaching different conclusions on the relevance and admissibility of retrograde extrapolation evidence.  The Catoe court did not endorse the use of retrograde extrapolation – it merely declined to set aside one particular trial court’s exercise of discretion to admit such evidence.  Catoe leaves ample room to attack retrograde extrapolation in current cases.

Unfortunately, subsequent North Carolina Court of Appeals cases have interpreted Catoe more broadly as an endorsement of the retrograde extrapolation methodology.  State v. Taylor, 165 N.C. App. 750 (2004), contains an expansion of the Catoe holding on retrograde extrapolation.

Admissibility of Retrograde Extrapolation Evidence

Howerton v. Arai Helmet, Ltd., 358 N.C. 440 (2004) articulates the standard for admissibility of scientific evidence.  Howerton sets forth the applicable three-step inquiry from State v. Goode, 341 N.C. 513 (1995) concerning the admissibility of expert testimony: "(1) Is the expert's proffered method of proof sufficiently reliable as an area for expert testimony? (2) Is the witness testifying at trial qualified as an expert in that area of testimony? (3) Is the expert's testimony relevant?" Howerton, 358 N.C. at 458 (internal citations omitted).

State v. Taylor, in turn, quotes Howerton and State v. Davis:

Regarding the first step, "when specific precedent justifies recognition of an established scientific theory or technique advanced by an expert, the trial court should favor its admissibility, provided the other requirements of admissibility are likewise satisfied."  Howerton, 358 N.C. at 459.  Our Court has "accepted the reliability of extrapolation evidence since 1985." State v. Davis, 142 N.C. App. 81, 90, disc. review denied, 353 N.C. 386 (2001).

Taylor, 165 N.C. App. at 753.  The Taylor court goes further to state that “We view Catoe as the type of ‘specific precedent’ indicated in Howerton which is meant to encourage a trial court to favor the admissibility of extrapolation evidence based on an average elimination rate.”  Id. at 754. 

Remember, however, that admission of retrograde extrapolation evidence does not end the case: defense counsel should still argue the unreliability of average values in determining individual guilt.  These arguments go toward the weight of retrograde extrapolation once admitted.

Potential Topics for Cross-examining the State's “expert”


  1. Human beings exhibit a broad range of alcohol metabolism rates

Paul Glover himself has testified that “non-drinkers” who are “dosed at low levels” exhibit an alcohol metabolism of approximately .014 to .015 grams of alcohol per 100 milliliters of blood per hour (short hand g/dl/hr), whereas “drinking drivers” exhibit rates between .018 for women and .020 for men. Thrasher, Tr. of Record at 14.  Mr. Glover likewise testified that there is “reliable” research showing alcohol metabolism rates as low as .010 per hour.  Thrasher, Tr. of Record at 27.  Mr. Glover acknowledges the existence of research showing rates of .009 per hour, but questions the “reliability” of this research.

Mr. Glover testified that he wouldn’t know what alcohol metabolism rate a person pulled off the street would exhibit (within the general range between .010 and .030) without doing an “individualized test” on that person.  Thrasher, Tr. of Record at 29.

Despite these variations, Mr. Glover has testified that .0165 is the rate he and his staff use to perform retrograde extrapolation, and that “it’s a conservative rate and it’s applied uniformly when we apply it to males and females regardless of the situation.”  Thrasher, Tr. of Record at 15.

We can already see the distortion in Mr. Glover’s calculations.  He has admitted that subjects exhibit a range of alcohol metabolism rates, and that men and women exhibit different rates, but that he and his staff apply the .0165 rate “regardless of the situation”.  Thrasher, Tr. of Record at 14

Note: it may be useful to cross the witness on the percentage margin of error inherent in using average values.  .010 is approximately 40% lower than .0165, and .030 is almost 100% higher.  Paul Glover himself has testified that alcohol metabolisms rates as low as .010 and as high as .030 are regularly found in “reliable” scientific studies.  Thrasher, Tr. of Record at 27-8.  Based on this evidence, it appears that retrograde extrapolation in the absence of individual data from the Defendant can overestimate individual alcohol metabolism rates by as much as 40% and underestimate them by 100% or more.

  1. Using Alcohol Toxicology for Prosecutors as a ‘learned treatise’ on cross.

The American Prosecutor’s Research Institute prepared Alcohol Toxicology for Prosecutors as a resource to help prosecutors “use the science of toxicology as a sword” against DWI Defendants.  APRI, Toxicology for Prosecutors, supra, at 1.  A monograph explicitly designed to dismember DWI Defendants makes an excellent resource to use against overreaching retrograde extrapolation “experts”.  Paul Glover himself is thanked for his “invaluable review and assistance with the technical material” included in the book.  Id. at 2. 

Among other things, the monograph contains important graphs on time to peak BAC and the effect of food on alcohol absorption.  See APRI, Toxicology for Prosecutors, supra, at 11, 13.   

In addition, it also contains crucial warnings about the limitations of the retrograde extrapolation method:

An expert’s estimation and testimony at trial will only be as good as the information provided. When there is a delay between the time of the crash and the test, a thorough police investigation is paramount.  The most important parts of the drinking history for purposes of retrograde extrapolation are alcohol consumption in the hour prior to the offense and any consumption after the offense but before the test.

Id. at 20.  In most DWI cases, the State will be missing exactly these important pieces of information. 

The book also contains the “first person prosecutor” testimonial by ADA Jane Starnes, in which she warns that “if your expert insists that there is just one standard elimination rate, be scared.”  Id. at 31.  You can argue to the judge that we should all “be scared” because the State is trying to fit every Defendant into the same suit of clothes.  Remind the judge that justice should fit better than a clown costume.  Use the story of Procrustes and his one size fits no one bed.

  1. Questioning the validity of the State’s controlled drinking experiments.

While the State's expert Bryan Smith will testify that he’s done hundreds of controlled drinking “experiments”, he will usually not be able to testify as to the exact number of subjects, their gender, ethnicity, age, genetic background, etc.  Nor will he be able to describe the results of any statistical analysis.  (See above for discussions of statistical analysis). 

It is important also to distinguish between controlled laboratory experiments conducted for research purposes and the kind of “educational events with respect to alcohol” that Mr. Smith and Mr. Glover put on a conferences and meetings.  Thrasher, Tr. of Record at 29.  Messrs. Smith and Glover may put on controlled drinking demonstrations, but those presentations do not have the same scientific rigor or reliability as controlled laboratory studies published in peer-reviewed journals. 

Statistical analysis is crucial to determine the degree of variation (how good an average is at describing the degree of variation around a mean) and to isolate potential confounding variables from consideration in making scientific conclusions. 

It is also unlikely that Mr. Smith has published any of his results.  If not, he has never subjected his methods, analysis and conclusions to the quality control process of peer review.

  1. Retrograde Extrapolation As Applied By Paul Glover Is An Unscientific Guessing Game.

When push comes to shove, Paul Glover is forced to admit that what he does in DWI cases is not based on established scientific research or procedure:

Q: I’m asking for one example where the scientist, the researchers knew nothing about the individual’s alcohol consumption, okay?  That is point one, they knew nothing about the consumption, time, amount, quantity, food in the stomach, they knew nothing.  They did only one sample from the subject’s blood.  They didn’t have two samples so they could individualize the elimination rate, they had only one.  Can you cite for us an example of a study wherein those researchers said, without knowing anything about this individual’s drinking and without knowing this individual’s established elimination rate, we can reliably predict and extrapolate what his BAC would have been at a previous point in time?

Paul Glover: No.

Thrasher, Tr. of Record at 44.

  1. Challenging Bryan Smith as an Expert

Few attorneys dispute that Paul Glover is “a witness qualified as an expert by knowledge, skill, experience, training or education” to testify on retrograde extrapolation under Rule of Evidence 702, and to render opinions to “assist the trier of fact to understand the evidence or to determine a fact at issue.”  Mr. Glover has B.S. and M.S. degrees in Biology, and significant experience in applying retrograde extrapolation to DWI cases.  Thrasher, Tr. of Record at 68.

Bryan Smith, on the other hand, has significantly less training and experience.  Based on past testimony, Mr. Smith has an associate’s degree in Criminal Justice, and as of last year his academic training in alcohol metabolism, retrograde extrapolation and related topics was limited to a ten hour course.

Mr. Smith makes a great deal of the controlled drinking “experiments” he has conducted.  He claims to have measured the alcohol metabolism rates of approximately 1000 individuals by administering a bolus dose of alcohol and then measuring the decrease in BAC over time.  Mr. Smith claims these sessions produced data that match the average alcohol metabolism rate he and Paul Glover use when testifying in retrograde extrapolation cases.

Mr. Smith’s data have several flaws:

  1. He has not previously been able to state the exact number of subjects he has tested
  2. He cannot testify to any demographic information about the test subjects such as age, race, gender, prior drinking habits, etc.
  3. He is unsure what or whether his subjects ate before, during and after testing.
  4. He has performed no statistical analysis on his data beyond a primitive averaging of alcohol metabolism rates gathered.

In reputable, peer-reviewed scientific journals, much of the body of an article is composed of statistical analysis.  Without such statistical analysis scientists would not make claims to significant or reliable results.  Mr. Smith has neglected to perform this essential analysis, and as a result his data are scientifically useless.  His data are merely “anecdotal”, a term used to describe scientific conclusions reached without proper vetting and analysis of experimental results.

While Mr. Smith has given valuable service to the State as a law enforcement officer, he lacks the education, training, experience, knowledge or skill to qualify him as an expert on retrograde extrapolation or any other pharmacokinetic analytical process.

A Seriously Dubious Practice

A.W. Jones is a well-known and influential expert on alcohol metabolism rates and the pharmacokinetics of ethanol.  Paul Glover often cites and relies on Dr. Jones’ work.  Dr. Jones has this to say about retrograde extrapolation:

Requests to back extrapolate a suspect’s BAC from the time of the sampling to the time of driving are often made, but this remains a dubious practice, owing to the many variables and unknowns involved.  The pharmacokinetic profile of ethanol is highly variable depending on the pattern of drinking and physiological characteristics of the subjects.  In a typical impaired driving trial, only a single measurement of BAC or BrAC is usually available making it very difficult to engage in retrograde calculations with sufficient certainty for a criminal prosecution. 

Jones, Disposition and Fate of Ethanol, supra, at 127.  Due process demands that in the absence of any proof of our clients’ actual alcohol metabolism rate on the day in question, the court should apply the lowest experimentally verified alcohol metabolism rate on record to our client.   

              10 of 11

[1] “Requests to back extrapolate a suspect’s BAC from the time of the sampling to the time of driving are often made, but this remains a dubious practice, owing to the many variables and unknowns involved.  The pharmacokinetic profile of ethanol is highly variable depending on the pattern of drinking and physiological characteristics of the subjects.”  A. W. Jones, Biochemical and Physiological Research on the Disposition and Fate of Ethanol in the Body, in Garriott's Medicolegal Aspects of Alcohol 47, 127 (James Garriott, 5th 2008) [hereinafter Jones, Disposition and Fate of Ethanol]. 

[2] See, e.g., Transcript of Record (Excerpts) at 59, State v. Thrasher, Nos. 01-CRS 877-80 (Currituck County NC Sup. Ct. April 7, 2003), where an ADA describes retrograde extrapolation this way: “You take a scientifically proven piece of information that human beings have a constant consistent elimination rate and then you apply it in the field.”