Dear Michigan Litigator:

Michigan statutes (MCL 600.2955) and rules of evidence (MRE 702 and 703) require expert reporting to include established, peer reviewed methods and Error Rates to demonstrate relevant and reliable evidence. Current and past Michigan State Police Forensic Lab gas chromatography – headspace BAC method and reporting practices (MSP, 2006) violate these requirements by implementing contraindicated techniques and biased reporting. Specifically misleading BAC analytical interpolation and unethical MSP practices are delineated below. These facts may be applicable to your current and/or past cases. Recently the U.S. Senate and House of Representatives initiated via the National Academy of Sciences a critical review of such forensic toxicology practices (http://www.nap.edu/catalog/12589.html). KETox presents these offensive MSP GC-HS BAC issues to address your current and past client’s forensic issues.

The MSP contraindicated and biased practices include:
I. MSP’s GC-HS BAC (2006) pretreatment step (2.1.5.1) involves the butyl rubber septa (BRS) contact with blood + internal standard:
A. BRS use for measuring BAC is explicitly contradicted within MSP’s cited reference (Seto, 1994).
B. BRS preferentially adsorbs IS, i.e., 1-propanol (GC5) or t-butanol (GC6)) causing serious analytical errors and mistaken BAC interpolations.
C. Fig. 1 graphically illustrates how the use of BRS can incorrectly interpolate (i.e., horizontal intercept with vertical arrows, point B) your client’s actual BAC (point A).
1. BRS preferentially adsorbs IS to reduce its detected signal that then elevates the [Esignal/ISsignal] ratio and mistakenly interpolates your client’s BAC.

II. MSP GC-HS fails to provide characterized Error Rates for single GC-HS Aqueous Ethanol Standards, NIST AES and Blood Controls for unbiased, random errors required for statistical control in BAC expert report per Daubert (Langhofer, 2004).
A. MSP (2006) Inferred vs. Observed GC-HS BAC Analysis :
1. GC-HS weekly calibration purposefully or unwittingly applies AES + IS and Linear Regression Analysis, (Fig. 2) [E]/[IS] v [E gm/dL], that deceptively looks better than the observed data.
a. MSP’s LRA graphically designates individual AES points derived from the resulting equation (Fig. 2) and fail to provide the observed GC-HS [E]/[IS] points and their variances that would describe individual BAC’s accuracy with Upper/Lower Error Limits (blue dashed lines). The depicted 8 exact line points is as impossible as 8 flips of a coin yielding 8 consecutive head observations!
b. MSP’s designated AESs optimize accuracy at 0.192 gm/dL (Draper, 1998) and become progressively larger above and below this point. Why isn’t maximum accuracy at 0.080 gm/dL?
c. The Lower and Upper Error Limits assure the accuracy of each measured BAC based on the methods characterization calibration Error Rate for each AES. In lieu of MSP’s AES ER, Fig 2 graphically illustrates the Upper (0.39 gm/dL) and Lower (0.25 gm/dL) accuracy range for an interpolated BAC of 0.32 gm/dL. Without GC-HS ER it is impossible to evaluate your client’s GC-HS BAC. Similarly for 0.080, the LEL is not intersected, cannot be demonstrated, and UEL exceeds 0.110!
d. MSP’s reported R2 (coefficient of determination ) always approximates ~ 1.0 and an apparent intercept at origin, i.e., – 0.0017. R2 = 0.9998 suggests a perfect correlation that is influenced by the designated AES.
1) MSP’s LRA excludes the highly variable 0.000 and 0.012, and includes low variance AESs of comatose, 0.393, and lethal, 0.59 gm/dL, BACs. MSP’s designated AES bias R2 ~ 1, inherently minimizes accuracy in critical sobriety testing ranges, i.e., 0.020, 0.040 and 0.080 gm/dL, and misleads its judicial BAC reporting.

2. MSP Daily GC-HS calibration checks (2.1.5, Method) for AES [E]/[IS] and interpolated NIST AES, and blood controls depend on their Acceptance Criteria established by GC-HS method calibration and characterization ER (item II.A.1). This warrants GC-HS interpolation of your client’s BAC but these critical Acceptance Criteria are not established nor provided per FOIAs.

The data presented above were derived from available data in FOIAs from MSP and the documented MSP GC-HS Method (MSP, 2006). The critique is based on cited reference within MSP (2006), industry accepted practices (Taylor, 1987; Jones and Liddicoat, 2009), and the intentions outlined by the National Academy of Sciences to improve forensic science. Comments are directed at the underpinnings of the GC-HS analytical process, and for the benefit of the MSP GC-HS method in reporting relevant and reliable BACs for critical judicial decision making processes.
Works Cited
Draper, N. S. (1998). Applied Regression Analysis. New York: Wiley.
Jones, G.R., Liddicoat, L. (2009). Quality Assurance, in Garriott’s Medicolegal Aspects of Alcohol, 5th ed. Garriott J.C. (ed). pp 269-274. Lawyers & Judges Publishing, Tuscon.
Langhofer, R. (2004). Michigan adopts Daubert Principles and evidence-based expert testimony, Revised MRE 702 and 703. Michigan Bar Journal , 34 – 37.
MSP. (2006). Michigan State Police Crime Lab, 2.0 Alcohol Protocol.
Seto, Y. (1994). Determination of volatile substances in biological samples by headspace gas chromatography. J. Chromatogr. A , 674, 25-62.
Taylor, J. (1987). Control Charts. In J. Taylor, Quality Assurance of Chemical Measurements. Chelsea, MI: Lewis Publisher.