LANSING — A calibration error in 4,001 alcohol cases resulted in inaccurate records in some of them, drawing criticism from defense attorneys and forcing prosecutors to take a second look. 

An examination of the cases, where testing was done at Michigan State Police labs, showed some results were incorrect by a range of -0.002 grams per deciliter (g/dL) to +0.004 g/dL, the June 30 letter signed by acting commander of the MSP Forensic Science Division, L. Scott Marier, states. 

Upon finding the error in April 2016, MSP took immediate corrective action to determine the most appropriate measures to remediate the issue and ensure the incorrect calibration model could not be used again, Michigan State Police Spokeswoman Shanon Banner said. 

“The integrity of our laboratory system is of the utmost importance…” Banner said, noting the corrective action. “We are confident in the amended results. The error that occurred was identified and corrected, and appropriate communication was made with our impacted stakeholders.

“Additional safeguards have been put in place to ensure this error does not happen again,” she said. 

Banner noted the error did not produce large differences in result figures, but said it was important to issue amended reports on any case with a change in result.

Among the cases, 2,007 require corrections, while 1,994 do not.

None of the cases originally reported near Michigan’s legal blood alcohol limit of 0.080 g/dL have amended results that move them across the legal threshold in either direction; and none of the cases had to be reanalyzed because existing raw data was accurate and available.

Of the cases originally reported near Michigan’s “Super drunk” legal alcohol limit of 0.170 g/dL, 18 have amended results that move them across the threshold.

“These cases are potentially impacted, but the original error was in favor of the defendant,” Banner said. “The individual would have still been over the drunk driving threshold of 0.080, but the original results report would not have put them at the super drunk threshold.”

Inaccurate readings were due to an incorrect calibration model. A review indicated that the processing method being utilized on that instrument had been incorrect since Dec. 14, 2015.

The incorrect calibration model was used on one of two machines, Banner said, used in blood alcohol cases. In these cases, from Dec. 14, 2015, to April 13, 2016, the labs used two machines and used the average of both to tally the results.