I would like to thank Dr. Mark Ellis and colleagues for their research identifying the important issues of search and identification of missing and unknown children. As the inventor of the Toothprints (Kerr, Orange, Calif.) bite impression technique, I do, however, have serious concerns regarding their most recent publication in September JADA, "An Evaluation of DNA Yield, DNA Quality and Bite Registration From a Dental Impression Wafer" (Ellis MA, Song F, Parks ET, Eckert G, Dean JA, Windsor LJ. JADA 2007;138[9]: 1234–1240).
These concerns specifically arise from the reporting of data collected in 2005 by the authors under Indiana University Institutional Review Board approval. These data were reported previously in two scientific venues: first, as a published abstract in Pediatric Dentistry1 and second, as an OMNII Oral Pharmaceuticals Pediatric Dentistry Postdoctoral Fellowship research presentation by Dr. Ellis at the annual meeting of the American Academy of Pediatric Dentistry (AAPD) in Orlando in 2005.2
No disclosure in the JADA article was made as to the previously reported conclusion that "Toothprints wafer impression material ... was able to produce an accurate representation of the dentition." In that the methods, analysis and conclusions differ in the published abstract, the presentation and the JADA article, it would necessarily precipitate a number of important questions addressed here for the authors’ reply.
First, why were the data and previous reports not cited in the literature review of the JADA article? Which method was used to collect the data on the children? I would ask the authors to define the actual sequence of data collection (rinsing prior to taking the bite impression would certainly be expected to reduce the quantity of DNA available for capture on the wafer).
Second, and perhaps most importantly, during the AAPD presentation, Dr. Ellis described the problems encountered with the data collection. This included "child didn’t really want to stay closed" and "warping" of the wafer. This also would help to explain the poor quality of the bite impression shown in Figure A of the JADA article (are there seven teeth in the bite impression and 10 teeth on the stone model?). Did the technique used for the bite impression produce the quality and clarity described in the published standard3 (this thermoplastic material itself is actually more accurate than alginate [Kerr Dental, unpublished data, 2004])? Because the data collection problems identified by Dr. Ellis in the presentation were not reported in the JADA article, could the distortions caused by a faulty technique have contributed to any of the mismatched comparisons?
Third, was the comparison technique "modified" after 2005? There was no indication of how a determination for match and unmatch was calculated. Was the examiner calibrated? What variance in overlap was recorded as a mismatch? To what sensitivity (micrometers, millimeters) were the comparisons made? Bite mark analysis is accepted to be a "physical comparison science with most identifications based on uniqueness" (sometimes only one or two teeth). The method described in the study, requiring matching all teeth at the 95 percent confidence level, might even bind future forensic testimony to an excessive standard. Perhaps the analysis presented by Dailey and McGivney4 at the American Academy for Forensic Sciences meeting in 2005 is a more useful tool? They also affirmed that "quality control is an absolute requirement."
Finally, at the American Board of Forensic Odontology 1999 Bitemark Workshop,5 an accuracy score of 0.86 was concluded to correlate with bite mark certainty and forensic value. It begs the question: are the authors’ conclusions even consistent with their results?
TOP
REFERENCES
