EEOC Stays Mum on GINA

In “an informal discussion” that it hopes “is helpful,” the Equal Employment Opportunity Commission (EEOC) has given an apparently negative answer to the following question:

Are DNA sequences that cannot be used and are not acquired by an employer to obtain medically relevant information on an employee “genetic information” as the term is used in the Genetic Information and Nondiscrimination Act of 2008 (GINA)?

The question came from the Scientific Working Group on DNA Analysis Methods (SWGDAM), which requested an “Informal Advisory Opinion ... on the GINA exemption for forensic laboratories.” GINA makes it generally illegal for “an employer to request, require, or purchase genetic information with respect to an individual.” § 202. It defines “genetic information” as “information about [an] individual’s genetic tests.” § 206(A)(i). Finally, it defines a “genetic test” as “an analysis of human DNA, RNA, chromosomes, proteins, or metabolites, that detects genotypes, mutations, or chromosomal changes.” § 201(7)(A).

Congress, in its finite wisdom, devoted almost no attention to the problem this law could create for police departments, laboratories that perform DNA testing, and manufacturers of reagents and materials used for DNA testing. Because personnel at these organizations can—and have—contaminated samples or equipment with their own DNA, it would be helpful to have DNA identification profiles from all employees who might possibly be a source of foreign DNA on record. A check of the records could the confirm or exclude any suspicions of such employee contamination. And even without a database, in a specific case, an employer might have good reason to ask a particular employee for a DNA sample to check whether that individual’s DNA seems like it could have affected a DNA test.

What Is a "Genetic Test"?

This situation could be resolved most simply by interpreting “genetic test” in light of the express Congressional goal of preventing employers from using health-related DNA sequence information to disadvantage applicants for jobs or employees. Because the DNA identification profile does not reveal such information and is not being used for such a purpose, this biometric identifier should not be considered a genetic test under Section 201(7). It is, of course, "an analysis of human DNA," but does it "detect[] genotypes, mutations, or chromosomal changes"? Current STR testing does not detect "genotypes" in the sense of alleles of genes. It does detect "mutations" inasmuch as all genomic variation comes from changes in base-pair sequences, but that cannot be what Congress meant by "mutations." Such a reading would render the preceding word, "genotypes," superfluous. Perhaps "mutations" is in the list to cover sequences that do not cause disease but have predictive power as a result of linkage disequilibrium, or perhaps it refers to somatic mutations that can produce diseases without altering intergenerationally inherited genotypes. Finally, "chromosomal changes" probably refers to breaks in chromosomes or changes in their number (aneuploidy), since these produce medical conditions in children.

Rather than focus on what Congress intended to accomplish with the words in the bills it passed, however, the EEOC passed the buck to a scientific body with no known legal expertise. The EEOC letter reads as follows: “we coordinated closely with experts at the National Human Genome Research Institute (NHGRI) ... who confirmed that forensic DNA analysis constitutes a genetic test.” The unnamed experts at NHGRI are no doubt whizzes at understanding how genetic and genomic research is conducted, but if the EEOC wanted to know whether scientists call forensic DNA profiling "genetic testing," all it had to do was glance at NHGRI’s website. A webpage states that “The term ‘genetic testing’ covers an array of techniques including analysis of human DNA, RNA or protein. Genetic tests are used as a health care tool to detect gene variants associated with a specific disease or condition, as well as for non-clinical uses such as paternity testing and forensics.”

But what should we make of the tautology that forensic DNA testing of human DNA tests human DNA? This issue is not how scientists use the words "genetic testing," but what usage best fits the structure and function of GINA. The fact that scientists define “genetic testing” one way for the purpose of medical research or diagnosis does not mean that the members of Congress used the words the same way for the purpose of combating a perceived problem in the workplace. Cf. Nix v. Hedden, 149 U.S. 304 (1893) (refusing to follow botanical definitions in interpreting an Act of Congress establishing different tariffs on fruits and vegetables).

What Is the CODIS Laboratory Exception and How Did It Get There?  

So what reason is there to believe that Congress intended to use "genetic tests" to refer not just to the tests "used as a health care tool," but also to those used for nonclinical purposes? Beyond invoking the impressive five-letter acronym NHGRI, the “informal discussion” adds that “the fact that Congress included an exception specifically permitting forensic DNA analysis suggests that it constitutes genetic testing that would be prohibited in the absence of the exception.” However, other explanations for this text are possible. As I suggested in an essay entitled GINA's Genotypes, 108 Mich. L. Rev. First Impressions 51 (2010), “most likely, individual legislators may not have had any view about the distinction between genomic and genetic information, but perceived no reason not to include the quality-control exception of section 202(b)(6) for laboratories conducting law enforcement and human remains testing.” If that is correct, then it would be questionable to read the exception as circumscribing the unresolved definition of genetic testing--especially considering the haphazard manner in which the exception came into existence.

Section 202(b)(6) of GINA, 42 U.S.C. 2000ff-1(b)(6), allows an employer to acquire genetic information

where the employer conducts DNA analysis for law enforcement purposes as a forensic laboratory or for purposes of human remains identification, and requests or requires genetic information of such employer's employees, but only to the extent that such genetic information is used for analysis of DNA identification markers for quality control to detect sample contamination.

On April 25, however, the House suspended its rules and passed a bill that differed from the one reported out of committee. The modified bill added (among other things) an exception for any kind of genetic analysis

(6) where the employer conducts DNA analysis for law enforcement purposes as a forensic laboratory, includes such analysis in the Combined DNA Index System pursuant to section 210304 of the Violent Crime Control and Law Enforcement Act of 1994 (42 U.S.C. 14132), and requests or requires genetic information of such employer's employees, but only to the extent that such genetic information is used for analysis of DNA identification markers for quality control to detect sample contamination.

The belatedly modified bill, approved by the House without committee consideration or floor debate on the amendment, went to the Senate. The Senate adopted the bill (with a few more amendments). This Senate version made no change to subsection 6.

The House then took up the Senate version of the original House bill. No committee of either house of Congress had ever discussed subsection 6, and there was not a word from any member about it. But somebody noticed one problem with it, for an eleventh hour change was made. House Concurrent Resolution 340 introduced the "technical amendment" of expanding the exception beyond laboratories that submit profiles for inclusion in CODIS databases. Representative George Miller explained:

[W]ith respect to the Department of Defense Labs, in our current bill, section 202(b)(6) and section 205(b)(6) of H.R. 493 provides an exclusion for an employer to conduct DNA analysis for law enforcement purposes as a forensic laboratory, which submits analyses to the Combined DNA Index System, known as CODIS, if the employer only uses that analysis of DNA identification markers for quality control to detect sample contamination.

However, we recently learned that the Armed Forces DNA Identification Laboratory, AFDIL, of the Armed Forces Medical Examiner System, which identifies soldiers' remains, would not be included in this exclusion because it does not submit DNA to the CODIS system.

It was not our intent to prevent the Armed Forces, AFDIL, from using DNA analysis for human remains identification. This technical change would allow them to continue their mission.

Cong. Rec., May 1, 2008, at H2982, 110th Congress.

On May 1, 2008, the House adopted the amended bill. The remarks about allowing AFDIL to continue analyzing DNA to identify the remains of soldiers, do support the EEOC’s view that Congress thought it needed to have Section 202(b)(6) because the definition of “genetic test” in Section 201(7) was so broad as to reach forensic testing of medically uninformative loci. But if the only purpose of the "technical" change to Section 202(b)(6) was to allow AFDIL to stay in business, then the 202(b)(6) exception applies only to AFDIL and the previously defined group of forensic laboratories that submit samples to the CODIS system. Other forensic laboratories have no safe harbor.

Years ago, I wrote the staff of Representative Slaughter, who fought for 14 years for GINA’s passage, for information on the process that led to the belated insertion of an exception into the bill after the House already had approved of it. I also wrote Representative Miller’s office (twice) but received no reply. Representative Slaughter’s staff provided no real answer either but referred me to Kathy Hudson, then the director of Johns Hopkins University’s Center for Genetics and Public Policy (and now NIH’s Deputy Director for Science, Outreach, and Policy). Dr. Hudson indicated that the provision was added because the Hopkins Center learned that forensic laboratories routinely collect employee DNA for “quality assurance.” She did not explain why this rationale does not also apply to employers such as police departments, whose officers come into contact with biological traces at crime scenes, and manufacturers of supplies used in forensic DNA profiling, whose employees’ DNA might lead to unwitting contamination in the laboratory.

In short, despite 14 years of introducing federal bills to stop employers from using “genetic information” to the detriment of employees and an explosion of state law on “genetic discrimination,” it appears that no one inside (or contributing to) the federal legislative process had a clear idea of the scope of the law-enforcement issues that GINA might create.

Where Does the EEOC Stand? 

Rather than attempt to fill this gap with meaningful, if informal guidance, the EEOC threw up its hands, writing that

Interpreting GINA to allow manufacturers of supplies used in DNA forensic analysis to acquire genetic information under the DNA testing exception might well constitute a reasonable interpretation of the statutory exception, but the issue you have identified was not raised during the public comment period on the regulation implementing Title II of GINA. Nor is there any legislative history to indicate that it was contemplated during the drafting of the statute. The Commission has not taken a position on this issue.

Overall, the letter seems to state that the Commission has adopted the broad and mechanical definition of "genetic test" by reason of its consultation with the scientists or administrators at NHGRI but that it remains willing to contemplate the possibility that the exception for some forensic laboratories and AFDIL is broader than its words suggest.

The letter concludes that “this ... discussion ... does not constitute an official opinion of the EEOC.” What. then, is it? An unofficial opinion? An official nonopinion? A nonofficial nonopinion? It hardly matters. The EEOC's analysis of the question posed is of little use to anyone who thinks seriously about GINA. If and when an employee--or the EEOC--sues a police department or a manufacturer for a practice that Congress clearly had no interest in preventing, one can only hope that the courts will supply a more thoughtful and complete analysis.

Florida Judge Disqualified, But Not for His Writings on Fingerprints

"Florida Trial Court Excludes the Opinion of a Latent Fingerprint Examiner — Maybe" (Oct. 31, 2012) described a vague "order" of Miami-Dade Circuit Court Judge Milton Hirsch. This order, embellished with some poetry, stated that when the time comes in a burglary case, the judge would exercise his “common sense” to fulfill his “gatekeeping function” for scientific evidence under Florida law to prevent “excessive and unsupportable claims made by fingerprint examiners.” Now that order--whatever it may have meant--has been vacated.

On January 16, 2013, the Florida District Court of Appeal ruled that Judge Hirsch should have disqualified himself rather than issue his earlier order. The court of appeal emphasized that it was not disqualifying the judge because he had written on the subject of scientific evidence and fingerprints. The problem was that in two other cases Judge Hirsch had suggested to prosecutors that that he would grant a motion to disqualify him "because of his preconceived opinions on the subject of fingerprint evidence."

These statements, the court of appeal concluded, "certainly would cause a reasonable person to question the judge’s ability to fairly and impartially adjudicate the issues surrounding the reliability and admissibility of fingerprint evidence in a judicial proceeding." This is not the first case in which remarks from Judge Hirsch led to his removal from a case he was trying.

Thanks to former Arizona Superior Court Judge Ronald Reinstein for directing me to this latest opinion in State v. Borrego.

"Scientists' Brief" on CODIS Loci: Q & A

On November 9, 2012, the Supreme Court voted to review a case posing the following question: “Does the Fourth Amendment allow the States to collect and analyze DNA from people arrested and charged with serious crimes?” In Maryland v. King, the state’s supreme court concluded that the protection against unreasonable searches and seizures forbids the state from collecting DNA from an individual whose true identity can be established with ordinary fingerprints. On December 28, 2012, the Supreme Court received a Brief of Genetics, Genomics and Forensic Science Researchers as Amici Curiae. Below are several questions and answers about the brief.

Who contributed to the brief?

I did, and Hank Greely was an additional author. The scientists who participated in the writing are all active and distinguished researchers at medical schools (including Harvard, Yale, and Johns Hopkins) or universities (including Duke, Penn State, and Kings College, London). They include a former president of the American Society of Human Genetics, a past president of the American Board of Medical Genetics, Fellows of the American Association for the Advancement of Science, and members of the Institute of Medicine and the American Academy of Arts and Sciences.

Why did these law professors, medical and statistical geneticists, and molecular biologists submit an amicus brief?

The brief is intended “to inform the Court of the possible medical and social significance of the DNA data stored in law enforcement databases.” (P. 1). Advocacy groups, legal scholars, and some judges have asserted that the small number of features used in law enforcement DNA databases are predictive of health status (or soon will be). The brief attempts to clarify this issue.

Which side does the brief support?

The brief was submitted in support of neither side. It describes the nature of genetic information, the features of the genome used in law enforcement DNA databases, how those features are used in medical research, and whether they currently permit police, employers, or insurers to discern significant facts about a person’s present or future health status.

What conclusions does it reach?

Amici conclude that “[u]nlike medical genetic tests, law enforcement identification profiles have no known value for medical diagnosis or prediction of future health.” (P. 2).

That’s today. What about the future?

Amici caution that “no one can say with certainty what the future will bring, and it is possible that specific loci will be found to affect the operation of certain genes or to display correlations to disease states.” (P. 2). Nevertheless, they suggest that “it is unlikely that the identification profiles will turn into powerful medical diagnostic or predictive tools that can be used to infer disease states or predispositions by examining forensic database records.” (P.2).

Does this mean that the “CODIS loci,” as the identifying features are called, have no medical significance?

Absolutely not. The DNA sequences have been used in medical research for some 20 years to hunt for disease-causing gene mutations. They have been studied for associations with diseases and traits such as longevity. The question the brief addresses is what kind of information can be gleaned from inspecting a database record.

Doesn’t the highly publicized ENCODE Project prove that there is no such thing as “junk DNA”?

The brief contends that “debate over the fraction of the genome that is, in an evolutionary sense, 'junk' ... is orthogonal to the matter before the Court.” (P. 26). A section of the brief explains that the data sets and papers recently released from the international Encyclopedia of DNA Elements Project are important to further research into gene regulation and other matters, but they do not indicate that all DNA sequences are critical to health or other important traits. What “[t]he ENCODE papers show [is] that 80% of the genome displays signs of certain types of biochemical activity—even though the activity may be insignificant, pointless, or unnecessary.” (P. 32).

Well, how about other uses? Don’t the CODIS loci tell scientists a lot about a person’s ancestry and race?

Not really. The CODIS loci can reveal something about bio-geographic ancestry, but anthropologists and population geneticists use far more probative ancestry-informative and lineage markers to study genetic histories. That “race” is not a biological category is now well known. As for socially perceived race, “[a] CODIS profile could be used to calculate probabilities that someone would be described as Caucasian, African-American, or Hispanic, but categorical inferences would not be very accurate, and attempts to predict the census-type race of a person from a CODIS profile would seem pointless considering that apparent race already would be known.” (P. 36).

So the brief shows that there is absolutely no important information that can be deduced from a CODIS profile?

No, amici do not say that either. The brief explains that “[b]ecause children inherit all their DNA from their biological parents, the CODIS loci can be powerful tools for determining whether two people could be genetically related as parent and child. ... [T]he most powerful genetic information other than identity that the CODIS profiles contain [would be] that two people are not parent and child” or “that two people were identical twins.” (Pp. 33-34).

Where can I find the brief?

Here is a pdf. It also should appear."soon," along with other briefs, on the American Bar Association's Preview of Supreme Court cases.

Postscript: The brief and an introduction to it is published as Henry T. Greely & David H. Kaye, A Brief of Genetics, Genomics and Forensic Science Researchers in Maryland v. King, 53 Jurimetrics J. 43 (2013). The publication is available at http://ssrn.com/abstract=2403063

The Judicial Reception of Acquiring Biometric Data on Arrest: Photographing, Sizing, and Fingerprinting Before 1933

A recent article suggests that in the first third of the Twentieth Century, American courts generally did not sanction photographing, fingerprinting, or making physical measurements of arrestees (Bertillonage). According to Wayne A. Logan, Policing Identity, 92 B.U. L. Rev. 1561 (2012), courts did not “sanction the common practice of ‘mugging’ every suspect whose picture and measurements the police would like to have. Nor d[id the courts] sustain the right to retain the prints and measurements after acquittal.” Id. at 1579 (quoting A.M. Kidd, The Right to Take Fingerprints, Measurements and Photographs, 8 Cal. L. Rev. 25, 32 (1919)). A mere “handful of decisions [adopted] a more generous stance.” Id. The early regime of general disapproval of police acquisition of biometric data then changed “[s]tarting in the 1930s, [as] courts began evincing a less critical and more accepting view. Most notably, in United States v. Kelly [55 F.2d 67 (2d Cir. 1932)] the Second Circuit rejected a challenge to the use of identity evidence, based on the absence of statutory authority to extract prints, brought by a defendant facing misdemeanor prosecution under the National Prohibition Act.” Logan, supra, at 1580.

The theory that United States v. Kelly, 55 F.2d 67 (2d Cir. 1932), represents a phase shift is open to serious question. True, a mere handful of decisions upheld the collection, retention, and widespread circulation of biometric records all at once. But then again there were only a handful of reported opinions—of any kind on each of these matters. An annotation from 1933 observed that

Considering the large number of arrests that are made, there are comparatively few cases in which the right of a person to prevent his description from taking its place in the "rogues' gallery" has been involved. The decisions of those few cases are not uniform. This is accounted for in part by the difficulty the courts find, on account of the form of the pleadings, in granting the relief sought. Both mandamus and injunction have been held not to be the proper remedies.

Annotation, Right to Take Finger Prints and Photographs of Accused Before Trial, or to Retain Same in Police Record After Acquittal or Discharge of Accused, 83 A.L.R. 127 (1933).

Having now read almost all the appellate opinions in the original annotation, my conclusion is that the clear majority rule favored the collection and retention of the data. Both before and after Kelly, appellate courts did not recognize high constitutional barriers to collecting biometric data from arrestees and rarely found merit in demands for the destruction of fingerprints, photographs, or bodily measurements.

Proceeding chronologically, in the earliest case collected in the Annotation, the Indiana Supreme Court evinced no doubt that a sheriff had the power to acquire biometric data on an arrestee. In State ex rel. Bruns v. Clausmeier, 57 N.E. 541 (Ind. 1900), the gravamen of the complaint was libel—that the sheriff damaged an acquitted defendant’s reputation by circulating his picture. The Indiana Supreme Court observed (in dictum) that “It would seem, therefore, if, in the discretion of the sheriff, he should deem it necessary to the safe-keeping of a prisoner and to prevent his escape, or to enable him the more readily to retake the prisoner if he should escape, to take his photograph, and a measurement of his height, and ascertain his weight, name, residence, place of birth, occupation, and the color of his eyes, hair, and beard, as was done in this case, he could lawfully do so.” The court cited no opinions to the contrary.

In Shaffer v. United States, 24 App. D.C. 417 (D.C. 1904), the defendant objected to the use of his arrest photograph (in which he had no beard) to help a witness identify him at trial (when he had a beard) on the ground that the state had no right to photograph him for that purpose and that using the photograph at trial violated his right not to be compelled to incriminate himself. The District of Columbia’s Court of Appeals forcefully rejected the argument, observing that photographing arrestees was “one of the usual means employed in the police service of the country, and it would be matter of regret to have its use unduly restricted upon any fanciful theory or constitutional privilege.”

Thus, the earliest appellate cases reflect no condemnation of the routine collection biometric identification information following an arrest. However, the Louisiana Supreme Court expressed a different view in Itzkovitch v. Whitaker, 42 So. 228 (La. 1906), and Schulman v. Whitaker, 42 So.2d 227 (La. 1906).  In these cases, the court saw no reason to take photographs (and share them with other police agencies) before conviction. The Louisiana Supreme Court cited no opinions previously adopting this position. Clausmeier and Schaffer notwithstanding, the Whitaker court “found no precedent directly pertinent to the issues here.”

The Arkansas Supreme Court in Mabry v. Kettering, 117 S.W. 746 (Ark. 1909), followed Clausmeier and Schaffer. Three men charged with state crimes and held in a county jail sought an injunction against developing the negatives of photographs that their jailers had taken. The state planned to give the photographs to federal officials “for the purpose of identifying appellants in the various localities where [federal] offenses are charged to have been committed.” The state supreme court held that they were not entitled to the injunction. It wrote that “[t]he authorities cited by appellants in support of their claim for a temporary injunction clearly recognize the principle that public officers, charged with the enforcement of criminal laws, and having in their custody individuals charged with crime, may use photographs for the purpose of identifying the individual accused.” The “identification” here plainly involved the use of the photographs in separate and unrelated investigations.

In the same year, Maryland’s highest court also rejected the Louisiana approach. In Downs v. Swann, 73 A. 653 (Md. 1909), the Court of Appeals affirmed the dissolving of an injunction against taking photographs and bodily measurements to identify an arrestee. This court defined “[t]he precise question” as “whether the police authorities of Baltimore city may lawfully provide themselves, for the use of their department of the city government, with the means of identification of a person arrested by them upon a charge of felony, but not yet tried or convicted, by photographing and measuring him under the Bertillon system.” The court perceived no constitutional defect in acquiring the identifying information. As for the general state of the law, the court explained that “The right of the police authorities to employ the Bertillon process for the identification of convicted criminals has been recognized in most, if not all, of the jurisdictions in which the subject has received consideration, although several courts and text-writers have either questioned or denied the right to subject to that process persons accused of crimes before their trial or conviction.” It cautioned, however, that it was not countenancing “the placing in the rogues' gallery of the photograph of any person, not a habitual criminal, who has been arrested, but not convicted, on a criminal charge, or the publication under those circumstances of his Bertillon record.”

In 1915, the Washington Supreme Court rejected an offender’s demand for the destruction of postconviction photographs. Although Hodgeman v. Olsen, 150 P. 1122 (Wash. 1915), is not a case on arrestee data collection—indeed, the court judiciously noted that did not need to reach the question of preconviction data collection—the opinion sheds light on the judicial understanding of that question at that time. The Washington court “call[ed] attention to . . . cases explicit in affirming the implied police power to take, preserve, and make reasonable use of such photographs and data for the identification of persons convicted of crime, and even of persons accused of crime, but not yet convicted.” 150 P. 1122. The only appellate cases to the contrary were the isolated Louisiana ones.

Finally, in Miller v. Gillespie, 163 N.W. 22 (Mich. 1917), the Michigan Supreme Court held that an arrestee was not entitled to the destruction of identifying records even though the charges against him were dismissed at trial. It perceived absolutely no authority—including the Louisiana cases—“for granting relief . . . , unless it can be said that the mere preservation in the files of the police department of a report proper to be made in the first instance—a true report—exposes plaintiff to ridicule, obloquy, or disgrace.” Naturally, the court was unwilling to say any such thing.

In sum, by the 1920s, the Louisiana cases were the exception to the rule. Those opinions, and ones from trial judges reaching similar results, never were the majority rule. The main innovation of United States v. Kelly, 55 F.2d 67 (2d Cir. 1932), was its deliberate extension of the majority rule to misdemeanor arrests. It described fingerprinting of arrestees as "widely known and frequently practiced both in jurisdictions where there are statutory provisions regulating it and where it has no sanction other than the common law." Id. at 70. The opinion was not radical. If it was persuasive in expanding the established doctrine, its impact may have been related to its unusually detailed analysis of the law and policy and to the prestige of the court. Judges Augustus Hand, Learned Hand, and Thomas Swan comprised the panel--a veritable judicial powerhouse.

References

Annotation, Right to Take Finger Prints and Photographs of Accused Before Trial, or to Retain Same in Police Record After Acquittal or Discharge of Accused, 83 A.L.R. 127 (1933).

David H. Kaye, The Constitutionality of DNA Sampling on Arrest, 10 Cornell J.L. & Pub. Pol'y 455 (2001)

A.M. Kidd, The Right to Take Fingerprints, Measurements and Photographs, 8 Cal. L. Rev. 25, 32 (1919)

Wayne A. Logan, Policing Identity, 92 B.U. L. Rev. 1561 (2012)

"Human Error, Bias, and Malfeasance" in DNA Databases and Law Reviews

A new article in the Boston University Law Review offers the following warning:

[E]xpansive police arrest authority—and the desire to continually enlarge identity evidence databases at very little cost in time and expense—should give pause for several reasons. First, contrary to common public perception, DNA is not infallible. Rather, like other evidence, it is subject to human error, bias, and malfeasance, and has figured in several wrongful accusations and convictions. As Professor David Kaye notes in his recent book:

How probable is it that two, correctly identified DNA genotypes would be the same if they originated from two unrelated individuals? By definition, [such matches] do not consider any uncertainty about the origins of the samples (the chain-of-custody issue), about the relatedness of the individuals who left or contributed the samples (the identical-alleles-by-descent issue), or about the determination of the genotypes themselves (the laboratory-error issue).

Wayne A. Logan, Policing Identity, 92 B.U. L. Rev. 1561, 1580-89 (2012) (footnote numbers omitted).

Having searched without success for a single case in the U.S. of a false conviction based on DNA evidence from a database search,¹ I was puzzled. Could I have missed several false convictions arising from erroneous DNA testing? Did these cases involve database trawls, where observer “bias” is not normally an issue?

Being a lawyer, I did what any reader of law review articles must do. I turned to the footnotes. The footnote on false convictions as a reason to constrain DNA databases reads as follows:

See Greg Hampikian et al., The Genetics of Innocence: Analysis of 194 DNA Exonerations, 12 Ann. Rev. Genomics & Hum. Genetics 97, 107 (2011) (mentioning existence of at least fifteen exonerations in which DNA resulted in conviction).

If Professor Logan (and the source-citation reviewers of the Boston University Law Review) are correct, Professor Hampikian discovered at least 15 cases of DNA evidence that resulted in false convictions. How could I have missed all these case in my earlier postings?

The Genomics and Human Genetics review article plainly does not even begin to support the claim that DNA testing produced 15 false convictions. It merely states that among previously analyzed cases of postconviction exonerations, "there were at least 15 cases where DNA was tested prior to conviction." Hampikian et al., supra, at 107. Let's look at the outcomes of this DNA testing, as presented by Dr. Hampikian and his colleagues:

• The cited article does not even discuss the outcome of the DNA tests in two of the 15 cases because there were no "transcripts or other accurate information on the DNA results available." Id. Counting two cases on which there is no information as showing that contemporary DNA databases produce false convictions is surprising.
  
• "The majority of these cases included proper testimony, with DNA results that excluded the exoneree (9 of the 13 cases). These exclusions were explained away by the state in various ways—perhaps the defendant had an unknown codefendant, the DNA could have come from a consensual sex partner, etc." Id. Claiming that DNA databases should be constrained because most DNA typing accurately showed that a defendant was not the source of an incriminating sample is a blunder.
  
• "In 5 of the 13 cases, DQ alpha tests included the exonerees as possible contributors. In 4 of these 5 cases, however, more discriminating tests performed postconviction excluded the exonerees. In the remaining case, a second round of DQ alpha testing exonerated the defendant after it was discovered that the original lab analysis was incorrect." Id. Before the DQA test was retired from forensic DNA testing, it was known to be relatively undiscriminating. See, e.g., Cecelia A. Crouse, Analysis of HLA DQ alpha Allele and Genotype Frequencies in Populations from Florida, 39 J. Forensic Sci. 731 (1994); NFSTC, DNA Analyst Training. Questioning databases stocked with CODIS profiles because a different, bi-allelic locus has different properties is off target.
• "There were four cases where improper DNA testimony was given at trial. In one, the analyst testified about a match based on DQ alpha testing; however, the analyst did not disclose that it was only a partial match. In another case, the analyst did not provide the proper statistic for the population included by the results of DQ alpha testing." To be sure, "improper" testimony is deplorable, but it is not clear that the analyst in the first case incorrectly stated the implications of the match or, more importantly for worries about databases, that analysts working with database matches would give incorrect estimates of genotype frequencies.
• "In a third case, the analyst testified that the DNA matched the exoneree, but failed to disclose an additional exclusionary DNA result." Withholding exculpatory evidence of any sort—DNA, fingerprint, toolmark, eyewitness, or anything else—is unconscionable and unconstitutional. But it is not much of an argument against inclusive DNA databases.
  
• “In the final case, the analyst misinterpreted the results of the testing (which was performed incorrectly—failing to separate the male and female DNA during differential extraction), falsely including the exoneree as a source of the DNA when in fact he should have been excluded.” Yes, if crime-scene DNA is mistyped, and if this error goes unnoticed, a database match could result.

Can DNA databases produce false convictions? Of course they can. Police can commit perjury about DNA evidence, just as they can about other evidence. If there were no databases, it might be slightly harder to fabricate such impressive evidence. DNA evidence, like all evidence, “is subject to human error, bias, and malfeasance.” So are law review articles. And so are blog postings—corrections are welcome.

Note

1. David. H. Kaye, Have DNA Databases Produced False Convictions?, Forensic Science, Statistics, and the Law, July 7, 2012 (cross-posted to The Double Helix Law Blog); David H. Kaye, Genetic Justice: Potential and Real, Forensic Science, Statistics, and the Law, June 5, 2011 (cross-posted to The Double Helix Law Blog).