A. analysis

Failure to test in the genomics era will likely mirror what we already see with genetic testing in medicine. Mostly, failure to test will be the result of the provider's failure to recognize a risk in an individual. In many cases, family history is the known indicator that there may be an inherited risk for specific individuals, even if this family history does not follow a Mendelian pattern. However, in the case of many conditions (most of them rare) finding the gene or genes involved is difficult and in the absence of a clear family history through multiple generations, risk to individuals is often missed. Of course, such a failure will result in liability only if it fails to meet the standard of care — not all adverse outcomes or wrong choices are the result of negligence.

A classic example of failure to test involves the belief on the part of the patient, and often the clinician, that pathogenic variants in breast cancer genes cannot be passed down through a male in the family. A clinician may also fail to consider genomic testing for an individual who is the only person affected in a family where everyone else is healthy. It is not uncommon for clinicians to fail to consider that a single affected person might have a genetic condition and therefore fail to test that patient. This could result in failure to make a diagnosis for the affected individual and failure to identify others in the family (including potentially future children) who are also at risk.

There is a lack of guidelines for when to use genomic testing and how best to interpret and manage results. Genomic testing is not always the best choice, interpreting the results is often confusing, and the diagnostic yield is variable depending upon the indications for testing and laboratory that completed the testing. How to follow up on a test that failed to yield a diagnosis may not be clear, and the fact that the results may look different when coming from different laboratories complicates the decision to test. Lastly, the literature continues to show that health care providers not trained in genetics lack the knowledge and skills needed to assess risk.^{Reference Mikat-Stevens, Larson and Tarini42}

B. recommendations

Recommendations for Health Care Actors or Institutions:

• In order to ensure quality care and to prevent liability, health care organizations should determine who within their organization is qualified to order genomic testing. Appropriate training is necessary to understand the type of genomic test to order (e.g., gene panel vs. GSS), communicate the results to patients, and incorporate the results into patient care. If a clinician does not have relevant training, the clinician should refer the patient or obtain a clinical consult with a provider (physician or genetic counselor) who does have the requisite training.

• Professional societies should give higher priority to issuing clinical practice guidelines, guidance, and/or recommendations to assist clinicians in making genomic testing decisions. In the absence of updated clinical guidelines, health care organizations and professional societies should provide guidance on appropriate use of genomic testing that incorporates the perspective of all relevant stakeholders.

• Health care institutions have responsibilities to offer or facilitate access to (1) continuing education on these topics, (2) genetic and genomic laboratory capacity and professionals, and (3) clinical processes (including clinical decision support (CDS) and adequate electronic health record (EHR) representation of genetic and genomic tests results) to support good clinical practice.

• Genetic counselors have a crucial role to play in clinician, patient, and family education and counseling.

• When genomic testing is indicated, the clinician should discuss this with the patient and offer a referral if the patient's health care institution does not offer genomic tests.

Recommendations for Legal Development and Change:

• When clinicians are accused of negligence or malpractice for failure to order genomic testing, attorneys and courts should access and consider clinical practice guidelines (CPGs), other recommendations and resources from professional societies, and the relevant medical literature, which may be germane, to but not determinative of, the standard of care.

• Because genomics is a developing field, doctrines recognizing multiple schools of professional thought may be relevant to adjudication. Highly competent genomics professionals may disagree on the question of when testing is indicated, the appropriate type of testing, and the interpretation of results.

• In some domains of genomics, standards are already clear on when testing is indicated. When that is not the case, it would be helpful to have professional guidance stating current uncertainties and options in order to make clear that providers must make professional judgments in light of uncertain evidence. In such situations of uncertainty, health care institutions and providers can protect themselves from liability by documenting that they had in place a reasonable process for deciding whether and what kind of genomic testing is warranted in light of the uncertain evidence, and that they explained the available choices to the patient.

A. analysis

When considering testing, it often seems that more is better. The more data that can be gleaned from genomic testing, the greater the likelihood of an answer for the patient and the family, or so one would think. However, in the current state of genomic testing, this is often not the case. Whole genome and exome testing can lead to information that is not wanted, does not aid the diagnosis or management of the case, and cannot be interpreted. In some cases, the results of GSS can give results that can be misinterpreted or over-interpreted, which can potentially harm the patient by encouraging additional testing and even medical procedures that may not be warranted and might even be dangerous.^{Reference Warring, Giudicessi, Wilde and Ackerman43}

An exome as opposed to a targeted test does not have the same coverage of potentially critical genomic regions. Exome testing could miss the critical area that might lead to a diagnosis. Exome testing sequences the coding regions of the genome, covering about one to two percent of the genome.^{Reference Manolio44} It is designed to provide a certain depth of coverage to ensure accuracy.^{Reference Sims45} However, this is less than the coverage that would occur when testing for a specific gene. The increased potential for error in exome sequencing could result in failure to identify an area critical to diagnosis. In contrast, a targeted panel is more comparable to a single gene test. It is more specific, and the coverage is higher. A significant risk to clinicians is believing that whole genome or exome sequencing “looked at everything” and thus assuming that if nothing was found, then there is no genetic basis for the phenotype. This assumption may be incorrect.

B. recommendations

Recommendations for Health Care Actors or Institutions:

• Clinicians should determine the scope of testing to order in a particular case by considering clinical practice guidelines (if any), other recommendations from professional societies and the scientific literature, and consultation with their testing laboratory.

• Clinicians ordering genomic analyses should be familiar with the testing options, their characteristics, and the trade-offs among them, in order to make a well-grounded determination of which tests to order.

• The scope of testing or sequencing ordered should optimize clinical utility (e.g., diagnostic yield), while minimizing results that are not relevant to the indication for testing. Parsimony in testing can be beneficial to prevent false positives and results that may be misinterpreted or over-interpreted.

• If clinicians, their laboratory, and/or their health care institution consider enlarging the scope of testing to include a set of secondary or incidental findings (e.g., those recommended by ACMG^{Reference Green, Green, Rehm, Kohane, Ginsburg, Willard and Green46}), this should be discussed with the patient as part of the informed consent process before testing. The patient should have the option of refusing some or all of these added analyses.

Recommendations for Legal Development and Change:

• Courts should recognize the over-testing (i.e., ordering tests not called for by the patient's symptoms, conditions, or history) can produce false positive results that can harm patients by requiring additional follow-up procedures that may prove iatrogenic.

• Most genetic malpractice cases to date involve a provider's failure to test for a particular genetic condition or variant. However, due to the uncertain nature and large volume of data generated by genomic tests such as GSS, it may be negligent to recommend over-testing in some situations.

A. analysis

Genomics creates an additional level of complexity compared to genetics in requiring that a physician or other health care provider choose which specific test(s) to order for a particular patient. In the past, the provider would assess whether the patient had a predisposition or illness that might be informed by genetic testing, and then would recommend a genetic test for that condition. Typically, this would involve testing a single gene. For example, a physician or genetic counselor may recommend carrier testing of a prospective parent for a mutation in the CFTR gene associated with cystic fibrosis or testing for mutations in the APC gene for a member of a family with a history of familial adenomatous polyposis (FAP). While different test manufacturers or labs may offer their own test for that particular gene, perhaps with different variants assayed by the test, there is usually no issue as to what gene to test for, only which test might include the most robust and relevant set of variants for that single gene.

An important difference in the genomic era is the shift from testing of a single gene to testing panels of genes or even exome or genome sequencing.^{Reference Domcheck, Nathanson, Minion and Kamps47} With the realization that the risk of almost all diseases (at least all non-Mendelian diseases) is affected by many genes, a patient believed to be at risk for a specific disease can best be assessed with a gene panel that packages into one test platform assays for a number of different genes that may affect the risk of that particular disease. Disease risk may be determined by “hits” within specific genes included in the panel, or by an algorithmic score that takes into account the testing results for all the genes in the panel. Test manufacturers and testing laboratories are increasingly using gene panels to test patients for a specific disease risk such as GeneDx's panel for epilepsy, which currently includes over 200 genes associated with multiple disorders that may involve they symptoms of epilepsy. As more genes are found to be associated with epilepsy, companies can add those genes to the panels. Another example is the CustomNext-Cardio panel by Ambry Genetics.⁴⁸ In this case, the panel allows the clinician to choose a group of disorders from a list of potential diagnoses such as lipid disorders, aortic aneurysms, cardiomyopathies, etc., to test for pathogenic variants associated with these diagnoses.

The novel liability risk associated with gene panels is that the health care provider may face liability if he or she recommends an inferior panel for an individual patient, and it turns out that the testing misses an important variant that would have been detected if a better panel had been recommended. Bringing a lawsuit for such a faulty recommendation would present some practical proof problems. The plaintiff (or their attorney) would need to discover that another gene panel was available and would be covered by the patient's insurer or funded by the patient, and that the alternative panel had genes in it that would have been more informative and affected the outcome if the physician had recommended that panel instead. Often potential plaintiffs and their attorneys will not be aware of the potential consequences of testing with a different gene panel, but if this is discovered by a plaintiff's expert, for example, who finds that the plaintiff was adversely affected by the failure to detect a mutation that was included in an alternative panel, this could be the basis of a potential medical malpractice claim. Specifically, the plaintiff would need to show that (1) an alternative gene panel was available and would have been utilized by the patient, (2) that this alternative panel would have disclosed information that would have resulted in a better outcome for the plaintiff, and (3) the physician should have had some reason to suspect that the alternative gene panel would have produced a superior result. This opens an additional potential claim for a plaintiff alleging genomic malpractice.

The choice of an appropriate panel may also implicate the lab or institution. Traditionally, the clinician orders a test (often designated by the patient's health insurer), and the lab performs that test and reports results back to the physician. As we move into the genomic era, however, the relationship between the clinician and test lab is becoming more complex and variable. In some cases, the lab may provide feedback or recommendations on the test panel the physician ordered, sometimes suggesting a different panel or perhaps adding in a few additional genes based on the individual patient. As labs perform this advisory role, physicians may actively consult the lab for recommendations on what gene panel to order. The choice of gene panel will be influenced by a number of factors in addition to the genes present on each panel, including each panel's coverage of relevant gene variants, the rate at which they detect specific variants, their cost and available insurance coverage, and how good they are at interpreting variants of uncertain significance (VUSs). In other institutions, the labs play an even more active role, and are in active partnership with clinicians in the patients' care. In still other contexts, the lab has developed its own test panel, and gives the physician the binary choice of whether to order that test panel or not. Depending on how active a role the lab had in selecting the specific test panel, the lab may assume some or all liability for an erroneous choice of a test panel. The health care institution may also bear responsibility if it creates a process or committee that reviews or recommends test orders.

B. recommendations

Recommendations for Health Care Actors or Institutions:

• Given the uncertainty and rapid evolution of genomic tests, greater transparency on the evidentiary basis and performance of available genomic tests would be highly beneficial, including those commercially available.

• The government or professional societies should publish and regularly update a listing and comparison of test panels and sequencing approaches that can be used to evaluate a particular genetic risk.

• Professional societies should whenever feasible identify specific tests or test parameters that are recommended for patients with specific risk profiles, and physician compliance with those recommendations should be evidence of reasonable care. These recommendations need to be updated on a regular basis as new tests and evidence become available.

• When the gene variants or algorithms incorporated in a test are changed, there should be a process and standardized format to notify providers of this change and the basis for the change.

Recommendations for Legal Development and Change:

• Given the trade-offs that may be inherent in constructing genetic test panels and algorithms, there should not be liability for test selection simply because it produces an inferior result in a specific patient.

• The choice of a test is particularly prone to the well-known phenomenon of “hindsight bias” in litigation,^{Reference Rachlinski, Harley, Roese and Vohs49} in that various testing options may have strengths and weaknesses that may look very different before and after the testing is done and outcomes occur. Courts should restrict, under the rule against prejudicial evidence, attorneys from insisting that all negative outcomes show malpractice and should instruct juries carefully to avoid hindsight bias.

• The learned intermediary doctrine^{Reference Dobbs, Hayden and Bublick50} should be interpreted as requiring test developers to be transparent with ordering clinicians and labs about the strengths and weaknesses of their tests in order to get protection from liability under the doctrine.

A. analysis

The shift to genomic medicine will provide a much larger set of data on which health care providers can rely for disease prediction, diagnosis, prognosis, and treatment. The availability of data from genomic tests will inevitably lead to second-guessing of how a provider used that data in a patient's care, especially when the patient had a bad outcome.

An example is reliance on a genomic test that predicts recurrence of breast cancer, and thus whether a patient should undergo chemotherapy after surgical removal of a tumor. In the past, many breast cancer patients unnecessarily incurred the costs and side effects of chemotherapy, because no information was available on their individual recurrence risk. Now genomic tests such as Oncotype Dx and MammaPrint predict the recurrence risk for an individual breast cancer patient. These tests are probabilistic not deterministic, and thus cannot predict with certainty whether a breast cancer will recur, but rather provide a probability known as a recurrence score. If a treating physician recommends that a patient undertake such testing, and if (for example) the test provides a low recurrence score so that the patient on the advice of the physician forgoes chemotherapy, can the patient sue the physician if the cancer nonetheless recurs? Or alternatively, if a physician fails to recommend such a test, and the test would have indicated a high recurrence score that would have indicated chemotherapy, would a patient have a claim against the physician for failing to recommend such a test?

An important principle of medical malpractice is that a physician is not, and should not be treated as, a guarantor of the patient's good outcome.^{Reference Pegalis, Moffett and Moore51} Most medical procedures and decisions involve inherent uncertainties and risks, and sometimes the treatment will result in a bad outcome even if the physician's judgment to recommend that course of treatment was reasonable at the time the decision was made. Nevertheless, when a patient does have a bad outcome, there will often be a temptation to second-guess the physician's decision on whether or not to have recommended a genomic test that may have resulted in a different outcome.^{Reference Marchant, Lindor, Marchant, Lindor, Lindor and Marchant52} A practical difficulty in bringing this type of lawsuit is that a patient may never know if their result would have been different if the test had or had not been recommended, because the test results are probabilistic rather than deterministic.

At least one such case has already been reported. In that case, a breast cancer patient was diagnosed as having a non-invasive cancer, which was treated but later returned and metastasized. The patient alleged that if she had been given the Oncotype Dx assay her tumor would have produced a recurrence score of 41, which would have indicated the need for chemo-therapy, which may have prevented the tumor recurrence.^{Reference Carrier53} She brought a medical malpractice lawsuit against her physician for failing to recommend an Oncotype Dx assay, and the case eventually settled.⁵⁴

B. recommendations

Recommendations for Health Care Actors or Institutions:

• The lack of clinical data to evaluate predictive value and clinical utility for many genomic tests presents a challenge to providers in determining whether a particular test is valid and reliable. Researchers, test developers, regulators, payers, and professional societies should work to ensure that sufficient clinical evidence supports available genomic tests, including those commercially available.

• Professional societies and others generating clinical practice guidance need to provide physicians with clear recommendations regarding what genomic tests are ready for application in the clinic, and for what indications or risk profiles.

• The reliability of genomic tests will depend on the robustness of the data on which they are based, including data that are representative of the patients for whom the test is recommended. Test developers and research funders should ensure that genomic tests are based on data from a diverse and representative cross-section of the population.

• Physicians need to explain carefully the limitations and uncertainties associated with the specific genomic tests they recommend to their patients, so the patient can make an informed decision on whether to assume the risk and make medical decisions on tests that are inherently imperfect (at least for the time being).

Recommendations for Legal Development and Change:

• Many genomic tests will give probabilistic rather than deterministic outcomes — in other words, the test will indicate that patients may have a certain risk. The fact that an individual patient has a bad outcome is not in and of itself evidence of negligence, and in this area of clinical practice as in others, the clinician cannot be a guarantor of the patient's positive outcome. Rather, the negligence inquiry should focus on whether the clinician's recommendation of and reliance on a test was reasonable in the circumstances of the individual patient and consistent with the prevailing standard of care.

• A key factor in whether a provider is negligent in recommending or not recommending a genomic test for a particular patient is whether the provider adequately explained the options, uncertainties, and risks of different courses of action. Documenting such discussions will often be critical to defending against liability.

A. analysis

Determining whether a particular gene variant is of clinical significance, and what should be communicated to the patient, is a major challenge in genomic medicine, particularly when exome or genome sequencing is conducted. Such sequencing will invariably (at least for the foreseeable future) generate a large number of VUSs.^{Reference Lindor, Thibodeau and Burke55} Both the testing laboratory (which usually makes the initial “call” on the clinical significance of variants) and the patient's physician or genetic counselor who communicates the test results to the patient will have responsibilities that may be second-guessed in litigation.

There is often inconsistency between reference databases on whether a particular variant is pathogenic, likely pathogenic, or of uncertain significance. If a test laboratory or provider reports to the patient that a variant was of uncertain significance (a VUS), when in fact the variant turns out to be clinically significant and the delay in understanding the risk results in injury, the patient could bring a lawsuit claiming the lab or provider erred in classifying the variant as a VUS. The playing field in such cases may be tilted against the defendant if subsequent information makes clear that the variant is significant — things always look clearer in hindsight, even though the legal fact-finder is supposed to evaluate the decision based on the information available to the defendant at the time the challenged action occurred. If the lab or provider did not report the variant to the patient at all because it was deemed to be of uncertain significance, this action could again be second-guessed in a legal claim for failure to disclose.

The first case presenting such issues is currently pending in South Carolina federal court—the Williams v. Quest case.^{Reference Thorogood, Cook-Deegan and Knoppers56} In that case, a child suffering from seizures was genetically tested, and a variant determined to be of uncertain significance was detected in a potentially relevant gene known to be associated with epilepsy (SCN1A). The child died from his condition in 2008, after being treated with a drug that is now known to be contra-indicated by the gene variant carried by the child. The testing lab issued a revised report in 2015 re-categorizing the VUS as a pathogenic mutation. The mother subsequently brought a lawsuit contending that the testing lab should have known the variant was pathogenic at the time it issued its original report, or that it should have updated its report promptly when it did discover that the variant was pathogenic and communicated that revised result to the patient.

B. recommendations

Recommendations for Health Care Actors or Institutions:

• Inconsistent classifications of the same variant by different sources and databases present a major problem for communicating test results to patients. Ongoing efforts to reconcile or integrate different variant databases will be very helpful in providing clearer guidance to labs and physicians in interpreting the clinical significance of genetic variants.

• A physician should be responsible for communicating and explaining to a patient any variants reasonably classified as pathogenic or likely pathogenic that are reported to the physician by the lab in response to the physician's order for genomic testing.

• The physician should also be able to explain “negative” results (benign or likely benign) if relevant to the patient's condition, risks, or concerns.

• There is controversy over whether and when providers should communicate VUSs. When the patient has a family history or other evidence of a specific genetic problem, and testing of the relevant gene(s) indicates only VUSs, the physician should communicate information about the specific VUS if clinically appropriate and urge the patient to continue to follow-up on the relevant variant(s) going forward in future interactions with health care providers.

• Policy makers, health care institutions, and professional societies should seek to integrate different interpretive databases to provide the most robust and consistent evidence on the pathogenicity of genetic variants.

Recommendations for Legal Development and Change:

• Courts and attorneys should pay careful attention to clinical practice guidelines and other recommendations and guidance from professional associations on making variant calls. These resources provide helpful evidence on reasonable practices even if they do not conclusively establish the standard of care.

• A physician may be held liable for failing to disclose and explain a variant that has been identified as “pathogenic” by the test laboratory. However, if the lab uses more than one database for making variant calls, and reports that the databases give inconsistent results on whether a particular variant is pathogenic, the physician generally should disclose to the patient both the pathogenic and non-pathogenic determinations and explain why different databases may give inconsistent results and what this means for the patient. A physician may offer their own professional opinion as to why a particular database result may or may not be compelling.

• A physician should not have a legal duty to go beyond the lab report and look at other sources that may classify a variant differently, unless that physician has good reason to know or suspect there are different interpretations available. Failure to disclose alternative variant interpretations should be the responsibility of the lab, not the physician. As with its other testing operations, a laboratory should have a legal duty to consult known and respected data sources in providing variant interpretation to the physician (and ultimately to the patient).

• A physician should not have a general duty to communicate and explain specific VUSs to a patient. However, in cases where the patient is known to be at risk of a disease, the physician should explain VUSs in genes of relevance, caution that some of these variants may become of known significance in the future, and inform a patient that he or she should consider seeking re-interpretation of these test results at a later time.

A. analysis

Genomic medicine will impose new opportunities and pressures for health care providers to recontact patients to provide updated interpretations of genomic information. Health care providers have always had a legal and ethical duty to try to recontact previous patients when they discover, or should have discovered, that their initial diagnosis or treatment recommendations were erroneous when originally provided.^{Reference Camp104} In the context of genomics, the question of whether the initial diagnosis, interpretation, or recommendation was erroneous will not be as clear-cut as in some other areas of medical malpractice. Given that a particular variant could be categorized as ranging from benign to pathogenic, and different databases may reflect divergent expert interpretations and classifications of the same result,^{Reference Van Dries105} it will not always be clear, even in retrospect, whether the initial decision of the physician to classify a variant or recommend a treatment regime based on that classification was erroneous at the time.

However, when the provider's actions were not erroneous at the time the advice was delivered, courts have recognized a duty to recontact patients in only two limited circumstances.¹⁰⁶ The first is when a physician administers a procedure or treatment that requires an established follow-up procedure or test, and if such follow-up is not provided, the physician could be liable for “abandonment.”^{Reference Drechsler107} The second situation arises when a patient is being treated with a drug or device prescribed by the physician, and the physician learns of new risks associated with that treatment, such as a new warning issued by the FDA.¹⁰⁸ The theory behind these two limited duties is that the patient is still undergoing treatment that the physician initiated.

Other than those limited situations, courts have not recognized a general duty for a health care provider to affirmatively recontact their patients with new information.¹⁰⁹ Genomic sequencing presents new questions about whether a physician and testing lab can and should update results and recontact patients. Genome sequencing will initially identify many VUSs, but their clinical significance will change over time. This then raises the questions — which professional societies have begun to address — of whether there is a duty for testing labs or physicians to determine whether a variant's reinterpretation applies to previous patients, and if so, if the new information is clinically relevant, whether the testing lab and/or physician has any ethical or legal duty to try to recontact the patient with that updated information.^{Reference Bombard, Klast, Wimberger, Arnold, Macklin, Davis and Carrieri110}

A recent survey of the literature on a physician's legal duties to recontact a former patient to provide later findings identified a number of articles discussing a potential duty to recontact, such that the failure to do so could result in liability to former patients under a theory of negligence.^{Reference Otten111} Approximately one half of the articles reached no clear conclusion regarding whether there is a duty to recontact former patients to provide updated genomics findings; of 28 that drew a conclusion, “6 concluded that a duty to recon-tact does apply and 22 concluded that it does not.”¹¹² Accordingly, scholars have concluded that “there is no generally held legal basis for recontacting in clinical genetics, although it is often considered desirable by both [health care providers] and patients.”¹¹³ However, no court has yet ruled on this potential legal duty in the context of updated genomic interpretations.^{Reference Ray114} A number of factors are relevant to any potential future ethical or legal duty to update genetic interpretations and recontact patients.

Imposing a responsibility on physicians to identify when new information changes the genetic risk for former patients is problematic, as most physicians will lack sufficient knowledge and confidence to identify and address the need for an update.^{Reference Selkirk115} Any potential future duty of physicians to update persons who receive genomic tests would likely vary depending on the nature, and inherent expectations, of the physician-patient relationship. The potential that the physician-patient relationship could give rise to a recognized duty to recontact or update patients could vary in at least three sets of circumstances, depending upon the relationship of the genomic results to the area of treatment.

First, a physician could order a genomic test that reveals a variant that is later determined to be associated with a disease risk unrelated to the physician's area of practice, and therefore outside the area of care that the patient received from the physician. In such a case, there would be a relatively lower risk that a court would impose liability on the physician for not recontacting and updating the former patient.

Second, the physician could order genomic sequencing in the course of providing continuing care to the patient. If the test identifies a genomic variant associated with a disease risk related to ongoing care, a court would be much more likely to determine that the physician has a duty to recontact and update the patient. However, a court might be less willing to impose liability on a physician in the context of offering ongoing treatment if the new genomic variant were in a field unrelated to the physician's area of practice and if the physician were unaware of the relevance of the new variant to the patient.

Third, there is the possibility of an intermediate scenario, in which the patient sees a physician for a discrete episode of care (e.g., cardiac surgery and follow-up) after which the physician's relationship with the patient ends. If a genomic test were ordered in the course of the treatment relationship and then a variant identified by the genomic test were later determined to be associated with a disease risk related to the problem for which the patient sought care, it is unlikely that the physician would have a duty to recontact the patient and update regarding the risk; the physician has no ongoing relationship with the patient, and one episode of care cannot impose a lifetime, continuing duty on a provider. Indeed, some scholars have suggested that any duty to contact would be “unlikely to extend in perpetuity or to require more than reasonable effort.”¹¹⁶

In contrast to physicians, testing labs generally do have the requisite expertise to recognize and conduct reinterpretations of previous VUSs. Indeed, some labs are now periodically rechecking their previous interpretations and sending updated interpretations that are clinically relevant to the physician who ordered the genetic testing.^{Reference Daniel117} However, there are no laboratory guidelines, court decisions, or other statements of standard of care that impose a duty on testing labs to periodically reinterpret genetic test results. Moreover, test labs generally do not have the immediate relationship or the proper training to report new genetic interpretations directly to patients.

If a responsibility to update genomic test results does emerge, it would likely put the onus on labs to identify the reinterpretation and then to send it to the ordering physician for communication to the patient. Yet, creating such a duty would involve significant practical and implementation issues.¹¹⁸ How long would the hypothetical duty to recontact extend into the future — for a limited time period or in perpetuity? Perhaps it would extend as long as there is a physician-patient relationship between the provider and patient at issue, but it is not always clear when such a relationship ceases. Is it when the doctor last sees the patient, or has not seen the patient for some length of time, or when the patient has changed their care to another physician? The end date of a physician-patient relationship is often not clear, and may depend on whether it was a one-time appointment with a specialist, a repeated series of appointments with a primary care physician, or perhaps a comprehensive care arrangement such as with an accountable care organization or institutional provider like Kaiser. In the past, a clear termination date for the physician-patient relationship may not have been legally significant in most cases, but given the looming possibility of an ongoing duty to update interpretations of genetic test results, it may behoove providers to contractually specify a more precise termination date for their care.¹¹⁹

Many other practical difficulties confound a possible ongoing duty to update. Who would pay for the additional efforts by the lab and physician? What if the patient has moved and is not easily located? What if the patient did not want to know the updated results? What if the patient has died in the interim? These and other practical difficulties in implementing a possible duty to update should be resolved before any such duty is imposed.

In the future, information technology and patient apps may help make the reinterpretation tasks more routine and patient-focused. It is likely that computer programs and apps will be developed that automatically notify patients when clinically significant new information becomes available that affects the interpretation of their genetic results. Indeed, a program called “GenomeConnect: The ClinGen Patient Registry” has already been launched that among other things alerts patients to significant updates in the interpretation of their genetic test results.¹²⁰

B. recommendations

Recommendations for Health Care Actors or Institutions:

• If a provider discovers that the previous information communicated to the patient was erroneous at the time of the communication, the provider should notify the patient of the corrected information as soon as possible.

• Health care providers who are responsible for ordering or interpreting genetic tests for a patient should expressly inform the patient that some of the results may have a different or changing interpretation in the future.

• Some testing labs are periodically updating their interpretations of previous patient test results in light of constantly evolving information. If during such a review, the lab determines that the interpretation of a patient's genetic information has changed to now provide an actionable response to a significant risk, the lab should make reasonable efforts to communicate that revised finding to the ordering physician.

• If a physician is notified by a testing laboratory that a previous interpretation of a patient's genetic variant has changed in such a manner that may materially affect the patient's health, the physician should make reasonable efforts to pass on the updated interpretation to the patient. In many or even most cases, this may not be feasible, due to patient loss to follow up and limited provider resources.

• A physician who is treating a patient for a condition and who refers the patient for a genetic consultation that provides no clinically relevant information should consider re-referring the patient for an updated genetic consultation every few years if clinically indicated (e.g., condition has not been resolved and is not being successfully treated and physician has reason to suspect there may be a previously undetected genetic underpinning).

• Any duty of a physician to recontact and update a patient regarding newly identified genomic testing results should depend on various factors including: (1) the correspondence, if any, between the physician's area of expertise and the genomic result, (2) whether the patient is under the physicians' current and continuing care, and (3) the period of time that has elapsed since the physician's care.

• In the future, software programs and apps may become available to automatically notify patients and their treating physician that a clinically relevant updated interpretation of their genomic data is available.

Recommendations for Legal Development and Change:

• Courts and legislatures have to date established no general legal duty for a test lab or health care provider to continuously update previous interpretations of a patient's genetic variant(s).

• A provider who is actively treating a patient for a specific condition, and that patient has been genetically tested in the past, should be held legally responsible for taking reasonable steps to communicate updated clinically-relevant genetic information to the patient if the updated information has been communicated to the physician and that information is within the physician's scope of expertise, or would otherwise be within the standard of care for the physician to be aware of the updated information.

• A health care provider who is provided with a materially updated interpretation of a patient's genetic information from a test lab should try to make reasonable efforts to pass that information on to the patient. Although no legal duty currently exists for a provider to communicate those updated test results to their patient, providers should be aware that courts may impose such a duty in the future, perhaps retroactively. However, legal systems should understand that this recontact is in many, even most, cases not feasible, given patient loss to follow up and limited provider resources, and thus should condition any such duty on a reasonableness determination.

• Courts and legislatures to date have imposed no general duty on testing labs to regularly update the results from previous patients or to alert ordering physicians of clinically relevant updates. However, given that some labs have started to do such updates as a regular part of their business, they may be establishing a new standard of care, especially in jurisdictions where test labs are considered to be health care providers that are subject to a standard of care based on local or national custom.

• Professional societies should continue to refine their guidance and provide recommendations on what is a reasonable effort that providers and test labs should undertake to update patient information with relevant new findings. Relevant factors should include the nature of the patient-provider interaction, the duration since the last patient appointment, whether the patient has provided the provider with their most recent address and contact information, the clinical significance of the new findings, and how often any update required should occur.

A. analysis

Health care providers have been concerned for many years about potential liability relating to failing to warn family members of patients about their genetic risks.^{Reference Clayton, Suter, Offit, Branum and Wolf121} Genetic information differs from most other medical data in that it is directly relevant to family members who may share the patient's genetic variants that confer a disease risk. Courts have held that a health care provider has a legal duty to warn relatives of a patient of genetic risks that become apparent through treatment of the patient by instructing the patient to warn their immediate family members of their potential genetic risk.¹²² However, with some limited exceptions discussed below, courts have generally not held that providers have a duty to communicate genetic information to a patient's relatives directly, and such disclosure without the patient's consent is generally prohibited by the Health Information Portability and Accountability Act (HIPAA). Some medical association guidelines such as those from the American Society of Human Genetics (ASHG) and National Society of Genetic Counselors (NSGC) suggest that a health care provider may (but it is not required to) directly disclose genetic risk information to a patient's relatives in the absence of the patient's consent in rare, exceptional circumstances.¹²³ Some experts contend, however, that any such narrow discretionary right to disclose genetic information to a patient's relatives was foreclosed by the 2003 privacy rule implementing HIPAA (discussed further below).^{Reference Rothstein124}

This duty to warn the patient to disclose family risks may be more difficult and complex to discharge in the era of genomics than genetics. With genetics, usually a mutation in a single gene is at issue, and therefore it is relatively straightforward for a provider to discern and communicate to the patient the inheritance pattern of that mutation (e.g., dominant vs. recessive, autosomal vs. sex-linked). The potential risks to immediate family members can then be conveyed to the patient with a recommendation that such information be passed on to blood relatives. With genomic test results, including test results of a large number of gene variants with varying degrees of significance, the task of determining what information (if any) should be communicated to the patient's relatives is much more difficult. For example, some of the mutations in a patient's genome may have been inherited, while others may be de novo, especially with mutations such as copy number variants (CNVs). Even for variants that appear to have been inherited, there is a lack of clarity on how much clinical significance and certainty about a variant is necessary to trigger a duty by the provider to counsel the patient to communicate the risk information to relatives. Even if a physician's duty is discharged by recommending information for the patient to tell his or her direct relatives, the shift from genetics to genomics will make this legal duty more complex and potentially subject to retrospective criticism in any litigation.

At the same time, the shift from single-gene genetics to genomics in health care may create more cases where there may be a duty to disclose to relatives for a couple of reasons. First, as a purely quantitative matter, as more people get their genes tested, there will be more cases where communication of those results to relatives will be clinically important. Second, to the extent that genomics progresses to various types of population-level screening, this too will create more opportunities and legal risks relating to disclosing information to relatives. For example, if sequencing of newborns becomes more prevalent, this will create enormous additional family disclosure issues. An example of a genomic test creating a duty to relatives is Polaski v. Whitson, where the court ruled that a physician treating a patient for heart disease may have legally harmed the patient's son by failing to order a gene panel test for the father for hypertrophic cardiomyopathy, which might have then informed the son's disease risk.¹²⁵

The increased number of potential cases where genetic information may be relevant to a patient's relatives is offset at least partially by the fact that as more people undergo genome testing, they will no longer need to receive their relatives' results to learn of risk, since they will have the results from their own genome. However, the long-standing fear of many physicians is that the duty to disclose to relatives could be imposed directly on the physician, such that the physician will be tasked with informing the relatives directly if the patient is unwilling or incapable of doing this. This fear was catalyzed by the 1996 decision in Safer v. Pack,¹²⁶ where a New Jersey court suggested that a doctor may have such a duty in at least some circumstances. The New Jersey legislature subsequently overturned the decision by enacting a genetic privacy statute that required patient consent for disclosure,¹²⁷ and no subsequent court case in the United States has held that a physician has a legal duty to inform relatives of their genetic risk over the patient's objections (or even without the patient's permission).¹²⁸

To obligate physicians to warn relatives directly would be problematic on both practical and legal grounds. First of all, disclosing over a patient's objections violates the clinician's millennia-old obligation to protect confidentiality. From a practical perspective, if a patient is uncooperative and unwilling to notify their own relatives, it will often be difficult for a physician to identify and locate at-risk relatives, who may be scattered in faraway places. The costs and burdens of contacting and communicating genetic risks to such relatives would likely not be reimbursed and could be quite substantial. Moreover, the relatives who receive the genetic warnings may not want to be informed of their genetic risks, having not consented to any communication, or may even be minors.

Legally, disclosing a patient's health information to relatives without the patient's approval could violate the HIPAA privacy rule.¹²⁹ There are HIPAA exceptions to the consent usually required for the disclosure of medical information by a health care provider or institution that could theoretically apply to the unconsented disclosure of genetic risk information to the patient's relatives, including disclosures “to avert a serious threat to health or safety” or where “required by law.” These exceptions have primarily been applied to imminent and serious risks from contagious diseases or mental health problems. While there has been some debate in the academic literature about whether either of these provisions could apply to familial disclosure of genetic information, the nature of harm from non-disclosure of genetic information is unlikely to rise to the level of “serious threat” of harm and is not included with the regulatory explanation of the “required by law” exception.¹³⁰ Furthermore, the 2013 amendments to the HIPAA rules do not expressly extend these HIPAA exceptions to cover sharing genetic results with a patient's relatives without the patient's consent.

HIPAA does allow a physician to send results about a patient's genetic risks to another provider treating a relative of the first patient, provided the patient has not forbidden such sharing.¹³¹ This allows sharing genetic results within families as more and more providers consider genetic data in their clinical care of patients. However, if the patient requests that information not be shared through this mechanism, and the physician agrees with that restriction, then the physician may not share genetic data with a physician treating a relative.¹³² Some other exceptions may apply in certain situations. Researchers and other actors outside the clinical context that work in institutions that are not “covered entities” under HIPAA are not precluded by HIPAA from sharing genetic test results with relatives in appropriate circumstances.¹³³ Moreover, some states assign the rights to medical information (including genetic data) to relatives of a deceased patient.¹³⁴

Finally, because HIPAA allows a personal representative of the patient to gain access to a patient's medical information, some relatives will gain access to patient genetic data through that mechanism.¹³⁵

However, as a general matter, a physician has no duty to disclose patient genetic risk information to the patient's relatives, but may risk violating HIPAA if the physician attempts to disclose such information without the patient's consent (and the patient is still alive¹³⁶).

B. recommendations

Recommendations for Health Care Actors or Institutions:

• Clinicians ordering genomic analyses including sequencing should be aware of key recommendations in the literature and from professional societies on sharing results and data or data files with family member(s) or their physician(s), including after the patient's loss of decisional capacity or death. Legal counsel should additionally be aware of the relevant law and legal analyses, in order to provide advice when needed.

• Genomic data and interpreted results qualify as an individual's protected health information (PHI) and should generally be protected as private and confidential information. Researchers and clinicians should encourage individuals to share their results with family member(s) when those results have implications for those family member(s), and should offer support for that process as needed. It may be particularly helpful to provide the patient with an informational document to share with family members.

• Clinicians should specify their approach to disclosure of results to family members when the patient is a minor, an adult without decisional capacity, or is deceased. Clinicians should consider eliciting the patient's preferences on sharing results with family members, including after participant loss of decisional capacity or death.

• Clinicians should address how they plan to handle requests for patient results from the patient's family member(s), parent(s) or guardian, the patient's Legally Authorized Representative (LAR), or the patient's Personal Representative (under state and federal law); the LAR and/or the Personal Representative may be a family member.

• Clinicians and test labs should address how they plan to handle requests for the patient's raw data or data files from any of those listed above.

• There may be cases in which the patient refuses to share results that have a high likelihood of averting imminent harm if shared. In these instances, the clinician may have a privilege to warn the family member(s) at risk, although some legal experts argue that any such disclosure would violate HIPAA. Ethics and legal consultation is recommended before sharing.

Recommendations for Legal Development and Change:

• Provisions in HIPAA and state law allowing broader sharing of genomic PHI with family members should be scrutinized and amended if necessary to ensure that sharing of genomic PHI without authorization from the source individual is limited to reliable results of high value to the family member(s) in question and that the benefit to the family member(s) warrants the privacy risks to the source individual, including after death.

• Courts and other actors should clarify where necessary that under current law health care providers do not have a duty to disclose genetic test results to relatives of a patient if the patient does not consent to such disclosure.

• Some courts have held that a person has a cause of action and standing to bring a lawsuit against a clinician for negligent genetic testing of a family member, on the theory that the erroneous testing or reporting of the test results injured the relative bringing the lawsuit (e.g., by depriving the plantiff of accurate genetic information about their relative that may have also pertained to them).¹³⁷ Such lawsuits have the potential to significantly expand the duty and liability exposure of clinicians. If similar claims are recognized in the future, courts should clearly establish limits on any such duties to protect clinicians from overly-expansive liability exposure

A. analysis

Direct-to-consumer (DTC) genetic testing is a relatively new arrangement for genetic testing that is likely to become even more common in the future as regulatory restrictions are relaxed and more consumers become interested in obtaining their own genomic information. In DTC genetic testing, a company sends a test kit directly to a consumer who purchases testing; the consumer provides a buccal swab, spit, or blood sample in a vial that is mailed back to the DTC company; then the DTC company after analyzing the sample provides the genetic test results directly to the consumer, usually through a password-protected web-site. The patient's personal physician is usually not involved in the ordering or initial receipt of the genetic test results, although the DTC company may have a physician on staff who technically “orders” the genetic testing to comply with regulatory requirements.^{Reference Hogarth, Javitt, Melzer, Allyse, Ramos and Weissman138} However, many patients may bring their DTC genetic results to their personal physician for consultation and advice.

Providing genetic results through DTC companies raises many controversial policy and ethical concerns,^{Reference Hogarth, Javitt, Melzer, Hawkins and Ho139} and the FDA has been increasingly involved in regulation of DTC testing. Our focus here is on potential liability issues faced by a health care provider who is presented with or reviews a patient's DTC test results.¹⁴⁰

DTC genetic testing can present liability risks for providers. For example, if a patient seeks to share with their physician DTC results that show disease susceptibility or PGx information about a certain pharmaceutical (e.g., warfarin), and the physician knows of these results but fails to take them into account in providing medical advice, the physician may be faced with a medical malpractice claim, if that failure directly resulted in adverse effects on the patient. However, given that many DTC results, especially if self-reported, may not be accurate or reliable, any such physician duty to take such results into account must be tempered, and indeed, no cases have been reported in which physicians or other providers have been held liable for failing to consider DTC results in their care and treatment. If the DTC test results are potentially clinically significant, the best practice is for the patient's physician to order the test be validated in a CLIA-approved or CLIA-exempt laboratory, so that the results will be reliable. Nonetheless, a physician who ignores a patient's DTC results without getting the results validated, especially when brought to the provider's attention by the patient, may be at risk of liability.

Alternatively, a physician could be liable for over-relying on DTC genetic test results and recommending prophylactic surgery, pharmaceutical treatment, or some other unnecessary and possibly risky procedure without adequately validating the DTC genetic results.^{Reference Tandy-Connor, Schleit, Naylor and Hisama141} Recent studies have shown that DTC results are often erroneous,^{Reference Kuehn142} and even the DTC companies generally recommend that patients validate DTC results before relying on them for clinical decision-making. A physician who relies on DTC results without validating the results therefore also faces a liability risk. However, the physician may be put in a catch-22 situation if the patient's insurer will not pay for the clinical validation, and the patient cannot afford to pay for the validation test out-of-pocket. The physician then faces the dilemma of either relying on a potentially erroneous test result or alternatively ignoring a test result that may be valid. Either way, the patient may be harmed and the physician may be sued.

B. recommendations

Recommendations for Health Care Actors or Institutions:

• DTC genetic test results will usually include results from tests that are not clinically indicated. Such testing provides results that are more likely to be false positives and are not clinically useful compared to genetic testing in symptomatic or at-risk patients that is clinically indicated. Clinicians should be wary of relying on results from DTC tests which are not clinically indicated, and should require a confirmatory test before making clinical diagnoses or recommendations. Obtaining confirmatory testing could be problematic if, for example, the patient does not have insurance willing to pay for such confirma-tory testing.

• DTC genomics companies should ensure that they clearly explain the limitations of their tests and coverage, and in particular the potential for erroneous results and the potential for other untested variants to present risk to the patient.

• Providers delivering general care as well as more specialized care should be informed of the limitations and risks of DTC genetic testing, as well as any benefits in terms of costs and privacy, and be prepared to counsel their patients about such issues if and when patients inquire about such testing.

• Providers should not refuse to consider genetic information that a patient offers to share with their physician when there are reasonable grounds to believe that information may relate to a specific complaint or condition the patient is presenting. However, clinicians generally do not have the knowledge or resources to be the primary explainer of the DTC test results that all their patients receive, and should not be expected to review and understand all the relevant information that may be contained in those test results.

• A provider should not base clinical decisions solely on DTC test results. Rather, if the DTC test results indicate a potential problem or meaningful health indication, the provider should seek to have the test result replicated by a CLIA-compliant test laboratory.

Recommendations for Legal Development and Change:

• When a provider fails to take into account specific actionable genetic test results (from any reliable source) that the patient brings to the provider's attention relating to a health condition or risk that the provider is treating, and the failure to consider that genetic information causes harm to the patient, the provider may be held liable for medical malpractice. However, providers should not be assigned the legal duty to be aware of and to take action on all the information in patient DTC genetic test results, in particular genetic test results from tests that are not clinically indicated, except for the narrow exception described above.

• A provider who recommends clinical action based on the results of a DTC genetic test, without first verifying the validity of those DTC test results, may be held liable for harm that results to the patient from the reliance on inaccurate and unverified DTC test results.