Box 1 About This Column
Mark A. Rothstein serves as the section editor for Currents in Contemporary Ethics. Professor Rothstein is the Herbert F. Boehl Chair of Law and Medicine and the Director of the Institute for Bioethics, Health Policy and Law at the University of Louisville School of Medicine in Kentucky. (mark.rothstein@louisville.edu)
On the first page of my book, Medical Screening and the Employee Health Cost Crisis, I wrote:
Increasingly, “predictive screening” attempts to identify whether currently capable individuals are at risk of developing a medical impairment at some future time. This newer form of medical screening is greatly affected by technological advancements and is likely to be an increasingly important part of the employee selection process.Reference Rothstein1
The book was published in 1989, before the Human Genome Project, precision medicine, polygenic risk scores, or big data predictive analytics. It was also before enactment of the Americans with Disabilities Act (ADA)(1990),2 Health Insurance Portability and Accountability Act (HIPAA)(1996),3 Genetic Information Nondiscrimination Act (GINA) (2008),4 Affordable Care Act (ACA) (2010),5 and other laws to prevent health-based discrimination in employment, health insurance, or other aspects of daily life.
Although I had some sense of the emerging problem, I could not anticipate the scientific developments shaping health risk assessment in the twenty-first century or the complicated legal issues arising under legislation enacted before the latest scientific advances and constrained by political considerations. In this article, I revisit employer use of predictive health information and attempt to follow up on my earlier work in three ways: (1) updating the science of predictive health risk assessment; (2) analyzing the legislative inadequacies evidenced in recent case law; and (3) proposing comprehensive solutions to the lack of antidiscrimination protection for individuals with an increased risk of future impairment.
Growth of Predictive Health Information
Assessing the current health status and likely future condition of patients have long been mainstays of medical practice, but the science of predicting health has been advanced greatly by two transformative technologies. First, the 2001 completion of a draft sequence of the human genome facilitated the study of genomic variation, including research using single nucleotide variants, genome-wide association studies (GWAS), genotype-phenotype associations, and other methods. Genomics also led to numerous other “omics” technologies that explore the dynamic functions and interactions of gene products, the gene-mediated effects of environmental exposures, and genetic variation beyond human DNA. Proteomics, transcriptomics, metabolomics, pharmacogenomics, toxicogenomics, epigenomics, and microbiomics are among dozens of omics fields of investigation.Reference Hasin, Seldin, Lusis and Nice6 Individually and in combination, omics increasingly can aid fine-grained predictive assessments, diagnostics, monitoring, and eventually therapeutics.Reference Donovan, Currie and Delles7
One of the newest predictive genomic technologies involves computing “polygenic risk scores.” Unlike Mendelian or monogenic disorders, more common complex conditions are the result of many genes with smaller effect size acting over time in combination with environmental factors.Reference Sugrue and Desikan8 Based on data from GWAS and using statistical models derived from large samples, an individual’s risk of certain diseases or ill health in general can be calculated and expressed as a polygenic risk score. The scores may have clinical utility in identifying individuals needing a heightened level of medical surveillance for specific conditions,Reference Torkamani, Wineinger, Topol and Martin9 such as Alzheimer’s disease,Reference Logue10 coronary artery disease,Reference Inouye and Mosely11 and obesity.Reference Torkamani and Topol12 They also have unmistakable significance for many entities, including employers, with a financial interest in an individual’s future health.
A second development causing a substantial growth in predictive health risk assessment is the application of multi-factorial big data analytics to health care, especially precision medicine. As exemplified by the All of Us research program directed by the National Institutes of Health, a million or more volunteers in the United States will have whole genome sequencing, and their sequence data will be combined with health data from electronic health records (EHRs), biometric measures, biological and environmental monitoring, wearable devices, and mobile health apps.13 In theory, new associations and directions for research will emerge from this application of big data. Other types of health information also could be relevant to precision medicine research and practice, including health histories and vital statistics of family members, military service records, employment records, financial and consumer information, educational records, travel information and geo-location data, social media postings, and various governmental records.Reference Rothstein14 Although the significance of any single data point may not be apparent, the essence of big data analytics is developing algorithms to decipher unexpected associations or correlations in disparate datasets.Reference Rothstein, Areheart and Roberts15 The possible application of big data health analytics also goes beyond research to the clinic.Reference Cohen, Topol and Hoffman16
Privacy and Security Concerns
The compilation of huge and diverse datasets of highly sensitive information raises significant concerns about privacy and security, regardless of whether the information leads to employment discrimination or other tangible harms.Reference Rothstein17 First, security issues may arise from lost or stolen mobile devices, hacking, or unauthorized uses of information by authorized users. Second, for individuals to obtain health benefits from health research using precision medicine, research records or other sources of nonclinical data would need to be integrated with individuals’ EHRs maintained by health care providers. Once health information is under the control of a healthcare provider (a HIPAA “covered entity”), it is subject to the HIPAA Privacy Rule.18 Contrary to common misconceptions, this does not mean that the information is well protected, because numerous exceptions to the Privacy Rule permit uses and disclosures of individually-identifiable health information without the knowledge or authorization of the individual.Reference Rothstein, Rothstein and Tovino19 Third, individuals may be compelled to sign a HIPAA-compliant authorization to disclose extensive health information as a condition of applying for employment, insurance, or other government or commercial uses. There are at least 25 million such compelled disclosures each year in the United States.Reference Rothstein and Talbott20
Employer Use of Predictive Health Information
There are two main categories of predictive medical screening of workers: (1) screening to determine whether a prospective or current employee is predisposed to injury or illness from workplace exposures, such as toxic substances;Reference Rothstein21 and (2) screening to determine whether a prospective or current employee is predisposed to ordinary diseases of life, such as cancer or cardiovascular disease.22 In addition to legitimate concerns about the health and well-being of their employees, many employers have substantial financial motivation for not hiring or retaining employees who are likely to become ill in the future. These financial interests include decreased productivity, increased absenteeism, increased presenteeism,Reference Hemp23 increased use of short-term and long-term disability benefits, increased turnover (requiring recruitment and retraining costs), increased healthcare costs,Reference Roberts24 increased workers’ compensation costs,25 and possible liability costs for harms to customers or members of the public caused by ill employees.
Prospective and current employees who are well, but who have an increased risk of future illness, may be difficult to identify. Nevertheless, employers can obtain predictive health information about employees through various lawful means, including voluntary disclosures by employees via application forms, interviews, social media, and casual remarks to supervisors; disclosures by references and coworkers; results of employerconducted medical examinations and health records generated or obtained at post-offer and periodic medical assessments; information disclosed in employee wellness programs; data generated by wearable devices and smartphone applications; Family and Medical Leave Act26 and other leave applications; disability insurance claims; workers’ compensation claims; and health insurance claims. Some employers also might be tempted to purchase employee health information from data brokers, reidentify deidentified employee health information, or engage in other dubious practices.Reference Hoffman27
Once in possession of predictive health information employers could take actions adversely affecting prospective or current employees. Besides refusing to hire or terminating an individual, an employer might disclose an employee’s health status to supervisors, managers, or other employees; reassign the employee; deny an employee’s requested promotion, transfer, or change in working conditions or hours; increase monitoring of an employee; or pressure an employee to resign or retire. The lawfulness of any of these actions depends, in the first instance, on whether the individual is covered by an antidiscrimination law. As discussed in the following section, current laws do not apply to employer use of predictive health information, except for genetic information.
Future Risk of Illness under the ADA and GINA
Americans with Disabilities Act (ADA)
The ADA makes it unlawful for an employer to discriminate “against a qualified individual with a disability because of the disability of such individual in regard to job application procedures, the hiring, advancement, or discharge of employees, employee compensation, job training, and other terms, conditions, and privileges of employment.”28 This broad language was adopted from the Rehabilitation Act of 1973,29 as was the ADA’s threepart definition of disability. Under the ADA, as amended, a person with a disability is defined as an individual: “(1) [with] a physical or mental impairment that substantially limits one or more of the major life activities of the individual; (2) [with] a record of such an impairment; or (3) [who is] regarded as having such an impairment.”30 Thus, the ADA prohibits discrimination because of an individual’s current impairment, past impairment, or “actual or perceived physical or mental impairment whether or not the impairment limits or is perceived to limit a major life activity.”Reference Befort31 Despite the ADA’s intent to provide broad coverage, the definition of disability does not mention a predisposition to or increased risk of a future impairment, nor is there any evidence the issue was considered by Congress.Reference Davis32
Recent judicial decisions under the ADA have highlighted the lack of coverage for individuals who allege discrimination in employment because of predictive health information. In Shell v. Burlington Northern Santa Fe Railroad Co.,33 the plaintiff successfully worked for 33 years in physically demanding positions at a railyard. When the ownership of the railyard changed hands, the plaintiff applied for employment with the successor employer as an intermodal equipment operator, another physically demanding position. At the required preplacement medical examination,Reference Rothstein, Roberts and Guidotti34 the plaintiff was measured as being 5’10” tall and weighing 331 pounds, giving him a body-mass index (BMI) of 47.5. The defendant railroad had a policy of refusing to hire individuals with a BMI over 40. Instead of rejecting him simply because of his obesity, however, the railroad said it considered him to be at a substantially higher risk of developing conditions such as sleep apnea, diabetes, and heart disease. Furthermore, it claimed that these conditions can have unpredictable onset, possibly resulting in sudden incapacitation, and thereby presenting serious threats to anyone at the railroad yard.35
The United States District Court for the Northern District of Illinois denied the defendant’s motion for summary judgment, and the defendant appealed to the Seventh Circuit. Under the circuit’s controlling precedent obesity alone is not a physical impairment under the ADA unless there is evidence that the individual’s obesity is caused by an underlying physiological condition or disorder.36 Consequently, the plaintiff argued that his “disability” consisted of the various conditions the defendant feared he could develop -- sleep apnea, diabetes, and heart disease. The Seventh Circuit summarized the key issue as “whether the ADA’s ‘regarded as’ prong covers a situation where an employer views an applicant as at risk for developing a qualifying impairment in the future.”37
Using a linguistic interpretation, the court stated that the present tense used in the definition of “regarded as” — “having an impairment” — precludes its applicability to future impairments. “If the impairment does not yet exist, it can be neither actual nor perceived.”38 The court also cited recent decisions of the Eighth,39 Ninth,40 and TenthReference Jordan and mutation41 Circuits for similar holdings. The court rejected the two arguments in the Equal Employment Opportunity Commission’s (EEOC’s) amicus curiae brief. First, the EEOC cited to its Compliance Manual, which provides that discrimination against an individual because of genetic susceptibility to cancer is covered under the “regarded as” prong of the definition of disability.42 The court held that this interpretation conflicts with the EEOC’s Interpretive Guidance, which provides that the definition of disability does not include “characteristic predisposition to illness or disease.”43 The court said that because the example of genetic susceptibility to cancer is “unmoored from the ADA’s text and in tension with other EEOC interpretive guidance, [ ] it lacks the power to persuade us away from the statute’s unambiguous text.”44 Construing genetic predisposition as an exception to the general Interpretive Guidance would eliminate the seeming inconsistency, but not the court’s concern about fidelity to the statutory language.
The EEOC’s second argument relied on the ADA’s overall purpose to combat “society’s accumulated myths and fears about disability and disease.”45 The court acknowledged that the ADA Amendments Act of 2008 intended to have the definition of disability construed in favor of broad coverage, but the court said it could not adopt an interpretation in conflict with the statute’s plain language.
Although no other circuit court decision has held that future health risk is covered under the ADA, there is a minority position. In EEOC v. Rockwell International Corp.,46 the EEOC brought an action to challenge an employer’s refusal to hire individuals who failed preplacement nerve conduction tests. The Seventh Circuit held that the individuals were not covered under the “regarded as” provision of the ADA. Judge Diane Wood dissented.
And it is not at all clear to me that as a matter of law the ADA permits an employer to refuse to hire a person who is fully qualified to perform certain work, simply because that individual might at some unspecified time in the future develop a physical or other disability that would render her unable at that later date to meet the employer’s reasonable expectations. This smacks of exactly the kind of speculation and stereotyping that the statute was designed to combat.47
Judge Wood is unquestionably correct about the legislative intent of the ADA, but it does not appear to be a promising strategy to wait for a favorable resolution by the Supreme Court or a change in sentiment by the circuit courts on the issue of ADA coverage of future risks.
Genetic Information Nondiscrimination Act (GINA)
Whereas the ADA has been construed as not applying to future impairments, GINA only applies to future impairments. As discussed below, inconsistencies and gaps in coverage often characterize the nation’s employment discrimination laws.Reference Rothstein48 Title II of GINA prohibits discrimination in employment based on genetic information. “The term ‘genetic information’ means, with respect to any individual, information about – (i) such individual’s genetic tests, (ii) the genetic tests of family members of such individual, and (iii) the manifestation of a disease or disorder in family members of such individual.”49 Notably, genetic information does not include information about the individual’s own manifestation of a disease or disorder and therefore discrimination based on a genetic condition that has “manifested” is not prohibited by GINA.Reference Prince and Berkman50 The reason for this seemingly illogical exclusion is that Title I of GINA prohibits discrimination in health insurance, and the effect of prohibiting exclusions based on manifested conditions would be to prohibit health insurers from denying coverage based on an individual’s preexisting condition. Congress was not prepared to enact such a ban in 2008 when it enacted GINA, but it did so in 2010 when it enacted the ACA.51
The “mirror-image” coverage provisions of the ADA and GINA would not be a problem if the composite picture was comprehensive coverage of genetic-related conditions, but that is not the case. GINA applies only to discrimination based on genetic information about a condition that has not manifested. By contrast, the ADA applies only to symptomatic individuals who have an impairment that constitutes a substantial limitation of a major life activity. Two gaps in coverage remain. First, individuals who have “manifested” a genetic-related condition that does not currently constitute a substantial limitation of a major life activity are not covered by either the ADA or GINA. This category of uncovered conditions is expanding because new technologies can identify biological changes earlier in the disease process by using, for example, biomarkers and endophenotypes indicative of subclinical expressions of genetic conditions.52 Second, asymptomatic individuals who are at an increased risk of a non-genetic related impairment also are not covered by either the ADA or GINA.
The Illogic of the ADA and GINA
The employment discrimination provisions of the ADA were enacted to eliminate and redress pervasive discrimination against individuals with substantially limiting impairments. They were not enacted to prohibit any adverse treatment based on an individual’s health status;Reference Roberts and Weeks53 more comprehensive legislation banning all healthbased discrimination undoubtedly would not have been enacted.Reference Rothstein54 From a cost perspective, it is illogical to prohibit discrimination against individuals who require ongoing reasonable accommodation while permitting discrimination against individuals who have a temporary health condition that will last only a few days. Yet, from a civil rights perspective, it is reasonable to target the most entrenched, insidious, and devastating forms of discrimination against individuals with substantially limiting impairments. Thus, employers are required to hire individuals with disabilities, including bearing the costs of providing reasonable accommodations, unless doing so would constitute an “undue hardship.”55
Many parts of the employment title of GINA are seemingly illogical, starting with the law’s intent to allay individuals’ fears of genetic discrimination, despite the lack of confirmed instances of such discrimination.Reference Roberts56 GINA prohibits discrimination based on information about future impairments, precisely the kind of discrimination that the ADA has been held not to cover for non-genetic impairments. Giving greater legal protection to individuals discriminated against because of genetic information is a form of “genetic exceptionalism.”Reference Murray, Roth stein and Rothstein57 Although it is appropriate to have genetic-specific policies in limited circumstances,Reference Evans, Burke, Khoury and Rothstein58 there is no justification for maintaining or adopting more general policies based on genetic exceptionalism. The increased availability of detailed genomic information and the growing use of predictive risk assessment using non-genetic information support the adoption of policies based on context.Reference Garrison59
GINA attempts to protect genetic privacy as well as to prohibit genetic discrimination.Reference Clayton60 Section 202 of GINA generally prohibits employers from requesting, requiring, or purchasing genetic information about an employee or a family member of an employee, except in limited circumstances. One circumstance, detailed in section 202(b)(5), permits an employer to acquire information for use in genetic monitoring of the biological effects of toxic substances in the workplace, but only if the employer provides written notice of the genetic monitoring to the employee; the employee provides written authorization; the employee is informed of individual monitoring results; and the employer receives only aggregate results that do not disclose the identity of employees.61 Employers, however, are not permitted to provide optional genetic testing at the preplacement stage, even with the same protections in place, so that prospective employees can learn whether they have a heightened risk if they are exposed to certain toxic substances in the workplace.Reference Rothstein62 Thus, GINA deprives prospective employees of the ability to obtain and act upon potentially valuable toxicogenomic information.Reference Rothstein63
Toward Reasonable, Consistent Public Policy
Many individuals with disabilities who are currently covered by the ADA have illnesses likely to worsen over time, and therefore they will be unable to work in the future with or without reasonable accommodation. The ADA protects these individuals, notwithstanding their prognosis. In addition, an EEOC regulation provides that “[a]n impairment that is episodic or in remission is a disability if it would substantially limit a major life activity when active.”64 Extending coverage to individuals who have a future health risk would be consistent with these other interpretations of the ADA.
Individuals with an increased health risk, currently not protected by the ADA, would be covered by the ADA if they became symptomatic with an impairment that constitutes a substantial limitation of a major life activity.65 It makes little sense that such individuals can be legally denied employment at a time when they are able to work, but when they become ill and are finally covered by the ADA, they might require accommodations or they might be unable to work at all.
Public policy should facilitate the employment of all individuals with the present ability to perform specific job-related tasks with or without reasonable accommodation and regardless of their future health risks. This position would be consistent with other federal antidiscrimination laws. For example, in 1978, the Pregnancy Discrimination Act (PDA) amended Title VII of the Civil Rights Act of 1964 to provide that pregnant women “shall be treated the same for all employment-related purposes” as other persons “not so affected but similar in their ability or inability to work.”66 Even though a pregnant woman would likely take maternity leave at some time in the future she would be protected as long as she was currently able to perform job-related functions safely and efficiently.Reference Hébert67 In addition, the Age Discrimination in Employment Act (ADEA), as amended, prohibits discrimination in employment against individuals at least 40 years old,68 regardless of their age or future health risks so long as they are currently able to perform job-related functions safely and efficiently.
By enacting the PDA, ADEA, ADA, and GINA, Congress recognized it was imposing additional costs on covered employers (and other entities and the public to the extent the costs are externalized), but believed it was necessary to provide employment opportunities to classes of individuals that have been subject to exclusionary policies and who have much to contribute to the economy and the nation. Extending coverage to individuals who are predicted to be at an increased risk of health problems would be consistent with these policies.
Moving Forward
Professor Sharona Hoffman has argued persuasively that Congress should broaden the ADA’s “regarded as” provision to include individuals who “are perceived as likely to develop physical or mental impairments in the future.”69 Although this would be an important step, more comprehensive measures also should be enacted. According to the EEOC’s regulations, there is no duty to provide reasonable accommodation to an individual whose coverage is under the “regarded as” prong of the definition of an individual with a disability.70 Nevertheless, many individuals at heightened risk of future illness would benefit from reasonable accommodations to reduce their risks, such as using respirators or other personal protective equipment when working with toxic substances, reducing certain physically demanding activities, or limiting exposure times in extreme environments (e.g., heat, cold, high altitude).Reference Rothstein and Rothstein71
If the ADA’s coverage is extended to individuals with future health risks, a question arises as to whether all future health risks should be covered regardless of their likely time of onset. For example, suppose an applicant seeks employment today, but there is medical evidence that the individual will only be able to work for a few months before becoming seriously ill. If the individual is seeking employment for a job with a long training period, it would be unreasonable to require the employer to hire and train the individual, only to have that person resign for health reasons before being able to perform the job.
In the ADA Amendments Act of 2008, Congress defined a “transitory impairment,” excluded from ADA coverage under the “regarded as” prong, as “an impairment with an actual or expected duration of 6 months or less.”72 Applying this standard to predictive health risks, an individual should be protected under the ADA’s “regarded as” prong if the individual is regarded as having a future health risk that would not manifest for at least six months. This extends protection of the ADA with out creating unreasonable burdens on employers.73 Another way of limiting employer access to information about an individual’s future health risks would be to amend the ADA to provide that all medical examinations and inquiries after a conditional offer of employment must be job-related and consistent with business necessity,74 the same standard applicable to current employees.75
The other main federal statute directly applicable to future health risk, GINA, is inadequate and outdated. It should be restructured to broaden and update the definition of genetic information;Reference Rothstein and Schlaberg76 extend coverage beyond health insurance and employment to other important areas, such as life, disability, and long-term care insurance;77 and permit the beneficial uses of genetic information to increase protection for sensitive employees and allow them to choose whether to work in personally hazardous environments.78 Other laws beyond the scope of this article also should be amended to advance important public policies.79
Recent cases holding that ADA protections do not apply to individuals at increased risk of substantially limiting impairments illustrate the incongruity between, on the one hand, the ADA and GINA, and on the other hand, modern technologies that generate predictive health assessments. Genomics and other omics analyses, polygenic risk scoring, and big data health algorithms are expanding the number of individuals who could be subject to employment discrimination, but who have no protection under federal law. It is imperative to amend the ADA and GINA to advance the public policy goals of nondiscrimination in an age of rapid technological change.
Conclusion
Recent cases holding that ADA protections do not apply to individuals at increased risk of substantially limiting impairments illustrate the incongruity between, on the one hand, the ADA and GINA, and on the other hand, modern technologies that generate predictive health assessments. Genomics and other omics analyses, polygenic risk scoring, and big data health algorithms are expanding the number of individuals who could be subject to employment discrimination, but who have no protection under federal law. It is imperative to amend the ADA and GINA to advance the public policy goals of nondiscrimination in an age of rapid technological change.
Acknowledgement
John Tyler Stocking, J.D. 2020, Louis D. Brandeis School of Law, University of Louisville, provided excellent research assistance.
