Good morning, Mr. Chairman and Members of the Subcommittee. Myname is Tom Miller. I am director of health policy studies at theCato Institute. It is a pleasure to appear before you today toexamine privacy concerns raised by the possible collection and useof genetic information by employers and insurers.
I say "possible" because there is little, if any, evidence thathealth insurers are using or likely to use presymptomatic geneticinformation in their medical underwriting. Evidence that employerstry to obtain, let alone use, such information generally is limitedto isolated anecdotes. One survey of human resources managersrevealed that limited evidence of genetic testing actuallyreflected misunderstanding of what truly is genetic information, asopposed to routine blood tests or testing for the presence of adisease (rather than a genetic susceptibility to it).
However, this topic usually is built on assumptions about thefuture and what some observers believe private insurers andemployers might do in the event that genetic testing and geneticinformation eventually became more accurate and precise inrevealing an individual's prospects for future disease, incapacity,and unusually high health care claims costs.
As Yogi Berra once said, or probably should have if he did not,"Predicting the future is hard, because it hasn't happened yet."Nevertheless, let's proceed by beginning with what we do know abouthow private insurance markets and labor markets operate.
It's reasonable to assume that more accurate, comprehensive, andinexpensive genetic testing will arrive one day; the question ismore one of the pace at which this will take place. When combinedwith reliable evidence from epidemiology about the probability andmagnitude of various maladies, the continuing genetic revolutionpromises to improve dramatically our ability to detect disease atan early stage, and treat it more effectively or even prevent it.Better, and earlier, knowledge about genetic predisposition toillness might help individuals take preventive measures to reducethe consequences of disease or even eliminate its onset. Enhanceduse of more predictive genetic information may assist individualsin making lifestyle plans and choices. It also holds great promisein fine-tuning health care treatment, such as through more narrowlytargeted "designer" drugs and gene therapy interventions.
But the offsetting concern encompassed in the catchall desire toprotect "genetic privacy" is that one's personal geneticinformation might be disclosed to others without one's consent andthen used to one's personal detriment.
Without question, any information that can be used may also beused badly. But a host of policy complications and administrativecomplexities arise if one attempts to craft a unique brand of legalprotection against the disclosure of personally identifiablegenetic information in the name of "genetic privacy."
First of all, any possible constitutional protections forgenetic privacy would be limited to apply only against governmentaction. The Fourth Amendment right of privacy (the right to be freefrom "unreasonable searches and seizures") and the more narrowright to "informational privacy" recognized in Whalen v. Roe do notapply to the private sector.
Moreover, the more accurate way to frame the issue of whetherstatutory law should prohibit or limit disclosure and use ofgenetic information by private parties is in terms of genetic"nondiscrimination" rather than genetic privacy. It also wouldnarrow the legislative and regulatory focus to consider whether andhow to protect consumers from harmful use of genetic information,rather than whether and how to maximize consumers' legal controlover that information per se.
But even erecting legal barriers against discrimination based ongenetic information would strain the limits of geneticexceptionalism, defy precise definition, pose serious threats tothe functioning of private insurance and labor markets, andoverlook more effective alternative remedies.
There is no clear line that separates genetic data from otherkinds of personal health information. Attempts to define "genetic"diseases must sort through a fuzzy mix of compulsive, addictive,and patterned behaviors, as well as characteristics likeintelligence, aggressiveness, and obesity that have both geneticand environmental roots.
Moreover, the sources of legally protected genetic informationmight be obtained from many current and commonly accepted medicalpractices that do not involve explicit "tests" of one's geneticmaterial. For example, personal medical histories, family medicalhistories, routine physical exams, and blood pressure tests allmight convey predictive health information that could be linked togenetic factors.
In efforts to protect against genetic discrimination, how shouldpolicymakers deal with the "predictability" of medical conditionsthat are influenced by multiple genes that work in conjunction withenvironmental factors? Or with genetic predispositions that onlyincrease the risk, rather than establish the certainty, ofdeveloping a disease? Or with genetic conditions that increaseone's probability of experiencing both adverse and beneficialoutcomes? Should legal prohibitions apply only to uses of geneticinformation that disadvantage the protected party?
Overly broad legal claims to genetic privacy and geneticnondiscrimination often may serve as a subterfuge for morefundamental opposition to various kinds of private health insuranceunderwriting based on individual health risk or to voluntarydisclosure of sensitive health information to one's employer. AsRichard Epstein observes, modern uses of privacy may be used tooverride freedom of contracts and even "act as a handmaiden tofraud." One's right to privacy should not include the right tomisrepresent oneself to the rest of the world, particularly in thecase of making statements to one's health insurer or employer thatone knows to be false, material to the listener, and relied on tothe listener's detriment. For example, individual insuranceconsumers who know or suspect their genetic risk factors should notbe able to engage in unilateral strategic behavior to choose thetiming, scope, and/or duration of the insurance coverage theypurchase.
A broad prohibition on any disclosure of genetic informationwould prevent good health risks from obtaining positive geneticinformation on their behalf and then voluntarily disclosing it topotential health insurers. Yet efforts to separate the treatment ofpredictive genetic information into prohibitive negative categoriesand permitted positive categories would defy administrativeconsistency and predictability. They would essentially lead to aregulatory regime of "Can't ask, may tell, may lie" in which truthis discouraged and lies are protected, promoted and necessary.
At this point of the discussion of genetic privacy and geneticdiscrimination, many private health insurance industryrepresentatives begin to retreat behind the current wall of federaland state legislation that already prohibits or limits most formsof genetic discrimination, particularly in the employer-sponsoredgroup insurance market. For example, the Health InsurancePortability and Accountability Act of 1996 (HIPAA) prohibitsdiscrimination against individual workers who are members of anemployer group plan - either on the basis of their current healthstatus or on the basis of their predisposition to a particulardisease based on genetic information. They cannot be denied accessto group health plan benefits or be required to pay higher premiumsdue to their individual health status. And, after they havesatisfied HIPAA's preexisting condition limitations once, they thenmay move to another employer's group health plan without facing newlimitations on coverage due to additional information about theirhealth status. Genetic information also is treated as protectedpersonal health information under HIPAA's health privacyregulations.
However, HIPAA does not govern the use of genetic information inthe individual health insurance market, which generally has moreoperating freedom in the underwriting process under the insurancerules of most states. Health insurers are more likely to opposeexplicit prohibitions on use of genetic information in theindividual market and to point out that tighter restrictions onunderwriting and pricing in such a smaller and often transitorymarket are more likely to reduce rather expand available coveragethere.
But even medical underwriting in the individual market tends tofocus on medical costs that are likely to occur within the firstfew years after a policy is purchased, and more expensive geneticscreening tests to predict much longer range health risks would notbe worthwhile to insurers, even if legally permitted. Yet insuranceindustry representatives generally will caution against outrightbans on genetic testing in the individual market, essentiallyarguing, "We don't use it, we don't plan to use it, but don'tprohibit us from using what we don't need to use." One might ask,why?
For one reason, first consider that, apart from HIPAA's legalprohibition on medical underwriting of individuals within employergroup health plans, the common practice of insurers is to rely onexperience rating for all but the smallest employer-sponsoredgroups. The expense and administrative burden of more intensiveunderwriting simply outweighs the practical value of whateverlimited and imprecise information about health risk that an insurermight obtain. Insurers also have no incentive to turn awayapparently healthy customers, or even somewhat less healthy ones(as long as rating flexibility permits some adjustments), basedmerely on possible long-term genetic risks that remain hard tomeasure.
However, it does remain possible that, as genetic testing andgenetic information becomes more predictive much further down theroad, and as private insurance coverage becomes more customized(such as through defined contribution plans and consumer-drivenhealth care options), more private insurers eventually will facegreater pressure either to sort high risks into high-premium andrestricted coverage risk classifications or to increase premiumsand restrict coverage further across the board. If individualcustomers become better armed with personal genetic information andcan engage in behavior that increases the previously predictablerange of insured claims, insurers will want to use, and they willneed to use, such information to correlate more accurately thosegenetic characteristics with future costs and premiums. Otherwise,low-risk customers increasingly will exit voluntary privateinsurance markets and the overall supply of coverage willcontract.
A good bit of this speculation about the future and theso-called "end of insurance" reflects the likely time lag we willexperience in making an uneven transition from the first round ofgenetic information and biomedical research that is better able todetect and diagnose genetic problems than later rounds ofscientific discovery will be able to cure or mitigate. As RobertaBerry has observed, we should more carefully weigh the consequencesof regulatory interventions that essentially are aimed at assuringthat genetically high-risk individuals are able to obtain bargainpremiums from and transfer their losses to a private pool of lowerrisk insurance customers or their employers.
Prohibiting use of predictive genetic information also rather"indiscriminately" provides a hidden subsidy to any individual whomight be discriminated against for genetic risk reasons, regardlessof their financial circumstances or particular needs. Yet wegenerally do not provide similar subsidies for other forms of "badluck" in life's genetic lottery, such as less desirable levels oftraits like intelligence, aggressiveness, or physical appearancethat have at least some genetic roots. It would also seem odd ifregulatory treatment were to become relatively more tolerant ofadverse treatment of symptomatic individuals than for asymptomaticindividuals who merely possess genetic risk factors.
Broad federal regulatory prohibition against geneticdiscrimination in health insurance may shut off valuable flows ofinformation and suppress financial incentives that could encourageindividuals to make better decisions about the insurance coveragethey buy, the investments they make in other health-promotingactivities, and the behaviors in which they engage.
Not very far below the surface of claims that geneticdiscrimination in health insurance should be prohibited is thelarger issue of whether risk classification based on health statusis permissible within at least some private sector portions of ourmixed system of voluntary private insurance and publicly financedhealth programs. Employer group health plans straddle the fence bymoving risk classification to the firm, as opposed to individual,level. But private insurers still need to be able to predict theapproximate level of health claims they are likely to pay if theyare to set their premiums high enough to stay in business, yetassure lower risk customers that the coverage offered to them isworth the price charged.
As enhanced availability and use of more predictive geneticinformation strains the fault lines of our private health insurancesystem, we first should consider other market-based, private-sectormechanisms, or at least more explicit and overt public subsidies,as alternatives to expanded regulatory cross subsidies. Publicpolicy might encourage the development of better voluntary poolingmechanisms outside of the employer-employee relationship. It couldfacilitate long-term health insurance contracts that offerguaranteed renewability options or other time-consistent insuranceincentives such as second-tier savings components that would remainsubject to illness-state-contingent "severance payments" for earlydeparture from a particular insurance pool. Or consumers mightconsider purchasing "genetic test insurance" to insure themselvesagainst any as-yet unknown risks before they took particulargenetic tests. More conventional approaches might include expansionof high-risk health insurance pools and greater incentives forcharitable contributions to nonprofit intermediaries that organizeand deliver safety net health care.
I will touch more briefly on the issue of genetic informationand genetic discrimination in employment settings (although theabove health insurance issues also would come into play foremployers that sponsor self-insured health plans). In general, Iconcur with Richard Epstein's view that employers should be able toseek whatever information they might find relevant to theiremployee's job performance. Prospective and current employees, ofcourse, may refuse to supply information that is requested. Onemight imagine some extreme circumstances that could necessitate theintervention of public authorities to do more than enforceemployment contracts and prevent fraudulent misrepresentations, butthat should be the narrow exception rather than the broad rule.Irrational prejudice and discrimination in labor markets mayoverwhelm economic logic on some occasions, but market forces alsowill impose significant costs on employers who persistently excludeproductive workers who might happen to possess genetic riskfactors. If employers are faced with prohibitions against usingvaluable genetic information, they are most likely to resort toother legal and second-best (or third-best) substitutes forforbidden information. Employers might restructure compensationpackages to adjust for higher health benefits costs, substitutepart-time labor and independent contractors for full-time jobs,rely more on capital investments or offshore operations, or hireyounger and presumably healthier workers.
If regulatory policy insists that employers should remain blindto the known costs of employing certain types of individuals, thoseemployers will resort to coping mechanisms to cut their losses butleave us all a bit poorer and less efficient.
Rather than rely on greater regulation of information flowssimply because they are labeled "genetic," we should restore andrenew our commitment to competitive markets, private propertyrights, and private contracts.