AIDS and the Moral Economy of Insurance

When a blood test to detect AIDS antibodies was first announced in 1985, the ensuing controversy over the use of the tests by insurance companies seemed to take a familiar shape. On one side were civil rights advocates claiming discrimination if the insurers were permitted to use the tests to screen applicants for life and health insurance. On the other side was an industry insisting on its right to be free of government regulation. But despite the seemingly familiar pattern, the conflict over AIDS testing really concerned a novel problem with repercussions for many people who do not see themselves as having any stake in the issue.

The AIDS antibody test is only one of a growing number of predictive diagnostic tests. These tests tell whether a person is highly susceptible to a disease or, very rarely, whether someone is certain to get sick. Other predictive tests include the recently identified genes and genetic markers for Alzheimer's disease, manic-depressive disorder, multiple sclerosis, muscular dystrophy, cystic fibrosis, and some forms of cancer, diabetes, and heart disease. Public health research has also identified numerous risk factors for chronic disease, such as high cholesterol, high blood pressure, obesity, and smoking. These risk factors are even easier than genetic abnormalities to detect in individuals.

Predictive diagnostic tests have created a new medical limbo between health and illness. Now it is possible to be labelled "at risk" without being ill or ever developing the disease in question. In fact, many predictive tests in clinical medicine are designed to be overinclusive to minimize the number of cases of illness missed by physicians, allowing them to treat or at least detect medical problems as early as possible. But when used by private insurers, predictive tests are turned to a contrary purpose: denying insurance coverage, or charging more for it, and thereby causing fewer people to receive the health care they need. The aim of preventive medicine is to extend the benefits of modem science. Yet, ironically, the more predictive tests there are available, and the more broadly risk categories are drawn by cautious clinicians and epidemiologists, the fewer people private insurance will serve.

In the United States, private insurance is the primary means that Americans use to pay for health care and to provide for needs that are too big to meet through normal work income and savings. It is also the principal vehicle for fulfilling family financial obligations. Through health insurance, life insurance, disability insurance, pensions, and their related dependent benefits, Americans create their own networks of social aid within the larger society of strangers. Social insurance programs, such as Social Security pensions, disability insurance, and Medicaid, are designed only to be safety nets, not primary sources of support.

The private insurance industry, operating under rules set by law and public policy, controls access to the vital first line of defense against financial catastrophe. By determining who gets privately insured, for what misfortunes, and at what price, the insurance industry also affects the balance of responsibility -- and costs -- between the private sector and government. The more the industry uses predictive testing to limit access to private insurance, the more people and troubles fall to public programs. Furthermore, just as insurers want to limit their payouts, so employers want to limit their insurance premiums. They may increasingly adopt predictive tests to screen job applicants.

The use of predictive testing thus raises major questions about the future of access to good jobs, health care, and financial security in America. The United States already has a patchwork system of health insurance that omits coverage of nearly one of every six citizens. If used to their full potential, predictive tests may relegate even more Americans to the ranks of the excluded. But the current framework of insurance, which concentrates costs on people with high health risks, is not the only possible design. We do have alternatives for creating a system that protects those at high risk -- and all of us -- from financial devastation and exclusion from health care.

Why Insurers Want to Test

The battle over AIDS testing initially seemed to go in favor of the opponents. Within a month after scientists announced the test, gay advocates in California had obtained state legislation to prevent insurance companies from using it; in the next two years Wisconsin, New York, Florida, Massachusetts, and the District of Columbia adopted similar restrictions. The commercial insurance industry, however, mounted an all-out effort to repeal the regulations, and by the end of 1989 only California's ban on testing health insurance applicants remained in place.

The money immediately at stake for insurers was not the chief reason for their concern. Insurance companies mobilized their political influence because they feared losing their ability to screen applicants and set rates according to the health risks that the applicants appear to represent. The companies consider control of those decisions crucial to their competitive strategies, even their financial survival.

In the jargon of the insurance business, the process of selecting risks is called "underwriting." Underwriting involves determining whether an applicant's likely loss experience is similar to that assumed by an insurance company in setting its standard rates. If the applicant represents a greater risk, the company may offer a "substandard" (that is, higher) rate or deny coverage altogether. By accurately predicting losses and setting premiums accordingly, insurers seek to maintain their solvency and profitability. They also use underwriting to design policies with specialized features for carefully selected groups of people. Indeed, life insurers compete not so much on price or service as by marketing special policies to target groups.

In health, disability, and life insurance, insurers use medical information and other factors such as age, gender, and occupation to determine coverage and rates. Besides asking applicants (and sometimes their physicians) to fill out questionnaires about their health, they may also require a physical exam, including laboratory and clinical tests, such as urinalyses or electrocardiograms. Tests are used in part to find out whether applicants are concealing important facts. For example, insurers sometimes screen blood for prescription drugs to determine whether applicants are being treated for a disease they did not disclose.

This information is not kept privately by the insurance firm checking out an applicant. The industry maintains a central laboratory, the Home Office Reference Laboratory (HORL), to perform most medical tests. The HORL shares its results with a central data bank, the Medical Information Bureau, which is a membership organization of about 700 companies. An applicant for health, life, or disability insurance to any of these companies must sign a consent form allowing the company to report its findings to the bureau. Once an applicant has filled out a questionnaire or had blood sent to the HORL, the results are available to other insurers. So despite the appearance of a highly competitive industry, the prospective purchaser of an individual policy effectively faces only one supplier.

BEFORE THE ADVENT OF the AIDS antibody test, insurers were already using supposed indicators of homosexuality as underwriting criteria. Single men between 25 and 45, particularly if employed in occupations deemed stereotypically gay, were being denied individual policies for life and health insurance. In one sense, the blood test was a blessing to both insurers and gays. It enabled insurers to rely on objective medical evidence and took the focus off sexual orientation. But gays still perceived a threat because the blood test was likely to be imposed selectively on the basis of presumed sexual orientation. Moreover, those who tested positive were not certain to get AIDS, for the test does not disclose who has the disease. It discloses only the presence of antibodies in the blood to the human immunodeficiency virus (FHV), which causes AIDS. Although scientists now think at least 75 percent of those who test positive will probably develop AIDS, it is uncertain exactly who will contract the disease.

ONCE GAYS DEFINED THE issue as discrimination, the industry's strategy was clear. This was not the first time that insurers had to defend their use of classifications that public sensitivities no longer readily accepted. In the late nineteenth century, several states banned the use of race in setting insurance rates. More recently, the use of gender in pension, disability, and automobile insurance has faced attack. The industry's case for AIDS testing reflected its arguments over the last century in defense of its underwriting practices.

HIV testing, the industry maintains, has nothing to do with attitudes towards homosexuals. For insurers, discrimination is the essence of the business, not a dirty word. It means differentiating among policy holders according to their risk of incurring the loss for which the policy will pay out. HIV tests are just one more means of determining risk status, and they are no more discriminatory than blood pressure readings. According to the industry, it would be unfair not to use the HIV tests.

Insurers distinguish between fair and unfair discrimination. As Spencer Kimball, a leading professor of insurance law, puts it, fair discrimination means measuring as accurately as possible "the burden shifted to the insurance fund by the policy holder and charging exactly for it." Fairness, in this view, means ideally that no one pays for anyone else. That, however, is not the only definition of equity.

By its very nature, insurance is redistributive. We could theoretically squirrel away our individual savings to provide financial security for any of the contingencies we commonly insure against. Through insurance, however, we join with others to "pool" our risks and our savings. Only some people in the pool will experience the insured harm (say, fire, theft, or illness). Since only those who experience the harm will receive a payout, the others necessarily pay to help them.

Some advocates for the industry call this redistribution built into insurance a "cross-subsidy" and deem it anathema to fairness when it can be foreseen. In a 1987 Harvard Law Review article, Karen Clifford and Russell Iuculano, who represent the American Council of Life Insurers, argue that insurers have a legal duty to separate policy holders with serious, identifiable health risks from those without such risks. "Failure to do so represents a forced subsidy from the healthy to the less healthy."

The argument makes sense only if we understand the purpose of insurance as allocating costs to the people who generate them, rather than spreading the costs of misfortune and thereby making them more manageable. All insurance entails cross-subsidy. That is what makes it insurance instead of personal savings. Insurers typically put the adjective "forced" in front of "subsidy" when defending an underwriting criterion against regulatory challenge. But there is no reason why any particular cross-subsidy is more coercive than all the other cross-subsidy that insurance entails.

The debate about HIV testing in insurance, then, comes down to a fundamental disagreement about the purpose of insurance, regardless of whether an insurance fund is operated as a commercial enterprise, a social program, or some hybrid. Ultimately, the disagreement concerns whether to distribute the benefits of insurance according to prior contributions or according to need. Medical testing of any kind is valid only to the degree that we want our insurance system to minimize redistribution from the healthy to the (potentially) sick. With enough predictive tests of sufficient accuracy, insurers could virtually eliminate risk-sharing and redistribution. We would each pay strictly for ourselves. The industry argument about fair discrimination assumes a vision of insurance as a personal savings plan operated by insurance companies instead of banks.

Fairness in Insurance

What kinds of differentiation are fair? The industry answer is not helpful: Fair discrimination is what each state's Unfair Trade Practices Act allows, and unfair discrimination is what it forbids. These laws were all adopted at the behest of industry between 1947 and 1960. They typically define unfair practices as "making unfair discrimination between individuals of the same class," "discrimination between similarly situated individuals," or -- one of my personal favorites for its tautological brilliance -- "discrimination between insureds having like insuring characteristics."

How does one know whether people belong to the same class, are similarly situated, or have like insuring characteristics? Sesame Street, that universal mentor of the preschool set, has something to say on the matter. In one segment, the kids are shown three cardboard stars and a moon and asked, "Which one of these things is different?" The lesson is not that the moon is different, but that several equally valid answers depend on which criterion a person uses to differentiate -- say, color, shape, or size. The moon is different only if the children select by shape.

In health and life insurance, many different factors could be used to answer the question, "Which people are most likely to get sick?" The industry most commonly uses age, gender, medical history, and occupation, but it could avail itself of others. Why not use race? Blacks have higher rates of heart disease and kidney disease and lower life expectancy than do whites. Insurers could also use residence. Cancer rates vary by state, and people living near major medical centers are at greater risk of expensive surgery than others living farther away The industry could use veteran status. Vietnam veterans have higher rates of accidents and premature death than do nonveterans. The industry could use marital status. Illness is much greater among the widowed and divorced that among the married or single. On what basis, then, do we say that an insurer's classifications are fair?

The classifications used by the industry are dictated by neither medical science nor financial principles. They are a policy choice. The industry cannot use race because it is legally forbidden to do so as a result of a political choice made outside the industry. It does not use veteran status because it does not dare to penalize political heroes. But it does use medical criteria.

Insurers have decided that certain diseases render people ineligible for life insurance. These generally include diabetes, leukemia, schizophrenia, emphysema, coronary artery disease, and now AIDS. People with these diseases are deemed "medically uninsurable." Risk factors, such as uncontrolled high blood pressure, are also a basis for exclusion. The rationale is that people with these diseases and risk factors have a very high probability of early death. If they die prematurely, they will not pay enough money in premiums to cover the losses that they will generate, unless the insurer were to charge them such high rates as to make the insurance unaffordable.

According to the industry view, admitting people at high risk to a general insurance pool would be unfair to the other, lower-risk policy holders whose premiums would go up. Industry representatives portray any effort to ban the use of HIV tests by insurance companies as granting "favored status" to carriers of one disease. Since the industry already screens applicants for heart disease, cancer, stroke, and other diseases, why should AIDS be privileged?

ONE CAN SEE HOW policy holders would not want to be burdened with the costs of people likely to be very sick or to die prematurely. Indeed, the industry tries to foster this lifeboat mentality by running advertisements explaining why we should not want to pay for people who run high risks. "If you don't take risks, why should you pay for someone else's?" asks one such advertisement, showing a man high up on a steel scaffolding. Never mind that the man is building an office tower that presumably contributes to our economy. (He's wearing a hardhat and a tool belt, so he's probably not climbing for the thrill of it.) In the insurance industry's view, fairness means concentrating the costs of accidents and illnesses on the individuals who bear the risks.

What if we step outside the privileged circle of people protected by private insurance policies? From that vantage point, equity might seem to require that those who are ill or at risk for illness and injury ought to have greater access to insurance, not less. If they face high medical expenses, they need health insurance coverage all the more. If they have dependents, they need life insurance all the more to protect their family's well-being. From a societal perspective, the people who require protection the most are precisely those whom commercial insurance companies find it economically necessary and fair to exclude.

Outside the privileged circle, people with diabetes or high blood pressure might feel that they have been singled out only because they have a condition that scientists and insurers now recognize as leading to early death or disease. But each of us is a living bundle of risk factors. We all have a multitude of characteristics -- socioeconomic status, heredity, race, gender, education, residence, family status, occupation, degree of happiness, eating habits, driving habits, work habits, and who knows what else -- that are or could be associated with illness and premature death. Scientists have investigated only some of these factors, and insurers have chosen to use only a few in setting rates. People with recognized risk factors are disadvantaged by our scientific knowledge. Given medical progress in identifying the precursors of disease, the number of people facing these new penalties of predictive knowledge can only grow. Groups of policy holders and the companies that insure them have every incentive to determine who is likely to be sick, disabled, and prematurely dead and to exclude these people from their risk-sharing plans or set higher insurance prices for them.

Many would argue that insurance companies are justified in charging more to high-risk people to encourage them to lead safe and healthy lives. When a person can reasonably be expected to understand the dangers of an action and refrain from it -- say, speeding, hang gliding, or smoking -- using that behavior as a basis for setting insurance premiums might serve the goals of education and prevention.

While incentives are sometimes a reasonable concern in designing insurance classifications, we ought to be wary of using them to set health insurance prices. Few health risks are truly voluntary. Even smoking, the favorite candidate, is doubtful. Nicotine is addictive, and the decision to start smoking, usually made when people are quite young, is heavily influenced by societal pressures, such as commercial advertising. Most known risk factors, including smoking, are heavily concentrated among the poor and less well educated. Many reasons for this disparity have their origins outside the sphere of individual choice. Alternative sources of satisfaction and stress reduction are less available to the poor. The poor tend to have more dangerous jobs. Even though we don't understand all the causal mechanisms, virtually every risk factor for disease has a high correlation with poverty. To increase health insurance prices for people already disadvantaged by poverty and poor health is to penalize them triply.

There is nothing wrong with creating incentives for healthy behavior, but health insurance is simply the wrong place for society to conduct is education of good habits. Health insurance should guarantee access to health care. Health is essential to life, happiness, and productivity. No matter whether people may have inflicted illness or injury upon themselves, we ought not to withhold compassion or medical care once they are sick. And denying health insurance on the basis of disease risk factors -- even the most controllable actions -- effectively denies care to the sick.

Here another tension of insurance becomes evident. Insurance is about sharing risks within a community. Underwriting is about exclusion. The industry term "uninsurable" applied to people deemed to be at very high risk suggests that insurability is a quality of individuals. In fact, insurability is the set of policy decisions by insurers about whom to accept. It is not a trait, but a concept of membership. It expresses the criteria used by a group to decide whom to include and exclude from its redistributive system. Treated as a scientific fact about individuals, the notion of insurability disguises fundamentally political decisions about membership in a community of mutual responsibility.

A system of competitive insurers based on medical underwriting guarantees that as insurers scramble for customers and seek to control their risks, society will be divided into more homogeneous risk classes, and more people will be left out of insurance pools altogether. From a commercial insurer's perspective, that may be good business practice. But from a social perspective, the splitting up of insurance pools means the erosion of mutual aid.

How Predictive Testing May Affect Health Insurance

The insurance industry's use of HIV tests and other new predictive tests will significantly affect access to health care. In the United States, eligibility for health coverage depends on work, age, or disability. Employee group plans, the most common form of protection, cover approximately 60 percent of the population. Through Medicare, Medicaid, the Veterans Administration, the Indian Health Service, and various other federal and state programs, government provides coverage for an additional 20 percent. Of the remaining population, some 5 percent obtain individual insurance policies, and 15 percent have no coverage at all.

Many of those lacking insurance either have no way to obtain it or face much higher insurance premiums than do the typical members of large employee groups. About two-thirds of the uninsured are employees or their dependents, but the smaller firms where they tend to work can purchase group insurance only at high rates. Individual policies are often prohibitively expensive. Moreover, of applicants for individual health insurance, around 8 percent are rejected outright as medically uninsurable. Commercial insurers designate another 9 percent as substandard risks and charge them even higher premiums than are normal in the individual market, while Blue Cross-Blue Shield plans rate about 20 percent of applicants as substandard. Of course, anticipating rejection or higher rates, many of the uninsured who are sick or disabled do not bother to apply for individual policies.

Until recently, only applicants for individual policies and groups under 25 or so employees were subject to medical underwriting. But a survey by the Office of Technology Assessment recently found that more health insurers are beginning to screen group applicants for high-risk status. Three of every four commercial and Blue Cross companies were either screening or planning to screen for high-risk applicants in small group plans. For large groups, 58 percent of commercial insurers and 7 percent of Blue Cross-Blue Shield plans were using or moving toward screening. Medical underwriting in the group market will raise greater obstacles to employers wishing to provide all their employees with health insurance.

Employers themselves may also now increasingly take health risks into account when deciding whether to hire a prospective employee. In recent years firms of all sizes have faced staggering increases in health insurance costs; one way to keep those costs down is to avoid employing people with high risks of illness. In countries with national health insurance, employers have less incentive to exclude the potentially sick from jobs. But employers in the United States pay directly for the health costs generated by their own workers. The group plans sold by insurance companies are typically "experience" rated; that is, the premiums charged by the insurer are based on each employee group's profile. Moreover, a growing number of employers operate their own health insurance plans, chiefly to take advantage of a provision in the 1974 Employee Retirement Income Security Act (ERISA). Under that law, if a firm "self-insures," its health plan is exempt from state regulation as well as state taxes on insurance premiums. Today more than half of all workers are covered by employer self-insurance arrangements. As a result, employers today are insurers, and all the difficulties surrounding the use of testing by insurers come up with employers, too.

Rising costs and self-insurance give employers strong incentives to use predictive testing to screen out high-risk applicants for jobs. State and federal handicap discrimination laws have begun to protect employees from being fired simply because they have some potentially costly health risk that does not affect their current job performance. But these laws do not necessarily bar employers from refusing to hire an applicant who appears to be a health risk according to one of the new tests. Thus people deemed "uninsurable" may also become "unemployable," at least at firms with good jobs that carry health insurance as a fringe benefit.

AS SCREENING AND UNDERWRITING exclude more people from private health insurance, the costs of their care fall primarily to Medicaid and public hospitals. Recent studies have estimated that of all people with AIDS, Medicaid covers 40 to 50 percent nationwide, while private insurance pays for only about 17 percent (and that share is probably declining). Taking all sources into account, hospitals are being paid only around 80 percent of their costs for AIDS cases in the Northeast, Midwest, and West and a mere 45 percent in the South. Public hospitals face the greatest burden. In cities hard hit by AIDS, such as New York, San Francisco, Newark, and Miami, municipal hospitals are having to shift resources from other services to AIDS care. This pattern is symptomatic of the wider problem. As private employers and insurers avoid the sick and high-risk individuals among us, they displace the costs onto the public sector, which simply lacks the resources to meet those demands on top of the others it already faces.

One possible response, advocated by many people in the insurance industry, is to create state high-risk pools to cover people whom insurance companies turn down. Such pools already exist in some fifteen states. Since people relegated to the pools are by definition (or at least assumption) at high risk for disease, they pay up to twice the usual rate for health insurance and face very high deductibles. These costs put high-risk pools out of the financial reach of most people. Still, the pools run at a loss. Insurance companies are then assessed to subsidize the pools, based usually on their pro rata share of business in the state. Some states also subsidize the pools out of state revenues.

High-risk pools permit insurance companies to continue skimming off the people who are least likely to become expensively sick and to shunt the others into a public program. Those people lucky enough to gain regular insurance protection pay cheaper premiums because they share their expense with others who also are unlikely to get sick or die early. True, they may face higher taxes to help pay for those who depend on public programs and public hospitals. But precisely because they enjoy the privilege of cheaper and better private insurance, they are not likely to be strong advocates for improved public services. The political effects of segmented insurance pools thus reinforce the economic forces at work when insurers are able to take the best risks and exclude the bad ones.

Life Insurance and the Bottom Line

Most medical testing by insurance companies occurs in the sale of life insurance, where the monetary stakes are much greater for the insurance industry. Insurers are particularly worried about people buying insurance policies when they know themselves to be at high risk for early death, while the insurer does not. (This is known as "moral hazard" in insurance jargon.) Such purchases produce the phenomenon that insurers dread more than any other: adverse selection, that is, a skewing of policy holders toward those with heavier than expected losses. Insurers raise the specter of legions of people exposed to HIV taking out large life insurance policies. Without HIV testing, these policies would be priced at standard rates, but the policy holders would likely die in a few years after having paid only a fraction of the premiums on which the companies were counting. Thus, without testing, the companies say their solvency is in jeopardy.

The little information currently available about life insurance payouts to AIDS victims suggests that the industry has not yet suffered major losses. The reason is most likely that few people at high risk of AIDS had life insurance policies in the first place. Historically, life insurance has been sold primarily to married people with children, and then only to those with enough regular disposable income to make monthly payments. That set of people does not include large numbers of gay men and needle-sharing drug users -- the two largest risk groups for AIDS. Nonetheless, insurers worry that if they were now to be denied the ability to require tests, carriers of the AIDS virus and other people identified as high-risk by predictive testing -- would sign up for policies and produce big losses.

THE ADVERSE SELECTION argument serves an important rhetorical function in the debate. It casts moral doubt on people with high risk for AIDS or any other life-threatening disease. The unspoken message is that people who buy insurance knowing they are high-risk are social parasites. Adverse selection, Clifford and Iuculano say, "unfairly burdens other policy holders who must support the increased claims through higher premiums." Thus the insurers' representatives subtly turn sick people into moral outcasts to justify excluding them from risk-sharing arrangements.

However, while they are healthy, many people simply have no opportunity to contribute to an affordable health insurance plan. When the uninsured later face extraordinary medical expenses, the only way they can spare their friends' and family's assets is to make them beneficiaries of life insurance policies. Of course, some people who discover they have a fatal disease might try to buy as much life insurance as they could afford, not only as a hedge against their own catastrophic expenses, but also to enrich their family or partner, or even to borrow against the policy and enhance their own current consumption. Insurance companies and other policy holders should be protected from such knowing exploitation of the risk-pooling mechanism. Though breadwinners have a legitimate interest in a moderate amount of life insurance, society need not create an entitlement to a huge payout when a person dies. But predictive medical testing by insurers has so many bad consequences that we ought to find other ways to prevent life insurance abuses.

Other mechanisms besides medical testing could ameliorate the problem. Insurance regulators could redesign "incontestability clauses" in life insurance policies. In most states, these clauses allow an insurer to refuse to pay if the policy holder dies within two years of the policy's issue and has misrepresented information on the application. After two years, the insurer may no longer contest the validity of the policy. Since the latency period for AIDS is considerably longer, we could extend these clauses for AIDS to some reasonable length, perhaps five years.

If we remember that fife insurance is primarily a mechanism to strengthen family income security, we could first ensure that basic levels of insurance are available to everyone. In addition to providing a guarantee of basic health insurance, we might expand the survivorship component of Social Security (broadening the concept of survivors to allow benefits to be paid to people with long-term commitments regardless of marriage, blood ties, or sexual orientation). With universal health insurance in place, the need to take out fife insurance at the moment of illness would diminish, thereby lessening concern about insurers' access to predictive medical information. With such strengthened arrangements for financial security, we should then permit private life insurers to test applicants for policies with large face values, perhaps all those higher than three times the median income.

Some Political Lessons

If ever there were an issue that ought to have propelled us to national health insurance, the AIDS epidemic should have been it. No recent experience so graphically demonstrates the limits of private health insurance as a method of paying for sickness.

The insurance industry made clear its concern to escape as much of the cost as possible. Nonetheless, efforts to stop insurers from testing for AIDS and excluding the victims proved a failure. What can we learn for the next round?

Perhaps the biggest mistake in the HIV testing controversy was the failure to grasp the full import of medical underwriting and predictive testing. Representatives of public hospitals, Medicaid agencies, and state health and welfare departments were nowhere to be seen in the legislative and regulatory hearing rooms as HIV testing was debated. Nor did any of the disease-based groups, such as the American Heart Association and American Cancer Society, see their stake in the testing issue. Gay rights and AIDS advocacy organizations were left to do battle alone.

As a result, HIV testing was treated solely as an issue of discrimination and privacy, not as the profound structural issue it also is. Because gays do face substantial discrimination, and because prejudice against people who test positive was running rampant, the advocacy groups focused their arguments on the injustice of burdening a minority, the insurers' use of crude stereotypes, and the lack of counseling and confidentiality for people tested by insurance companies.

The charge of discrimination is often a powerful political resource in American politics, but it backfired here. Insurers were able to trump the charge with their own wild card: they threatened not to write business in states that restricted testing. Moreover, because opponents of testing framed the issue as discrimination against gays, they lost the opportunity for alliances with other groups whose members stand to lose from increased medical underwriting but do not see themselves as victims of discrimination.

THE UNDERWRITING ISSUE is bound to come up again, given the rapid pace of discovery of genetic markers for disease and the near-weekly announcements of environmental, dietary, and behavioral hazards to health. If we understand the broad values at stake, we should be better prepared to frame these emerging issues and mobilize alliances to defend a wider vision of social protection.

In the struggle over AIDS testing, the insurance industry adopted the argument used by all industries seeking to resist public regulation: we cannot operate efficiently, perhaps not even at all, if we are burdened with social objectives. This argument might be persuasive if there were a wall between the public sector and private sector. But that is not our world. Private and public forms of social aid are intimately entwined, since the people and troubles that commercial policies do not cover get pushed into the public sector. Either public insurance programs fill the gaps in private insurance, or the victims wind up on the streets, the welfare rolls, the doorsteps of voluntary agencies, and the beds of public hospitals.

Private insurance companies might point out that their policy holders still pay taxes to care for the "uninsurables," but that argument misses an important point. Why should medical underwriting be used at all for health insurance? Why should certain diseases, such as AIDS, be socially financed, while others are privately financed? And why should insurers be the ones to decide, through their underwriting policies, which diseases taxpayers will have to finance and which ones private insurers will cover?

The political debate is likely to focus on which risk factors and tests insurers ought to be permitted to use in selecting applicants. The choice of permissible underwriting factors is not neutral. It defines a set of people likely to be excluded from the better coverage of most private programs. Medical underwriting on the basis of diagnoses has a particularly cruel and perverse result. After determining that a person is sick or at high risk, insurers turn around and deny aid for exactly that need. True, if insurers accepted an applicant knowing he or she was already sick, they would no longer be insuring but simply providing a payment mechanism. But that point merely illustrates the limits of private insurance as a method for financing care of the sick. Private insurers do not hide their interest in denying coverage to the high-risk; they insist it is their obligation to turn their backs on people once it is clear that they are or will become expensively sick.

The problems in health insurance are so severe that many major insurance companies are beginning to realize they must reform their practices or have their business either taken over or regulated by government. Small business associations and even some insurance trade associations are actively pressing for state and federal laws that would stop some of the cream-skimming. As Robert Laszewski, executive vice president of Liberty Mutual, recently told The New York Times, "The notion that an insurance company should be making a profit by figuring out which Americans not to cover is no longer viable."

When used to exclude people from such basic services as health insurance, predictive testing divides our society in dangerous and undesirable ways. The debate should focus, not on which tests insurers should use, but on how medical testing undermines our institutions of social protection.

TO TACKLE THE INSURANCE conundrum, we need to take community as our starting point. From there, it is clear that the purpose of health, disability, and life insurance, at least at levels providing security against devastating losses, is precisely to distribute according to need. An effective campaign for broader risk sharing has to demonstrate that insurance practices are issues of membership, and that the predictable result of medical underwriting is to exclude those people who need help the most. Such a perspective should help build coalitions among all the disease and disability groups who are similarly affected by insurance underwriting. Finally, we must reveal the distinction between private and public social aid systems as wholly artifice. The selection of people and troubles to be covered by each sector ought to be a matter of conscious public policy, not the result of efforts by the insurance industry to skim off the good risks. That process inevitably puts the high-risk and the poor into a public sector lacking both adequate resources and majority political support.