Health care and the technologies that support it are of growing concern to most Americans. Until recently, I was something of an exception. I lead a fairly sheltered existence; I seldom think about the issues one encounters in the media. I have focused instead on my research: first in the field of computing; then, since 1990, in some of the fundamental issues of biology. During most of that time I’ve felt that I had nothing of special importance to say about medical issues.
Now, however, I strongly believe that the rate of progress in the field of medicine is much slower than it needs to be, and that this lag affects all of us profoundly. Many wonderful advances have been made. Yet I am disturbed by the discoveries that have not happened, or have happened but have not yet been allowed to reach the market. We must discuss the forces that constrain and retard possible breakthroughs in medical treatment, especially at this point in time, when we are on the verge of profound shifts in medicine driven by advances in technology.
One could expend a great deal of effort on detailed arguments about regulation, innovation, and desirable goals, but most of the detail would miss the essential point: we are all affected, and many of us are dying, because the system optimizes the wrong set of goals. A group of powerful people have established professional goals for medicine that seem appropriate to them and, in fact, to most Americans; yet the very selection of these aims – minimizing medical accidents, for example, or reducing the distribution of ineffective cures – has had a host of unintended, unfortunate consequences. These consequences must be discussed.
To this end, I invite you to consider three fantasies, three visions of a world in which alternative approaches to science and technology were taken. These fantasies have to do with (1) the impact of government regulation, (2) the cost of risk aversion, and (3) the choices that people make about the treatment of the terminally ill. In other words, these are fantasies about real things, and I will be making some real-world observations about them.
Fantasy 1: The Speed of Innovation
I began my career in computing in 1967. It is hard to convey how rapidly the field developed. This is how Chris Evans summarized it in 1979:
[S]uppose for a moment that the automobile industry had developed at the same rate as computers and over the same time period: how much cheaper and more efficient would the current models be? If you have not already heard the analog the answer is shattering. Today you would be able to buy a Rolls-Royce for $2.75, it would do three million miles to the gallon, and it would deliver enough power to drive the Queen Elizabeth II.
That was written just as the micro revolution was begin- ning. Since then, computers have affected the world in ways that were, and remain, almost unimaginable.
Now, fantasize for a moment about what would have happened if the following arguments had been successfully advanced back in the 1970s and 1980s:
1. Computers will inevitably perform central functions in many products, and the issue of their reliability is too critical to be left to the marketplace. Indeed, we will see computers exercising critical functions in airplanes, spacecraft, coordination of rail traffic, maintenance of nuclear reactors, and·an almost unlimited number of other settings. Failures in either hardware or software can cause inconvenience, injur~or even death. Therefore, the new Federal Computer Authority (FCA) will be responsible for licensing computer products. This agency will be assigned the task of certifying the reliability of each new product.
2. Because computers will playa crucial role in every consumer’s life, the software used to control them must be developed by professionals who have been certified as capable of delivering state-of-the-art products. Therefore, only certified graduates of a limited number of licensed institutions will be allowed to produce software for commercial uses, and people trained in foreign institutions will be required to pass strict examinations guaranteeing their familiarity with the best current practices.
3. Normal consumers are obviously not qualified to make judgments about whether or not the software they purchase will live up to expectations. The FCA must therefore have the power to determine the efficacy of programs before they are marketed. Products that are considered critical will be sold only through licensed outlets, where certified professionals can prescribe products that they consider necessary· to address consumers’ needs, products that have been tested and approved for these applications. For relatively non-critical applications, we will allow mass-marketing outlets to handle the appropriate products.
How would things develop from that point? At first, the FCA would probably take a pragmatic course, attempting to weed out the truly bad software and hardware while minimally constraining forward progress. The product development cycle would certainly become longer. But the really important, and unfortunate, result would be the enshrinement of the idea that the FCA was responsible for minimizing or even eliminating serious product failures.
Once the FCA acquired responsibility for preventing accidents, incentives would exist for it to make increasingly cautious judgments. As the agency became more risk-averse, the population would draw greater and greater solace from media reports of its careful regulation and quality-control. Some of the reports might be false; corruption might set in within the FCA. Because it would effectively control the release of all new technology, its approval would be something worth paying for. But if the FCA was doing its job, the product cycle would continue to lengthen, product development would become steadily more expensive, and the number of new products would be kept to a decent and approved minimum.
Now, it is precisely this fantasy situation, this kind of regulatory environment, that currently exists in the health care industry. As a result, health care technology is progressing at only a fraction of the pace that might be achievable.
In the case of computing we see what can be accomplished with minimal regulation. In the case of health care we observe the outcome of a highly regulated process. Most cost-benefit analyses of regulatory protocols consider the trade-off between the prevention of accidents and delay of the products that reach the market. But if you reflect on the computing industry you will realize that the effect of regulation is the simple nonexistence of many products that could have reached the market, but did not. In such cases, you cannot quantify the effect of regulation, because the advances just cease to occur; the innovations just cease to happen. The delay of events that do eventually occur is certainly important, but the dramatic reduction in innovation is far more so.
Right now, reduction of innovation is an issue of grave importance. We have reached the stage in our understanding of the human genome where a dramatic acceleration of medical technology is possible. Let me explain.
Within the cells of every person’s body is a collection of DNA that determines many of the details that support life. This collection of DNA, called a genome, acts as blueprint for processes that allow a living cell to begin, grow, and divide. We can now gain access to the information stored in a genome through a process called sequencing. The cost of sequencing a person’s genome is dropping rapidly, and it is the information learned in this process that will drive much of modern medicine. The basic vision goes like this:
1. Many, if not most, drugs have a favorable effect on a small percentage of the population but a neutral or unfavorable effect on other people. If we can, through the sequencing of genomes, predict which individuals will react favorably to a specific drug, the usefulness of drugs will be dramatically increased.
2. We will be sequencing a large number of diverse human genomes. We will then look at them and see how they differ. We have tabulated specific spots that reflect differences, even
though we cannot now identify the effect of each difference. Let us call such a difference an SNP (this happens to stand for “single-nucleotide polymorphism,” a fact that you need not
We are all affected, and many of us are dying because the health care system optimizes the wrong set of goals.
remember). We will determine a set of, say, a million SNPs; then we will check them against specific human beings, creating an SNP profile for each person.
3. When we have accumulated thousands, eventually millions, of SNP profiles, we will correlate them with the ways in which people react to specific drugs. In this way, we will be able to predict, with gradually improving accuracy, which people can benefit from those drugs. The generation and use of these SNP profiles has the potential for revolutionizing modern medicine. It may result in the saving of millions of lives.
This is only one of many technologies that will grow out of our ability to sequence large numbers of genomes at rapidly decreasing costs. Such innovations are likely to occur, however, only in the regulation-free environment that has characterized computer technology.
Fantasy 2: The Price of Risk Aversion
This second fantasy involves anticancer drugs. Suppose that two such drugs are available. Further suppose that each of them has the following properties:
• The drug is a total cure for exactly 500/0 of the population.
• The drug is lethal for the other 500/0 of the population.
And let us suppose that the two drugs are complementary – together they could cure all members of the population, assuming that a sick person could know which of the two drugs to take – and that the use of SNP profiles will eventually allow prediction of benefits with 100% accuracy. In other words, suppose we can foresee a time when we will be able to make an entirely accurate prediction of the efficacy of these drugs, though right now our ability to predict is pretty poor.
This is clearly a contrived fantasy, but I believe it captures the essence of the current situation. We do have many drugs that offer dramatic therapeutic benefits to one or another subpopulation, but potentially lethal damage to other subpopulations. We do have an emerging technology (SNPs) that will increase the predictability of benefits.
How might this field of investigation progress, in the absence of regulation? Here is how I fantasize the course of events involving the two complementary drugs:
1. At first the manufacturers of each drug would offer it to individuals who willingly risk a 50% chance of death in order to gain a real chance of being cured, given that, absent the use of either drug, they would almost certainly die soon. These potential consumers would be informed of the risks; some would accept them; and 50% would be cured – while 50% would die.
2. Because the value of the drugs would be directly related to the chance of curing the patients, strong incentives would exist to improve the odds of success. This would encourage efforts to characterize SNPs for all the patients who took the drugs. Long before the cause of success or failure could be accurately determined, it would be possible to identify subpopulations that exhibited much higher and much lower success rates for each drug.
3. The odds of predicting which of the two drugs would work for any specific patient would begin to improve. This would lead to more patients taking the risk of ingesting one or the other of the drugs, which would lead in turn to a rapidly growing body of statistics that could support more accurate analysis of correlations between SNPs and therapeutic outcomes. A positive feedback loop would be established, quickly leading to vastly improved outcomes.
Now the question I wish to pose is this: is there any way this can happen, given existing principles of regulation?
Under current law, a drug may not be brought to market unless it is judged both safe and effective. Consider what would happen if a drug showed up in a lab, it looked promising, and a doctor gave it (probably illegally) to two friends who were dying from cancer, one of whom died, while the other was totally cured. If the story became known, I suspect the doctor would face prosecution, even if both his friends had been fully informed of all the data he possessed, and both had elected to take the risk. In my view, a well-informed adult should have the right to take potentially fatal risks. But that is yet another fantasy, because this right is wholly unrecognized in the present medical environment.
To summarize: the normal progression of events in an unregulated, risk-friendly environment would have every possibility of producing a 100% cure rate, while the existing, regulated, risk-averse environment would bar any progress toward using the full therapeutic properties of either of the two drugs in question. This alone constitutes grounds for rethinking the priorities of the existing system.
Fantasy 3: The Purpose of Saving Lives
In my last fantasy, I want you to think of yourself as a person who runs a government agency with a substantial budget, an agency that is responsible for saving the lives of as many people as possible who have been diagnosed with cancer and
The effect of regulation is the simple non- existence of many products that could have reached the market, but did not.
have a probable life expectancy of less than a year. The lives of thousands of these people depend on your judgment. (Remember, every day 1,500 people die from cancer in America.) Your job is to make decisions that can save their lives. The U.S. government alone spends about $5 billion a year just on cancer research. Let us suppose that you have a serious portion of that budget to spend. What will you do?
Take some time and think about it. I don’t believe that the answer is completely obvious, but I would guess that if people think seriously about this fantasy, they will identify a number of ideas that clearly should be tried. I believe that if hundreds of highly qualified scientists, doctors, and entrepreneurs were asked to describe how they would proceed, many useful options and insights would emerge. I have performed this exercise a number of times myself. No matter how often I reconsider the issue, I just do not see people arriving at something like this:
I have it; I see the key principle. What we need to do first is to make it impossible for anyone to take a drug that might kill him. That is where we should start. Then we need to construct a list of cures that have been proposed but probably won’t work. We will make sure to prevent such scams.
The idea that sensible people might come to such conclusions strikes me as completely preposterous. Yet as you know, these are the policies that are now in place, supported and mandated by government and the health care establishment. The result is predictable: our ability to provide medical technology to patients in need is dramatically constrained, and the cost-benefit analyses that underlie all business decisions, and should underlie all practical decisions about health care, are heavily skewed to reflect the costs of conforming to regulatory requirements.
It is clear that most Americans sincerely believe there is nothing but goodwill behind this system. I have no doubt, however, that your own fantasy of what you might do to help the terminally ill will stand in judgment of things as they are.
I have used three fantasies to argue that the current health care industry in the United States is crippled by regulation and risk aversion. The effect has been to constrain product development and innovation.
What is truly upsetting about this situation is, that we are commonly confronted with arguments in which the U.s. health care system is designated as “private”, and contrasted with the “public” systems of other countries. This distinction focuses on the delivery of the existing technology of health care. Instead, the focus should be on the set of industries that advance the technology of medicine and medical procedures. In that context the critical issue is the degree of regulation, which in turn affects the rate of innovation.
Medicine, like computing, is entering a period in which extremely rapid innovation impossible, most notably advances relating to the availability of economic data. I believe the advances that will inevitably occur will be so stunning that we will lose sight of how paltry they are – compared with what might have been, if innovation had been unconstrained.