« Happy Fourth! | Main | Chicago renaissance? »

Silly economist


One of the things I did this weekend was read this book, What's Your Life Worth? Health Care Rationing... Who Lives? Who Dies? Who Decides? by David Dranove, a healthcare economist. There were good insights here, and some good comic relief, too.

The central point of the book is that we're not spending our health care dollars as efficiently as we could be, because we haven't embraced the techniques of cost-benefit analysis/cost-effectiveness analysis ("CBA/CEA") with enough enthusiasm. Everyone is familiar with these techniques. They're just the old "biggest bang for the buck" approach dressed up with a lot of jargon, and applied to health care spending. Dranove argues that if healthcare payers (the government, employers, and insurers) would explicitly adopt what he constantly (and misleadingly) refers to as the "scientific" approach of CBA/CEA, we could "save many more lives" with the money we choose to spend on health care.

So far, so sleepy. Everyone expects these kinds of arguments from an economist, but Dranove actually says some interesting things in this short book written for the layman. Before I get to the parts that had me laughing out loud in the Starbucks, I'll talk about what I think are the really worthwhile insights of Dranove's book. (You may want to skip ahead to "CEA and QALYs" if you're only reading this for the laughs.)

First, we ration health care--always have, always will. Since we are rationing, it makes more sense to do it explicitly, using techniques that most of us will accept as reasonable, than to do it implicitly by less rational means, all the while denying that we're rationing at all. That's where CBA/CEA comes in. Dranove argues that spending ought to be directed at those treatments that give us the greatest benefit for the least cost, and he thinks the techniques he describes in his book are the best way to do this. All of this is great. Dranove's description of the techniques, though, were good for some really good laughs. Keep reading to find out why...

Second, Dranove argues that we may be spending too little on health care, not too much. It's refreshing to hear this contrarian point of view. I don't know if Dranove is right or not. "Too much" or "too little" are value judgments about which reasonable people will disagree. My own intuition is that health and longevity are so important for most people that if their preferences could be accurately measured, health care spending would, in fact, increase. So I agree with Dranove about this, but I think the argument he uses to get himself there is a little goofy. Ok, a lot goofy. It's the old CBA/CEA thing again, which Dranove fetishizes like an old lady at the slot machine kissing the quarters, or the courtside fans waiting for Kobe to throw them his sweaty jersey. Really; get a grip on yourself, people. Like so many economists, Dranove takes a sensible idea and wields it irrationally, to my great delight in the Starbucks. I just had to laugh.

CEA and QALYs

Ok, here's the funny part: QALYs, which stands for "Quality-Adjusted Life-Years." They're funny because Dranove presents them as a "scientific" alternative to messy and unreliable political appraisals of what medical treatment is worth. His enthusiastic economist's soul loves the numbers so much, he starts to sound like a twelve-year-old boy who's just fallen in love with Cameron Diaz (Cameron Diaz is mentioned in the book on p. 143, but I'm sure it's just a coincidence).

As Dranove explains them, QALYs are a way of measuring the relative desirability of various "health states." I'll let Dranove explain them:

The QALY scale provides a numerical score for different states of health, or what researchers call health states. Examples of health states are "dead," "completely healthy," and "complete visual impairment" (i.e., blind). Health states may include multiple conditions, such as "visual impairment and needs a wheelchair." . . .

A year of full health is worth 1 QALY. Death is worth 0 QALYs. Depending on how individuals respond to survey questions, a year of blindness might score 0.5 QALYs and a year of incontinence might score 0.75 QALYs. . . .

QALYs can be manipulated mathematically. They can be added: if one year of blindness scores 0.5 QALYs, then four years of blindness would score 2 QALYs. Similarly, if two blind individuals each live one year, they have a total of 1 QALY. . . .

Researchers can use QALYs to compute cost-effectiveness scores for treating different health states. . . .

Researchers measure QALY scores by asking carefully worded questions of a large cross-section of respondents. . . . The relative scale approach is by far the easiest. . . . Here is one example you can try yourself. [Gives example of a horizontal line with "0" on one end, marked "Blindness," and "1" on the other, marked "Occasional blurred vision," with ".5" in the middle. Under the line is the instruction: "Use a pencil to indicate where you would place the following health states on this line: (1) frequent blurred vision (2) constant blurred vision."] . . .

The relative scale approach is easy to describe and can even be conducted through the mail. However, the approach has two important weaknesses. First, it easily bogs down when the respondent must rank more than five health states. Second, respondents often get frustrated trying to place health states along the scale [endnote omitted]. . . . This ambiguity makes mathematical manipulations of the scores somewhat tenuous.

To more rigorously quantify the QALY scores, researchers rely on one of two other approaches. Here is the standard gamble approach [endnote omitted]. Once again, you can try this method yourself.

Consider the following alternatives: (1) You will be blind for the rest of your life (2) You will undergo a procedure to restore your vision. The procedure will be successful with probability p, where p is a number between 0 and 1. The procedure will fail with probability 1-p. If the procedure fails, you die instantly.

At what value of p are you indifferent between alternatives 1 and 2?

Many respondents are unfamiliar with the concept of indifference. . . . Many respondents are also unfamiliar with the concept of probabilities. . . .

The third approach is called the time trade-off. To understand the time-tradeoff-approach, we return to the prospect of living with blindness.

Consider the following alternatives: (1) You could live 50 more years with blindness, and then die. (2) You could live y more years in full health, and then die.

How big does y have to be for you to be indifferent between the two choices?

Dranove goes on like this, inviting us to rank blindness, death, "frequent vomiting," and "confined to bed with poor memory." He glowingly describes these procedures as "scientific" and therefore far superior to the more "intuitive" rankings that other methods, such as the political process, tend to provide. Frankly, I don't think I could answer his relatively simple example questions very easily, and if I tried to do it twice, my answers would probably be so different that I'd have to answer them fifty or sixty times before I could even say anything about my ranking of health states, let alone some "objective" society-wide ranking. That ranking would just be a numerical average, and might never actually equal any of my actual rankings at any given point in time.

In fairness to Dranove, he does point out the limitations of QALY surveys for ranking health states, among them unfamiliarity with the condition, ignorance of probability, wide discrepancies in rank between individuals, the enormous influence of the wording of the questions, etc. It's hard to see why he is so enthusiastic about these techniques when he can acknowledge they are plagued with so many apparently unsolvable problems. But wait. There's more.

QALYs are concerned only with the "benefit" side of the cost/benefit analysis, or "CBA." The other side, "costs," aren't explored in depth in Dranove's book. He does occasionally note that reliable cost data is often unavailable. This, at least, seems undoubtedly correct. When you consider the opacity of drug prices and the ubiquity of billing errors, it's not surprising that "good cost data" is often "lacking." Given the extremely complicated nature of our health care system, the costs of treatment may be more difficult to quantify than the benefits.

Even though both sides of the cost/benefit analysis that Dranove holds out as the silver bullet for many of our health care spending problems are deeply problematic, Dranove himself seems oblivious. Nowhere is this more obvious than in his account of Oregon's experiment with the kind of explicit rationing that he argues for in his book.

Oregon, under the leadership of emergency physician John Kitzhaber (who was later elected Governor), adopted a system of explicit rationing for its Medicaid program in the early '90s. Oregon came up with a list of treatments and conditions, and ranked them according to several criteria. It then refused to pay for any treatments that were ranked lower than a certain number. The goal was to use scarce Medicaid dollars for the most important treatments only, enabling the state to expand Medicaid coverage to everyone in Oregon that qualified for the program.

The way Dranove explains the development of the Oregon health plan, it seemed to this reader to demonstrate that the kinds of CBA/CEA techniques that Dranove describes in his book can only be improved by a healthy injection of intuition and political discussion. To Dranove, however, the Oregon health plan is a case study of how "science" lost out to "intuition" and "politics."

Basically, the first few lists that Oregon came up with were unacceptable to real Oregonians. Strict application of CBA/CEA techniques in a vacuum inevitably produced lists that real people weren't happy with--things that people thought should not be excluded were excluded, and vice-versa. Before the Oregon health plan was passed, the lists were subjected to community scrutiny through the political process, and revised. Only then were people comfortable enough with what was covered and what was excluded to support the plan.

Despite the fact that the quantification of "benefits" is supposed to ultimately depend upon what real people think, Dranove treats this after-the-fact revision of the lists as an example of "poorly formed opinions" replacing "scientific inquiry" as the basis for public policy. I'll leave it to you to respond to that one on your own. As for myself--I laughed heartily. Silly economist!

Despite the drawbacks of Dranove's book, there's a lot to recommend it. If you've never had any exposure to QALYs or "costs of illness" (COI), as I hadn't, you'll probably learn something useful from this book. For example, you'll learn that according to virtually every way of measuring that's ever been tried, from questionnaires to studies of salary premiums for high-risk jobs, the cost of an extra year of life is always approximately $150,000. If you believe this number, we're paying too little for health care in this country, since per-capita spending on treatments that offer extra years of life is much less than this.

You'll learn something from Dranove's book. But you'll learn even more if you read it with a healthy degree of skepticism.