Medical expenditures and their rate of growth in the United States are high. Many health care professionals claim that early detection of medical conditions would reduce health care expenditures and improve patient welfare. But empirical research conflicts with this belief, according to Overdiagnosed, one of the most important books about health care in the last several years.

The authors clearly state their argument early in the book:

Americans have been trained to be concerned about our health. All sorts of hidden dangers lurk inside of us. The conventional wisdom is that it's always better to know about these dangers so that something can be done. And the earlier we know, the better. … Americans love diagnosis, especially early diagnosis. ...

But the truth is that early diagnosis is a double-edged sword. While it has the potential to help some, it always has a hidden danger: overdiagnosis—the detection of abnormalities that are not destined to ever bother us.

Normal becomes abnormal | Early diagnosis arises from two distinct trends in medicine. The first is the "renorming" of standards; that is, the redefinition of what readings from medical tests are designated as abnormal and worthy of treatment even though other clinical symptoms are absent.

For example, before the late 1990s, a blood pressure reading of 160-over-100 was considered the threshold for the diagnosis of hypertension. Today it is 140-over-90. The five-year risk of a bad event (death, heart attack, and stroke) for diastolic pressure of between 90 and 100 is only 9 percent. If the pressure is lowered with medication, only 3 percent of people experience a bad event for a treatment benefit of 6 percent. Thus, just one in 18 patients with diastolic pressure in this range who is treated with medication avoids a bad health event in the next five years that would have happened without treatment.

That benefit may be worth the money spent on medication for all 18 patients, but the cost of treatment goes beyond the pharmacist's bill. Medical intervention is never precise, i.e., there is a range of effects from taking medication. So, for instance, blood pressure medicine lowers some people's pressure so far that they faint—which is more than just an inconvenience. Widespread treatment of all people with pressures above the new guidelines would result in more fainting.

Cholesterol levels have also been renormed. In 1998 a trial showed that reducing total cholesterol below 200 mg/dL reduced major health events over five years from 5 percent to 3 percent of the studied population. Choosing a cutoff of 200 as abnormal had large effects on medical practice because 200 was near the middle of the population frequency distribution. The change from a standard of 240 to 200 increased the number of abnormal people by over 42 million, or 86 percent. Over a lifetime (24 years of treatment for the average 58-year-old in the trial), the results from treating everyone with a total cholesterol level between 200 and 240 mg/dL are that (for every 100 patients) eight will benefit, 14 will have bad events despite treatment, and 78 will be "overdiagnosed"—they wouldn't have experienced bad events even in the absence of treatment.

Seeing trouble everywhere | The second source of early diagnosis is advances in medical imaging. Computerized tomography (CT) and magnetic resonance imaging (MRI) scans allow doctors to see abnormalities in people. Many of these abnormalities will ultimately have no adverse effects on people's health. Yet this reservoir of clinically unimportant abnormalities combined with more aggressive "preventive" imaging leads to diagnosis and a cascade of further testing and intervention. Some 40 percent of people with no knee pain have meniscus damage in their knees. Some 50 percent of people with no back pain have bulging discs. And 7 percent of people under age 50 in the Framingham health study had silent strokes with no symptoms. The lifetime risk of dying from prostate cancer is 3 percent, but the reservoir of undiagnosed prostate cancer is enormous—almost 50 percent of men age 50–59 have prostate cancer, for example, but it isn't clinically important. Even lung cancer has a large reservoir of abnormalities in people who are well. The landmark study of British physicians by Richard Doll and Austin Hill in the 1950s found that the lung cancer death rate over five years in smokers was 17 times higher than in those who never smoked. Yet in the early 2000s a study of 5,000 people with spiral CT scans found a cancer rate incidence in smokers that was only 1.1 times that of nonsmokers (11.5 per thousand vs. 10.5). The conclusion is that many nonsmokers have lung cancer that isn't clinically important.

The net result of a large reservoir of abnormalities, improved imaging, and increased testing of those without symptoms is that the incidence of "disease" increases dramatically, survival rates increase dramatically, but mortality rates from those diseases remain constant. Kidney and thyroid cancer as well as melanoma incidence is rising dramatically, but death rates haven't changed. The introduction of mammography increased breast cancer incidence by 50 percent, but death rates have decreased just slightly. Some 90 percent of mammography-diagnosed cancer is of two types: overdiagnosed (the cancer will not affect the patient's health), or the cancer will harm the patient's health regardless of treatment. For mammography, only one in 1,000 women who are screened every year for 10 years will benefit, while two will be overdiagnosed, 5–15 will be diagnosed correctly and early but not have their prognosis changed by medical intervention, and 250–500 will experience a false alarm.

Ironically, all of this overdiagnosis and these false alarms contribute to more overdiagnosis and false alarms. Oftentimes, people who initially receive bad health news but later receive good news (the illness is not spreading or the initial bad test proved incorrect) are so relieved that they encourage their friends and family to get tested. Overdiagnosis and false alarms then propagate. However, if someone forgoes testing and is later diagnosed with a late-stage case of some disease, he naturally wonders what would have happened if the disease had been "caught early." Yet in truth, the outcomes are basically unaffected by early detection.

Consumer/voter revolt? | How do we get out of this state of affairs? University of Illinois law professor and Cato adjunct fellow David Hyman has argued ("In Medicine, Money Matters," Winter 2010–2011) that American medicine's fee-for-service system, which rewards health care providers for performing tests and providing treatment even if they provide no health benefits, incentivizes overdiagnosis and false alarms. So what prevents a different business model from arising and saving money by not testing for and finding clinically unimportant abnormalities? In the 1980s and early 1990s, employers experimented with incentivizing employees to join health maintenance organizations (HMOs) in which physicians were salaried and acted as gatekeepers to reduce access to specialists. But consumers did not like the utilization restriction strategies. Employers stopped imposing them as labor markets tightened in the late 1990s boom and the "patients' bill of rights" movement threatened to make utilization controls subject to government regulation. So while economists are correct that changing incentives would change outcomes, consumers have resisted the imposition of such incentives both in the labor market and politically.

The late economist Herbert Stein said, "If something cannot go on forever, it will stop." Increasing health care costs cannot go on forever because there really is a budget constraint, even in the United States. One strategy to implement the insights of the evidence-based medicine movement is the Independent Payment Advisory Board created by the 2010 Patient Protection and Affordable Care Act. The board is charged with devising ways to constrain Medicare and Medicaid expenditures that will automatically be implemented unless overturned by a vote of Congress. To me this feels like the second installment of the managed care movement. The trick will be to avoid the result from the first attempt, which was consumer revolt. Unless average voters embrace the claims of this book, this second attempt to impose constraint on the "medical industrial complex" will fare no better than the first.

Readings

  • "Adverse Consequences of Adverse Selection," by Mark Pauly and Sean Nicholson. Journal of Health Politics, Policy and Law, Vol. 24, No. 5 (October 1999).
  • "Health Care Costs: On the Rise Again," by Sherry Glied. Journal of Economic Perspectives, Vol. 17, No. 2 (Spring 2003).