Book Review

Health Care Politics

Educated Guesses: Making Policy about Medical Screening Tests

N Engl J Med 1994; 331:1461November 24, 1994

Article

Educated Guesses: Making Policy about Medical Screening Tests
By Louise B. Russell. 128 pp. Berkeley, University of California Press, 1994. $30 (cloth); $10 (paper). ISBN: 0-520-08365-2 (cloth)

This book is a stunner: concise, lucid, wise, important. Run, do not walk, to your nearest bookstore to buy a copy, and read it promptly. It will tell you all you ever wanted to know about screening tests but were afraid to ask lest you appear to be unsophisticated. Even those with a fear of mathematics can proceed without trepidation; there are no formulas; to understand the book requires no more than basic arithmetic.

Screening tests are all the rage these days. It has become an article of faith that early detection of disease (or the risk of disease) permits interventions that can be expected to improve outcomes. That is obviously a great thing to do. Everybody knows that an ounce of prevention is worth a pound of cure and that a stitch in time saves nine. But is “everybody” right? Is it ever preferable to stand there and do nothing rather than rush in to do something? How to decide when these aphorisms are right and when they are wrong is what Professor Russell explains clearly in less than 100 pages. She does this by examining in some detail three widely recommended screening procedures for cervical cancer, prostate cancer, and high blood levels of cholesterol. For the first, there is good evidence that periodic Papanicolaou smears save lives when treatment is provided; for the second, it may be that screening results in more harm than good; for the third, the evidence is at best equivocal. Understanding the logic behind the analysis of these three well-chosen examples will enable medical practitioners to assess for themselves which screening tests make sense for their patients.

All physicians know (or at least should know) that “unless early treatment makes a difference, screening is pointless.” The International Agency on Research on Cancer has shown that screening with Papanicolaou tests and early treatment can reduce deaths from cervical cancer. Then, what is controversial? Why not go full steam ahead? To begin with, as with any test, there are false negative and false positive results, the former leading to missed cases and false reassurance, the latter leading to repeated testing, considerable anxiety for the patient, and sometimes unneeded surgery. Second, laboratory quality control is still unsatisfactory. Third, because population-wide testing is costly, the key issue is to determine the optimal interval between tests. Available data indicate that there is a 90.8 percent reduction in cases of invasive cancer when testing is conducted every three years, a 92.5 percent reduction when testing is conducted every two years, and a 93.5 percent reduction when testing is conducted every year. Correspondingly, the cost per year of life saved is about three times higher for yearly testing than for testing every three years. As a public health investment, far more would be gained by investing the funds required for annual as opposed to every-other-year testing to reach out to the one quarter of women who are not reached once in five years, or at all.

In the case of screening for prostate cancer, the basic problem is that we simply have no good evidence of the effectiveness of treatment itself in improving survival; under these circumstances, debates about screening become almost irrelevant. Given the sizable number of serious side effects from treatment (impotence and incontinence), assessing the worth of treatment must come first. But how can it be true that detection is not worthwhile? Data on surgical outcomes demonstrate that five-year survival is greater the smaller and the more contained the cancer is at the time of removal. The uncertainty arises from the difficulty of distinguishing “innocent” from life-threatening cancers. Autopsy studies show that almost a third of men between the ages of 50 and 70 years have asymptomatic cancer of the prostate; the rates go up with age. Therefore, screening will necessarily be associated with an apparent increase in five-year survival even if treatment is ineffective, because of lead-time bias (the cancer is found earlier in its course), “length” bias (slow-growing cancers are more likely to be detected), and “overdiagnosis” bias (tumors are detected that would never have caused the patient any trouble). Only a randomized, controlled trial of screening followed by intervention can determine whether screening by digital rectal examination, measurement of prostate-specific antigen, or transrectal ultrasonography actually improves survival. This August, the Food and Drug Administration approved the use of prostate-specific antigen for screening; as a result, it will be more and more widely used. There is no doubt it will increase the number of cases found. But will it save lives or worsen the quality of life for men who will be treated for a growth that would have been innocuous if left undisturbed?

Controversy is greatest with regard to screening for high blood levels of cholesterol. Why should it be necessary? After all, 10 years ago the Lipid Research Clinics trial showed that lowering cholesterol levels was associated with a reduction in heart attacks (when both fatal and nonfatal attacks are summed together). The trouble is that neither that trial nor a half-dozen others showed that lowering cholesterol levels extended life; the reduction in mortality due to coronary disease was offset by an increase in mortality from other causes in these studies. In fact, a clear benefit from lowering blood cholesterol levels is found only for the subpopulation of men who have had heart attacks. For asymptomatic persons (except those with markedly elevated cholesterol levels or other important risk factors), the jury is still out. Under such circumstances, recommending the use of drugs and even diet to lower blood cholesterol levels in adults, let alone children, makes little public health sense. Yet it is widely recommended.

The costs involved in current screening policies are enormous. Professor Russell points out that “the billions of dollars involved in these three screening tests alone are enough to finance a system of basic care for the poor and uninsured.” In this season of my discontent with the fiasco in Washington involving what is laughingly called health care reform, these words are worth pondering.

In these days of disclosing conflicts of interest, I am obliged to confess my biases. Since I read and admired Professor Russell's earlier book, Is Prevention Better than Cure? (Washington, D.C.: Brookings Institution, 1986), I was predisposed in her favor. My neutrality as a reviewer has been jeopardized by my predilection for authors who do not stand between the reader and the topic, for authors who simplify a topic rather than make it more complex, for authors who are willing to let the thing speak for itself. These qualities are so rare that once again I urge readers to buy a copy of this book before it sells out, as I hope it will.

Leon Eisenberg, M.D.
Harvard Medical School, Boston, MA 02115