Effective Clinical Practice
Effective Clinical Practice, March/April 1999
H. Gilbert Welch, MD, MPH
John D. Birkmeyer, MD
Lisa M. Schwartz, MD, MS
William C. Black, MD
Steven Woloshin, MD, MS
Physicians have traditionally cared for the sick. Patients have sought our help to alleviate pain, ameliorate symptoms, and when possible, to cure disease. Over the past couple of decades, however, there has been increasing public pressure for the medical profession to pay more attention to the well. Specifically, we are being encouraged to prevent disease. Although this occasionally refers to persuading people to adopt a “healthy lifestyle” (e.g., exercise regularly, eat right, and avoid risky behaviors), more commonly it means that physicians are seeking disease in their patients.
Should the standards of evidence be different for interventions targeting the well? I think so. Although our symptomatic patients generally ask for help, asymptomatic people are being told that they need our help. When suffering patients call for help, it seems reasonable to act with a lower standard of literature-based proof. The demand for proof is further reduced when direct evidence of the intervention's usefulness, albeit circumstantial, may be provided from the patient's response to therapy (i.e., whether symptoms persist or resolve). For people who are well, however, benefit cannot be observed directly because the expectation is that most people will stay well regardless of what is done. Instead, there is an implied pledge made to individuals who are well, (1,) (2) a pledge that services delivered now will prevent disease in the future. The burden of proof must lay with proponents of early detection to demonstrate that it works.
In this issue of ecp, Black makes it clear how much the quality of the evidence matters. (3) He reminds us how cursory evaluations of screening are subject to some of the most powerful biases in medicine and may lead to dramatically wrong conclusions (see Primer ). The combined effect of lead-time, length, and overdiagnosis biases can make negative effects seem to produce survival benefit. In short, early detection could actually harm people while appearing to help.
The randomized trial must be the fundamental standard to demonstrate that early detection can provide net benefit. Only a randomized trial can disentangle the real from the apparent effects. As Black points out, however, randomized trials have serious limitations of their own. One obvious limitation is generalizibility—a trial provides a test of screening under well-specified conditions, whereas clinical practice may be more haphazard. Another is the need for large numbers of study participants. Black argues that evaluations of screening will ultimately need to be based on a combination of trial data and decision modeling. Perhaps the best we can expect is to have the real effect of an early detection strategy demonstrated under a few well-specified conditions and then make careful inferences about how changing conditions (e.g., target population, screening frequency, new tests) will affect net benefit.
At first glance, there is every reason to believe that early detection should work. If people are examined carefully enough by using advanced laboratory or imaging technologies, then most disease ought to be "caught" at an early stage. It also stands to reason that disease found earlier will be easier to treat. Consequently, much of the mortality and morbidity of advanced disease should be preventable. The idea of early detection is so appealing that there has been a dramatic growth in the use of diagnostic tests—as part of systematic efforts (the Appendix Table provides the current cancer screening recommendations of the U.S. Preventive Services Task Force and the American Cancer Society) or as more routine testing in general (witness the finding, also in this issue, that one quarter of the elderly in Miami undergo echocardiography each year ).
But there are downsides to early detection. First, many people must be involved but only a few can benefit. To encourage people to be screened, proponents must articulate a message that motivates people to do so (exemplified by the "1-in-9" statistic for breast cancer). Too often this persuasion involves overstating the risk for the target disorder and exaggerating the potential benefit of screening. In an effort to find the few, a sort of 'disease' is spread across the population.
Second, early detection identifies more people as being sick. To find cases earlier, the threshold for making the diagnosis must be lower. The "advance" in diagnostic testing is the ability to find disease that was previously undetectable. Not surprisingly, advanced diagnostic tests invariably find more cases. The problem is larger than the familiar one of false-positive test results—it includes true-positive results where the “disease” identified has uncertain significance (as evidenced by the large reservoir of undetected thyroid, (5) breast, (6) and prostate (7) cancers). There will be some patients who would never become aware of their disease were it not for the test. In short, there may be people who previously felt healthy and are being unnecessarily informed that they are sick.
Also in this issue of ecp, Schwartz and Woloshin (8) demonstrate just how ubiquitous disease could become. Using the National Health and Nutrition Examination Survey, they calculated the impact of new disease definitions on the number of Americans who have four common conditions: diabetes, hyperlipidemia, hypertension, and obesity. The bottom line is that three out of four adults meet the criteria for one of these conditions. After digesting these data, some of you may not feel so well. (I myself thought seriously about a new diet.)
Third, early detection strategies imply that more people will be treated. Some can benefit—those whose disease is both 1) destined to cause morbidity and mortality and 2) not as effectively treated once it becomes clinically evident. Some will be treated unnecessarily—either because their disease was always destined to remain subclinical or because later treatment would have worked just as well. Others will be harmed—those for whom the potential risks (however rare) of unnecessary treatment become a reality.
Finally, the potential exists for early detection to become distracting to clinicians. (9) Consider how the interest in measuring quality interacts with the focus on prevention. Measures of prevention are easily understood and have wide appeal. Moreover, they are relatively easy to obtain. It is clearly easier for health plans to measure how well providers comply with screening recommendations than it is to determine how good they are in helping sick patients. Similarly, it is easier to improve physician compliance on an agreed-on preventive goal than it is to improve the quality of care given the sick. But that doesn't mean that prevention should become the focus of health care. Although all physicians would like to see more disease prevented, we should ask ourselves if the focus on finding early disease is distracting us from caring for the sick.
1. Cochrane AL, Holland WW. Validation of screening procedures. Br Med Bull. 1971;27:1-8.
2. Lee JM. Screening and informed consent. N Engl J Med. 1993;328:438-40.
3. Black WC. Should this patient be screened for cancer? Effective Clinical Practice. 1999;2:86-95.
4. Lucas L, Wennberg DE, Malenka DJ. Variation in the use of echocardiography. Effective Clinical Practice. 1999;2:71-75.
5. Harach HR, Frarissila KO, Wasenius VM. Occult papillary carcinoma of the thyroid: a "normal" finding in Finland. A systematic autopsy study. Cancer. 1985;276:1008-13.
6. Nielsen M, Thomsen JL, Primdahl S, Dyreborg U, Andersen JA. Breast cancer and atypia among young and middle-aged women: a study of 110 medicolegal autopsies. Br J Cancer. 1987;56:814-9.
7. Montle JE, Wood DPJ, Pontes E, Boyet JM, Levin HS. Adenocarcinoma of the prostate in cytoprostatectomy specimens removed for bladder cancer. Cancer. 1989;63:381-5.
8. Schwartz LM, Woloshin S. Changing disease definitions: implications on disease prevalence. Analysis of the Third National Health and Nutrition Examination survey. 1988-1994. Effective Clinical Practice. 1999;2:76-85.
9. Fisher ES, Welch HG. Avoiding the unintended consequences of growth in medical care: how might more be worse? JAMA. 1999;281:446-53.