A Value Framework for Cancer Screening: Advice for High-Value Care From the American College of PhysiciansFREE

Screening sets off a chain of events—a cascade—that can lead to benefit (such as longer life) or harm. When one is assessing the value of screening, it is important to count the benefits versus the harms and costs from the entire cascade. Benefit accrues only to a subset of persons with true-positive results (Figure 2). At optimal intensity, screening detects many of the cancer cases that could benefit from earlier treatment, leaving fewer additional cases to detect. When intensity is further increased to detect these few remaining cases, screening tests, diagnostic work-ups, and detection of cancer that would not benefit from earlier treatment are disproportionately increased, resulting in rapidly increasing harms and costs.

Optimal-intensity screening strategies seek to find the subset of abnormalities that have the greatest probability of eventually causing health problems and that are more treatable at an early, asymptomatic stage. One can think of many cancer cases as having 3 general rates of progression from asymptomatic to fatal disease (Figure 3): rapid (often difficult to treat [patient 1]), intermediate (more amenable to treatment if found before symptoms [patient 2]), or slow (treatment not needed because the cancer will never cause symptoms [patient 3]). Detection of this third type of cancer constitutes overdiagnosis (detection of cancer that will not progress to symptoms during the patient's lifetime). Screening preferentially detects these slowly progressive cancer cases (and precancer cases) because they spend more time in the “detectable but not symptomatic” zone; higher-intensity screening increases this tendency. This is one of the factors that explains the steeper slope of the harms and costs curves with increasingly intensive screening (Figure 1). Screening confers benefit only when it detects cancer with intermediate progression.

Optimal-intensity screening strategies focus on persons with sufficient risk for potentially fatal cancer who also have low competing health risks from other causes. Overdiagnosis is due to detection of not only slowly progressive cancer (Figure 3, patient 3) but also any type of cancer in patients with serious noncancer health risks that will end their life before the cancer becomes symptomatic (Figure 3, patient 4). An example is a hypothetical woman whose breast cancer is detected as a palpable lump (that is, by symptoms) at age 55 years and who would die of the cancer at age 65 years. With detection by screening at age 52 years and effective treatment, she would instead live until age 80 years and die of another cause, such as a stroke. Thus, her benefit (living 15 years beyond age 65 years) would come 13 years after she is screened at age 52 years. However, if this woman has severe congestive heart failure that would end her life at age 70 years, her benefit from breast cancer screening would be reduced from 15 years to 5 years. If her noncancer risk is even more serious (for example, if she has severe diabetes, end-stage renal disease, and cirrhosis), her life span may be decreased such that there is no benefit from breast cancer screening (Figure 3, patient 4). In fact, there may be unintentional harm if anxiety from and treatment of breast cancer reduce the quality and length of her life. This is another reason for the decrease in the slope of the benefit curve and the increase in the slope of the harms and costs curves with increasing screening intensity (Figure 1): The screening population is enlarged to include more persons with serious noncancer health risks.

For some persons and some cancer screening strategies, screening can have important benefits by reducing cancer morbidity and mortality, although the number whose lives are extended by screening may be surprisingly small. For example, compared with no screening, annual screening mammography for 10 years prevents about 2 breast cancer deaths for every 1000 women aged 50 years (24). To achieve this benefit, however, there are real harms, and for many more patients than those whose lives are extended (23). A patient moving through the screening cascade (Figure 2) may encounter several types of harms, including physical harms (such as complications from the screening test or the diagnostic work-up), psychological harms (such as anxiety and sleepless nights while waiting for results from the screening test or work-up), opportunity costs (such as distraction from other meaningful life events), and financial strains (such as anticipated financial problems from positive screening or work-up results) (23). We briefly discuss 2 of the harms that have received attention: false-positive screening results and overdiagnosis.

False-positive screening results are common. For every 1000 women aged 50 years who are screened annually for breast cancer for 10 years, about 600 have at least 1 false-positive mammography result (24). Although anxiety from false-positive results is transient for many women, condition-specific worries and intrusive thoughts may persist for others, even after they are told that the positive result was not due to cancer (25, 26).

Overdiagnosis leads to different and more persistent harms. For every 1000 women aged 50 years who receive annual breast cancer screening with mammography for 10 years, about 7 are overdiagnosed with breast cancer that will never progress to being clinically important (24) (this estimate carries uncertainty but is probably the correct order of magnitude). In the psychological realm, the harm of overdiagnosis is labeling—the woman's life changes suddenly after she is labeled as a patient with cancer (23). In the physical realm, she is subjected to the harms of unnecessary treatment.

Determining value requires assessing the balance between all benefits versus all harms and costs in the screening cascade. This involves considering the number of persons receiving the benefit (or harm) and the type of benefit (or harm). The number of persons harmed in some way by screening is always larger than the number of cancer deaths prevented (21, 23), but the weight of a single person's benefit is often greater than a single person's harm when counted in cancer deaths prevented. Thus, the balance often comes down to many persons experiencing some degree of harm versus a few experiencing a greater degree of benefit.

The ultimate arbiter of this balance is the voice of the informed public, but how to include it is uncertain. One extreme position is that scientific experts can examine the tradeoffs and determine the most likely opinion of the informed public. The other extreme is that informed individual patients should decide the value of every strategy for themselves. A tenable middle ground is that experts (with adequate oversight) could determine value in situations that are reasonably clear. In less clear, close-call situations, public input from individual shared decision making or (for policy decisions) “deliberative democracy” methods (27) could determine value. This approach is reflected in the flat top of the value curve in Figure 1. The exact strategy that optimizes value may vary among individual patients and groups.

Although cost is an important factor in determining value, finding the best approach to include it in the balance of benefits versus harms and costs has been a problem in the United States. The use of cost-effectiveness analyses (especially those that include all benefits, harms, and costs with a societal perspective) and outcomes tables can provide information about the tradeoffs. Both economists and the informed public (through deliberative methods) could be involved in adding cost into the assessment of value.