Presenting Benefits and Downsides to Facilitate High-Quality Decision-Making About Cancer Screening

In this issue of the Journal, Caverly and colleagues examine how US clinical practice guidelines present the benefits and downsides of five major cancer screening topics: breast, cervical, colon, lung, and prostate cancer screening (1). They reviewed 32 guidelines from major organizations, including the US Preventive Services Task Force (USPSTF), the American Cancer Society, the American College of Physicians, and the National Comprehensive Cancer Network. Notably, they found that less than one-third of positive recommendations presented both benefits and downsides in an appropriate manner, defined as the use of comparable absolute effects. The remaining recommendations either had incomplete information or presented the benefits and harms asymmetrically, such as showing benefits in terms of relative risk reduction and harms in terms of absolute risk increases. Caverly and colleagues concluded that improved presentation of benefits and downsides in guidelines “would better ensure that clinicians and patients have access to the information required for making informed decisions.”

Clearly, we have substantial room for improvement in the quality of medical decision-making, including cancer screening. Numerous studies have documented poor communication about the benefits and downsides of different cancer screening options among clinicians, patients, and members of the general public. In a recent national population-based survey of men and women age 50 years and older who had a cancer screening discussion in the past two years, participants reported that although most discussions included some information about the benefits of screening, only about 10% included any discussion of potential harms (2). Studies employing audio-taping of clinical encounters reach similar conclusions: Elston-Lafata and colleagues, for example, have shown that fewer than 10% of provider-patient discussions about colorectal cancer screening had any mention of potential downsides of screening (3).

In the context of suboptimal provider-patient communication about screening, it should not be surprising that patient surveys suggest limited knowledge of the magnitude of benefits and harms for these key cancer screening topics. In most cases, patients or members of the general public overestimate the benefits of screening and may underestimate or not recognize common potential downsides, including false-positive test results and risk of overdiagnosis and resultant overtreatment (4–5). Multiple studies have shown that proving patients with well-balanced, easy-to-understand information about the benefits and harms of different cancer screening services in the form of patient decision aids can improve patients’ knowledge of the magnitude of potential benefits and downsides and, in some cases, influence their willingness to undergo screening (6).

Caverly and colleagues’ findings of the limitations in how information about benefits and downsides are presented in clinical guidelines could be an important root cause of inadequate patient-provider communication and low patient knowledge about cancer screening services. If providers rely on information from clinical guidelines to inform their evaluation of the net benefit of cancer screening services, then asymmetry between how benefits and downsides are presented may lead such providers to misjudge the net effects of a given service, which may lead to inadequate or inappropriate communication of these effects to patients. Of course, this mechanism presupposes that providers use guidelines (directly or indirectly) for this purpose and that asymmetric or missing information directly affects how providers communicate about screening. Further research is required to determine if this mechanism is indeed operational, but it is certainly plausible.

If we assume that guidelines are important sources of information for communication about cancer screening, we must then ask why the presentation of information about benefits and downsides in guidelines seems to be suboptimal. One explanation may be that guideline developers are simply not aware of the potential negative effects of presenting benefits and downsides incompletely or asymmetrically. A second explanation may be the lack of high-quality, generalizable studies that accurately estimate important downsides: For example, it is difficult to estimate the risks of radiation-induced cancers for radiological screening tests such as CT scans for lung cancer or CT colonography for colon cancer.

Another dilemma relates to the challenge of synthesizing estimates of effect from different studies. Guideline developers may need to combine estimates of benefit, such as cancer-specific mortality reductions, across a range of studies with different underlying risks of disease. In such cases, it is more appropriate to combine estimates of relative risk reduction into a summary effect measure rather than trying to combine absolute effects, which are critically dependent on the underlying risk in the population being studied (7). Guideline developers may encounter this type of problem more frequently for estimates of benefit; in contrast, they may rely on other sources for information about downsides, particularly rare harms, that are not as prone to have differences in risk across studies and hence can be expressed effectively as absolute risk increases.

One technique to overcome this type of asymmetry is to first determine the summary relative risk reduction for the measures of effect that differ by risk level, then use simple modeling or outcome tables to show comparisons of absolute benefits and downsides for different levels of risk. Such an approach has the added benefit of showing how the net benefit may differ across risk levels. Bach and colleagues illustrated the potential advantages of this type of approach for the case of lung cancer screening. They showed that the ratio of benefits to downsides differed considerably across different risk groups, thus allowing better targeting of screening to those most likely to benefit (8). In order to apply this type of information clinically, providers must be able to assess their patients’ risk. To address this concern, Caverly and colleagues have developed an interactive, web-based lung cancer screening decision aid (www.shouldiscreen.com) that calculates lung cancer risk and the potential benefits and downsides of screening based on individual patient characteristics (9).

One challenge to this type of approach for presenting benefits and downsides is that the effects are often obtained from trials of limited duration. When presenting information about benefits and downsides, guideline makers would ideally like to present information over a uniform and clinically relevant time range. In the example of lung cancer screening, the National Lung Screening Trial examined the effect of three rounds of screening and follow-up over seven years. Ideally, we would like to inform providers and patients about the potential benefits and downsides over the full duration of a screening program, which for lung cancer might begin at age 55 years and continue for 20 years in some cases.

One option for overcoming this challenge is to use modeling. To inform the USPSTF recommendations, the CISNET collaborative modeled the benefits and downsides of lung cancer screening (10). The USPSTF used the modeling results to develop an outcomes table in their recommendation statement showing the absolute numbers of lung cancer deaths prevented (benefits), as well as imaging tests required and radiation-induced lung cancers caused (downsides) (11). Of course, such estimates depend on whether the results seen in the trials can be extrapolated over longer time periods and in populations that may differ from those enrolled in the trials. How best to estimate and represent the uncertainty inherent in extrapolation from the primary evidence, as well as the collective uncertainty around the parameters included, remains a topic in need of additional research. As Caverly and colleagues illustrate, we still have substantial room for improvement in how we translate and communicate evidence.

Dr. Pignone is a member of the US Preventive Services Task Force. The views presented here are his and not necessarily those of the USPSTF.