Screening for prostate cancer with prostate-specific antigen (PSA) remains a problematic aspect of primary care. With the exception of the US Preventive Services Task Force (USPSTF), which made an influential(1) but highly-criticized(2) recommendation against any early detection efforts based on PSA, most guidelines recommend that an informed decision be made by the individual patient following a discussion with a doctor. For example, the American College of Physicians (ACP) states that “doctors and patients should discuss the potential benefits and harms of screening” before PSA testing. This recommendation is made even though the ACP is generally skeptical of the benefits of PSA (“for most … men, the harms will outweigh the benefits”)(3). Likewise, the American Urologic Association (AUA), while more favorably predisposed to screening, makes a similar statement in “strongly recommend[ing] shared decision-making.”(4)

Yet implementing shared decision-making in primary care is not straightforward, as it must take into account the wide range of information and data that could be discussed, the complex trade-off between immediate harms and long-term benefits, and the limited time primary care clinicians have for in-depth discussions about PSA in the context of a multitude of other issues in a typical visit. Recent years have seen a considerable literature develop on PSA decision-making, including specific advice to primary care providers as to what they should say to individual patients. Our multidisciplinary group, a statistician specializing in localized prostate cancer, a bioethicist who has conducted empirical research about decision-making as to PSA screening, an academic urologic oncologist and epidemiologist, and an academic primary care physician, has followed this literature closely. We think recommendations either specify too little information to allow patients to make a decision or so much as to overwhelm patients’ ability to decide rationally. Recommendations requiring extensive information also have low clinical feasibility — one suggests that doctors inform patients on 16 separate points and ask 12 questions concerning preferences(5) - include data that might be hard for patients to understand or assign a value to, such as the risk of deep vein thrombosis(6), or cite estimates that are conflicting and questionable, such as PSA leading to either 30(7) or, alternatively, 110(6), extra diagnoses of prostate cancer per 1000 men screened.

Given the inadequacies of current recommendations, and attendant poor compliance, we pose an alternative approach for informed decisions about PSA testing in primary care. This is based on three main principles. First, what is said to the patient must be based on best evidence and, as far as is possible in such a controversial field, must be beyond dispute. This would avoid the situation of many decision tools, such as the infographic provided by the National Cancer Institute(6), in which many of the key numbers cited, such as the risk of overdiagnosis, are subject to considerable controversy. Second, the patient should be presented with a clear framework for a decision. This contrasts explicitly with complex decision aids that provide patients with a large number of estimates and then asks them to integrate these somehow into a discrete choice. Third, the schema should be appropriate for primary care. It should not assume detailed knowledge by the provider, nor require more than a few minutes.

As a starting point, we assume that primary care providers correctly identify eligible patients: men in their mid-forties through mid-seventies with minimal comorbidity. We also assume that providers adopt the “ask-tell-ask” approach that has been previously advocated(8). Starting with the initial “ask”, the clinician finds out critical information regarding what the patient already knows about PSA or what their level of concern or interest may be. This allows the clinicians to tailor their “tell” portion of the conversation more succinctly and directly to the patient’s particular needs and level of current understanding. The “tell” portion of the conversation then follows the simple schema outlined below. The Web appendix provides evidence supporting each point.