Challenging the 10-year rule: The accuracy of patient life expectancy predictions by physicians in relation to prostate cancer management

Abstract

Introduction: : We assess physicians' ability to accurately predict life expectancies. In prostate cancer this prediction is especially important as it affects screening decisions. No previous studies have examined accuracy in the context of real cases and concrete end points.

Methods: : Seven clinical scenarios were summarized from charts of deceased patients. We recruited 100 medical professionals to review these scenarios and estimate each patient's life expectancy. Responses were analyzed with respect to the patients' actual survival end points, then stratified based on the demographic information provided.

Results: : Respondent factors, such as sex, level of training, location of work or specialty, made no significant difference on prediction accuracy. Furthermore, respondents were typically pessimistic in their estimations with a negative linear trend between estimated life expectancy and actual survival. Overall, respondents were within 1 year of actual life expectancy only 15.9% of the time; on average, respondents were 67.4% inaccurate in relation to actual survival. If framed in terms of correctly identifying which patients would live more than or less than 10 years (dichotomous accuracy), physicians were correct 68.3% of the time.

Conclusions: : Physicians do poorly at predicting life expectancy and tend to underestimate how long patients have left to live. This overall inaccuracy raises the question of whether physicians should refine screening and treatment criteria, find a better proxy or dispose of the criteria altogether.

Fig. 1

The dichotomous accuracy of life expectancy predictions in relation to a spectrum of life expectancy cut-offs. As actual survival moves towards the 10 year mark, dichotomous accuracy decreases precipitously; however, as actual survival increases past this point, dichotomous accuracy slowly climbs back up to near 100%.