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

Résumé

Introduction: We assess physicians’ ability to accurately predictlife expectancies. In prostate cancer this prediction is especiallyimportant as it affects screening decisions. No previous studieshave examined accuracy in the context of real cases and concreteend points.

Methods: Seven clinical scenarios were summarized from chartsof deceased patients. We recruited 100 medical professionals toreview these scenarios and estimate each patient’s life expectancy.Responses were analyzed with respect to the patients’ actual survivalend points, then stratified based on the demographic informationprovided.

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

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