Do Attending and Resident Physicians Agree on Patient Prognoses?
CCCF ePoster library. Yarnell C. Nov 7, 2018; 233401; 22
Christopher Yarnell
Christopher Yarnell
Login now to access Regular content available to all registered users.

You may also access this content "anytime, anywhere" with the Free MULTILEARNING App for iOS and Android
Rate & Comment (0)



Physician prognostication is fallible and differences in prognoses estimated by attending and trainee physicians may represent an opportunity for education and improved end-of-life care. This study assessed prognostic agreement between attending and trainee physicians using the surprise question, a popular end-of-life screening tool that asked “Would you be surprised if this patient died during this hospital admission?”




Determine the prevalence of agreement and factors associated with discordance between attending and trainee physician estimates of patient prognosis using the surprise question.




This multicentre prospective observational survey asked “Would you be surprised if this patient died on this admission?” to general internal medicine attending and senior trainee physicians for each of the patients for whom they were responsible. Baseline data for patients included age, gender, functional status, residence type, admission diagnosis, comorbidities, length of stay, creatinine, albumin and goals of care. Baseline data for physicians included year of medical school graduation and duration of time on current service. Primary analysis included calculation of overall agreement and Cohen’s kappa. Secondary analyses assessed the association between baseline covariates and surprise question response agreement using mixed effects logistic regression to account for clustering by physician pair.




The study enrolled 21 pairs of attending and trainee physicians based at 5 different hospitals caring for a total of 419 patients (median team size 20, IQR 17-22). Patients had median age 75 years (IQR 61-85) and 201 (48%) were female; 320 (76%) came to hospital from their own residence and 106 (25%) were dependent for activities of daily living. Missing data for the surprise question responses was rare (1/838 responses). Trainee and attending physician responses for admission mortality agreed in 335 of 419 patients (80%, standard deviation 12%). The Cohen’s kappa was 0.40 (95% confidence interval 0.31 to 0.5) for agreement on admission prognosis, although kappa values for each physician pair ranged from -0.17 to +1.0. In discordant responses, attending physicians were more likely to respond “No, I would not be surprised if this patient died during this hospital admission” (OR 5.76, p = 0.016). Logistic regression modeling with random intercepts clustering by physician pair showed association between prognostic agreement and female sex (OR 2.04, 95% CI 1.05 to 3.73) and CPR status “Yes” (OR 2.43, 95% CI 1.25 to 5.09). Variation between physician pairs measured by median odds ratio (MOR 1.78, 95% CI 1.02 to 2.76) was similar in magnitude to variation by patient-level characteristics.




Trainee and attending physicians show moderate agreement on the prognoses of their patients as measured by the surprise question. The adjusted model suggests that sex and CPR status may correlate with prognostic agreement. Limitations include the subjectivity of the surprise question and future work includes collection of 12-month mortality data. A deeper understanding of variability in physician prognostication will benefit all physicians who care for critically ill patients.

    This eLearning portal is powered by:
    This eLearning portal is powered by MULTIEPORTAL
Anonymous User Privacy Preferences

Strictly Necessary Cookies (Always Active)

MULTILEARNING platforms and tools hereinafter referred as “MLG SOFTWARE” are provided to you as pure educational platforms/services requiring cookies to operate. In the case of the MLG SOFTWARE, cookies are essential for the Platform to function properly for the provision of education. If these cookies are disabled, a large subset of the functionality provided by the Platform will either be unavailable or cease to work as expected. The MLG SOFTWARE do not capture non-essential activities such as menu items and listings you click on or pages viewed.

Performance Cookies

Performance cookies are used to analyse how visitors use a website in order to provide a better user experience.

Save Settings