BACKGROUND: Predicted patient life expectancy, based on a patient's medical history, is an important component of medical decision making. This study therefore aimed to determine the consistency, accuracy and precision with which doctors, nurses and medical students predict life expectancy (LE). METHODS: 20 doctors, 20 nurses and 20 medical students (4th and 5th year) independently examined 70 hypothetical patient case scenarios containing age, sex and comorbidity; this included 13 duplicate scenarios. Accuracy and consistency of prediction was assessed by comparison with statistical LE estimates generated using evidence-based actuarial and life insurance industry methods in collaboration with a team of professional actuaries. RESULTS: Doctors, nurses and medical students underestimated LE by a mean (95% confidence interval) of -1.46 (-0.31 to -2.61), -1.79 (-0.52 to -3.06) and -2.24 (-1.16 to -3.32) years with an equivalent root mean squared error (RMSE) of 4.74, 5.49 and 5.08 years respectively. LE predictions were equal to actuarial LE in less than 10% of cases and accurate to within 25% of actuarial LE in less than 45% of cases. Intra-observer reliability was 91%, 85% and 87% for doctors, nurses and medical students respectively. Inter-observer reliability was 66%, 57% and 57% for the three groups. CONCLUSION: Doctors, nurses and medical students were inconsistent, inaccurate and imprecise in their prediction of LE with a tendency toward underestimation. This may lead to patients being managed inappropriately. There is a need for improved training and objective outcome prediction models.
BACKGROUND: Predicted patient life expectancy, based on a patient's medical history, is an important component of medical decision making. This study therefore aimed to determine the consistency, accuracy and precision with which doctors, nurses and medical students predict life expectancy (LE). METHODS: 20 doctors, 20 nurses and 20 medical students (4th and 5th year) independently examined 70 hypothetical patient case scenarios containing age, sex and comorbidity; this included 13 duplicate scenarios. Accuracy and consistency of prediction was assessed by comparison with statistical LE estimates generated using evidence-based actuarial and life insurance industry methods in collaboration with a team of professional actuaries. RESULTS: Doctors, nurses and medical students underestimated LE by a mean (95% confidence interval) of -1.46 (-0.31 to -2.61), -1.79 (-0.52 to -3.06) and -2.24 (-1.16 to -3.32) years with an equivalent root mean squared error (RMSE) of 4.74, 5.49 and 5.08 years respectively. LE predictions were equal to actuarial LE in less than 10% of cases and accurate to within 25% of actuarial LE in less than 45% of cases. Intra-observer reliability was 91%, 85% and 87% for doctors, nurses and medical students respectively. Inter-observer reliability was 66%, 57% and 57% for the three groups. CONCLUSION: Doctors, nurses and medical students were inconsistent, inaccurate and imprecise in their prediction of LE with a tendency toward underestimation. This may lead to patients being managed inappropriately. There is a need for improved training and objective outcome prediction models.
Authors: Jennifer L Moss; Siddhartha Roy; Chan Shen; Joie D Cooper; Robert P Lennon; Eugene J Lengerich; Alan Adelman; William Curry; Mack T Ruffin Journal: JAMA Netw Open Date: 2020-07-01
Authors: Laura E Brotzman; Rachel C Shelton; Jessica D Austin; Carmen B Rodriguez; Mariangela Agovino; Nathalie Moise; Parisa Tehranifar Journal: Cancer Med Date: 2022-05-26 Impact factor: 4.711
Authors: Sarah A M Hallen; Norbert A M Hootsmans; Laura Blaisdell; Caitlin M Gutheil; Paul K J Han Journal: Health Expect Date: 2014-05-12 Impact factor: 3.377