Abstract 353: Physicians Do Not Accurately Predict Length of Stay of Patients Admitted with Heart Failure
Objective: Determine the accuracy of physicians’ predictions of length of stay of inpatients with heart failure and compare those predictions based on physician experience.
Background: Discharging a medical inpatient is a complex process often dependent upon a multidisciplinary team. Discharge planning begins with the physician who must anticipate patients’ lengths of stay, response to acute therapy, complications, and post-hospitalization needs. Increasing focus on patient safety emphasizes eliminating preventable harm including 30-day re-admissions, medication errors, and insufficient post-discharge care. To achieve these goals, early discharge planning is critical and is contingent upon an accurate estimate of length of stay at the time of admission. However, the accuracy of physician predictions of length of stay has not previously been studied.
Methods: In this prospective, single-center, observational study we measured physicians' predictions of length of stay of patients admitted to an academic referral hospital’s heart failure teaching service. We compared admitting interns’ (PGY1, less than 6 months experience), supervising residents’ (PGY2-4, 1-3 years experience), and attending cardiologists’ (mean 19±9.7 years experience) predictions at admission with actual length of stay. The primary outcome was accuracy of physician predictions stratified by experience. Secondary outcomes included comparisons among predictions made by interns, residents, and cardiologists.
Results: Predictions were made on 166 consecutively admitted patients within 24 hours of admission. Of those 166 patients, 98 (59%) had predictions by physicians at all three levels, with predictions made by 21 interns, 24 residents, and 8 cardiologists. Patients had a mean age of 56±14 years old and were 62% male (61 of 98). Median length of stay was 7.5 days (interquartile range 4-13). Mean difference between predicted and actual length of stay was statistically significant for all groups: for interns, -5.8 days (95% CI -8.2 to -3.4, p<0.0001), residents, -4.6 days (95% CI -7.1 to -2.0, p=0.0001), and cardiologists, -4.3 days (95% CI -6.5 to -2.1, p=0.0003). There were no statistically significant differences among intern, resident, and attending predictions, as determined by GEE modeling (p=0.61).
Conclusion: Physicians, regardless of experience level, systematically underestimate length of stay of patients admitted with heart failure. Prediction models could assist with improved discharge planning.
Author Disclosures: M.S. Durstenfeld: None. M.D. Saybolt: None. A. Praestgaard: None. S.E. Kimmel: None.
- © 2014 by American Heart Association, Inc.