Realizing the potential of clinical judgment: a real-time strategy for predicting outcomes and cost for medical inpatients

Am J Med. 2000 Aug 15;109(3):189-95. doi: 10.1016/s0002-9343(00)00477-0.


Purpose: We sought to determine whether illness severity and anticipated level of function, as evaluated at the time of admission, were associated with outcomes and costs of care for patients admitted to the medical service.

Methods: All 1,759 patients admitted to the medical service at a large urban academic medical center between July 1, 1997, and September 30, 1997 (excluding those admitted directly to the intensive care units or for protocol chemotherapy), were evaluated and categorized by the admitting intern by illness severity (not ill, mildly ill, moderately ill, severely ill, or moribund) and anticipated level of function at discharge (excellent, good, fair, or poor) as part of their routine sign-out process. Interns' ratings were always available within 24 to 28 hours of admission. In-hospital mortality, length of stay, cost of hospitalization, and anticipated billing revenue were evaluated.

Results: Patients who were more severely ill had significantly greater in-hospital mortality. For example, mortality was 1.1% (11 of 972) among those who were not ill or mildly ill, 3.6% (26 of 724) among those who were moderately ill, and 15% (9 of 60) among those who were severely ill. Illness severity (P = 0.003) and anticipated functional status (P < 0.01) were significant predictors of in-hospital mortality. Illness severity and function were also significant predictors of greater length of stay and greater costs of hospitalization (all P < 0.0001). The 389 patients who were moderately ill with fair or poor anticipated function were associated with the largest cumulative losses (about $330,000 during the 3-month period), whereas the 798 mildly ill patients with good or excellent function were associated with the largest cumulative profits ($550,000).

Conclusion: Physicians' estimates of patients' illness severity and anticipated function at the time of discharge, as made by interns using a system designed to help them sign out to their colleagues, predict outcomes and costs of hospitalization. Such a system may be useful in developing new approaches to management strategies based on prognosis.

MeSH terms

  • Academic Medical Centers / economics*
  • Academic Medical Centers / statistics & numerical data*
  • Adult
  • Aged
  • Aged, 80 and over
  • Clinical Competence*
  • Female
  • Hospital Costs*
  • Hospital Mortality
  • Hospitals, Urban / economics
  • Hospitals, Urban / statistics & numerical data
  • Humans
  • Judgment*
  • Length of Stay / statistics & numerical data
  • Male
  • Middle Aged
  • New York City
  • Odds Ratio
  • ROC Curve
  • Severity of Illness Index*
  • Treatment Outcome*
  • Utilization Review / statistics & numerical data