PURPOSE: COPD is a common disease and one of the leading causes of hospitalization in the US. An easy-to-use severity stratification tool may facilitate COPD disease management. We sought to develop and validate a bedside application tool using data routinely available on admission.
METHODS: We analyzed 57,791 COPD admissions in 2004-2005 across 191 hospitals (79 teaching and 112 non-teaching hospitals). We used Classification and Regression Tree (CART) approach to derive and cross validate the prediction rule for in-hospital mortality. We included age, gender, laboratory findings, vital signs, and comorbidities on admission as candidate variables.
RESULTS: Among the 57,791 admission, the overall crude mortality was 2.4%. The median age was 72 years old with an inter-quartile range of 63-79. There were 55% women. The CART analysis yielded three risk factors with the highest discrimination power: BUN >25 mg/dl, presence of altered mental status, and pulse >109 per minute. The algorithm classified patients as the Low Risk category when none of the three risk factors presented on admission. It classified patients in the Intermediate I category when any one of the three risk factors presented. It further classified patients in the Intermediate II category when any of the two risk factors presented. Finally, it classified patients in the High Risk category when all three risk factors presented. The prevalence was 51.6%, 39.7%, 7.9%, and .8% for the Low, Intermediate I, Intermediate II, and High Risk categories respectively. The corresponding mortality was 1.0%, 2.7%, 8.2%, and 17.6% respectively.
CONCLUSION: Three risk factors assessing renal, cardiac, and mental status can differentiate mortality risks among COPD patients on admission.
CLINICAL IMPLICATIONS: This risk classification rule is developed using a large clinical database. It is easy to remember and can be used at bed side for rapid risk assessment and stratification.
DISCLOSURE: Xiaowu Sun, No Financial Disclosure Information; No Product/Research Disclosure Information