Effect of Concurrent Computerized Documentation of Comorbid Conditions on the Risk of Mortality Index

Objective: To evaluate the effect of adding concurrent computerized documentation of comorbid conditions to standard medical record review on the severity of illness index and risk of mortality index. • Design: Cross-sectional study. • Setting and participants: 377 surgical patients at 2 academic centers undergoing inpatient surgery from September 2004 to January 2005. Standard medical record review was performed by the hospital’s professional medical records coders. Concurrent computerized coding was performed by an anesthesiologist doing preoperative assessment using a software tool. • Measurements: Severity of illness index, risk of mortality index, and number of comorbid conditions. • Results: Concurrent coding averaged 5.3 additional comordid conditions that were not identified by medical record review. 9 of 13 comorbid conditions increased with concurrent coding. Mean severity of illness index and risk of mortality index scores were 1.92 and 1.44 with medical record review and 2.23 and 1.68 for concurrent coding. Concurrent coding increased severity of illness by 16% and risk of mortality by 17%; severity of illness and risk of mortality increased by 1 category in 27% and 23% of patients, respectively. • Conclusion: Concurrent coding significantly increased the severity of illness index and risk of mortality index, which can have important implications for health outcomes research, perceived quality of care, and financial reimbursement. Numerous studies have indicated shortcomings in the quality of the U.S. health care system [1]. Concern about these deficiencies have led to numerous quality improvement initiatives, including widespread use of quality measures and the public reporting of quality data. A primary goal of quality measurement is to increase the number of patients exposed to high-quality providers [2,3], either by converting low-quality providers into highquality providers or by moving patients away from providers of low quality to providers of high quality. Yet, measuring quality of care is an imprecise and evolving science. While most hospitals want to measure health outcomes such as mortality, the challenges of measuring these outcomes are significant [4]. Although studies have raised concern about the validity of mortality as an outcome measure, its use continues [5,6]. As with all outcome measures, hospital mortality rates may be affected by random or systematic error. One systematic error in measuring hospital mortality results from inadequately estimating a patient’s risk for death. A common method to account for patient differences in risk is to use risk adjustment [7]. The risk of mortality index is a relatively new method of risk adjustment that is part of the All Patient Refined Diagnosis-Related Group (APR-DRG) system (3M, St. Paul, MN) [8]. The APR-DRG expands the basic CMS-DRG structure by adding subclasses to each DRG that describe patient differences relating to risk of mortality and severity of illness. The methodology yields 2 subcomponent scores: the risk of mortality index and the severity of illness index. Hospital payment rates for inpatient care are based on DRGs. Relative weights are assigned to each DRG to determine the reimbursement rate. These weights are affected by the patient severity of illness index score. Because the severity of illness index score influences hospital reimbursement, hospitals have an incentive to code more thoroughly and accurately. A primary strategy has been the implementation of tools to help better identify comorbid conditions. However, unless such tools are uniformly applied by all hospitals, originAl reseArch From the Department of Anesthesiology and Critical Care Medicine (Drs. Stonemetz, Ulatowski, and Pronovost) and the Department of Emergency Medicine (Dr. Pham), Johns Hopkins University School of Medicine, Baltimore, MD; and Oschner Health System, New Orleans, LA (Dr. Marino).