PURPOSE: Epidemiological and outcomes studies often rely on ICD codes to identify patients with specific diseases. In Oklahoma City we observed a fourfold variation in the incidence of severe sepsis among hospitals of similar size. To examine the hypothesis that these differences were due to variability in coding, we compared the incidence of sepsis diagnosed by critical care physicians to the discharge diagnoses assigned by the hospital coders.
METHODS: With IRB approval all patients admitted to the MICU at OU Medical Center between October 2009 and March 2010 were screened and those with suspected sepsis were enrolled if informed consent could be obtained. The diagnoses of no sepsis, sepsis, severe sepsis and septic shock were established by critical care physicians based on ACCP/SCCM consensus definitions and compared to the discharge diagnoses determined by hospital coders based on retrospective review of the medical record.
RESULTS: A total of 85 patients were enrolled, of which 71(84%) met consensus criteria for sepsis including 39 with septic shock and 27 with severe sepsis. The discharge diagnoses by the coders identified only 39/71 (55%) septic patients, including 15/39 (38%) with septic shock and 25/27 (93%) with severe sepsis. Mortality was 13/39 (33%) in patients coded as septic but only 5/32 (16%) in septic patients given alternative diagnoses for a true overall mortality of 25%. Pneumonia was the most common missed cause of sepsis (22/32, 69%). All 21 patients with a physician dictated discharge diagnosis of sepsis were correctly coded in contrast to only 18/50 (36%) patients without a diagnosis of sepsis in the discharge summary.
CONCLUSION: A significant number of patients with sepsis were not identified by hospital coders, resulting in a misleading picture regarding outcomes. Lack of proper documentation by physicians contributes to this discrepancy.
CLINICAL IMPLICATIONS: Epidemiological and outcome studies performed utilizing coding data may not have a representative patient population.
DISCLOSURE: Ashley Ellis, No Financial Disclosure Information; No Product/Research Disclosure Information