1076. Risk-Adjustment for Ventilator-Associated Events Using Electronic Medical Record Data
Session: Poster Abstract Session: Surveillance of HAIs: Evaluating National Strategy
Friday, October 4, 2013
Room: The Moscone Center: Poster Hall C
Background: The National Healthcare Safety Network (NHSN) recently released new surveillance definitions for ventilator-associated events designed to make surveillance more efficient and objective.  Regulators may consider using these new definitions to benchmark care between hospitals. However, meaningful inter-hospital comparison requires adjustment for underlying risk.  Existing risk-adjustment models tend to either be very simplistic or require extensive manual data collection. Using only predictors commonly available in electronic health records, we aimed to develop an efficient risk-adjustment model for severity of underlying disease that could be of potential use in NHSN.

Methods: Retrospective cohort study of all patients receiving mechanical ventilation in an academic medical center (2006 – 2011). Data were extracted from electronic health records for the one day-window surrounding initiation of mechanical ventilation. Logistic regression models were developed to predict in-hospital and 30-day mortality using combinations of patient demographic information, ventilator data, pharmacy dispensing, administrative data, laboratory and blood culture results. Internal validation was performed by bootstrapping.

Results: 17049 episodes of mechanical ventilation in unique patients were included for analysis. 2604 (15.3%) patients died in hospital and 2791 (16.4%) within 30 days of mechanical ventilation. A model comprising only age, surgical status, 8 common laboratory tests (hematocrit, white blood count, platelets, creatinine, blood urea nitrogen, INR, sodium and glucose), blood culture data and medication use (antimicrobials, vasopressors, amiodarone, vitamin K, steroids and anticonvulsants) achieved an area under the ROC curve of 0.87 (95% CI 0.87 – 0.88)  for both in-hospital and 30-day mortality.

Conclusion: Precise risk-adjustment in ventilated patients using only data routinely stored in electronic health records is feasible. Implementing such a model in NHSN will allow for more valid interpretation of complication rates across hospitals. Future validation studies are needed to assess generalizability to other settings.

Maaike S.M. van Mourik, MD, MSc1,2, Michael V. Murphy, BA2, Michael Klompas, MD, MPH, FRCPC, FIDSA2,3 and For the CDC Prevention Epicenters, (1)Department of Medical Microbiology, University Medical Center Utrecht, Utrecht, Netherlands, (2)Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA, (3)Department of Medicine, Brigham and Women's Hospital, Boston, MA


M. S. M. van Mourik, None

M. V. Murphy, None

M. Klompas, None

<< Previous Abstract | Next Abstract

Findings in the abstracts are embargoed until 12:01 a.m. PST, Oct. 2nd with the exception of research findings presented at the IDWeek press conferences.