USE OF AN ELECTRONIC HEALTH RECORD (EHR) DATABASE TO IDENTIFY HOSPITAL-ACQUIRED PNEUMONIA (HAP)
Author(s)
Campbell RS1, Spoeri RK1, Harris AD2, Kothari S3, Fiske S1, Hays HD1, Emons MF11Cerner LifeSciences, Beverly Hills, CA, USA, 2University of Maryland, Baltimore, MD, USA, 3Astellas Pharma US, Deerfield, IL, USA
Objective: HAP confers significant mortality, morbidity and cost. Observational studies using real-world data have been limited. We sought to develop a methodology to identify and characterize patients with HAP in a large multihospital database without reliance upon radiographs or reviewer designation of infection origin. Methods: We conducted a retrospective analysis (2005-2008) of the Health Facts® database (Cerner Corp., Kansas City, MO), which contains comprehensive clinical records from 115 US hospitals. Adult patients with a secondary discharge diagnosis consistent with pneumonia and a positive respiratory or blood culture obtained >48 hours after admission were included as HAP. Respiratory cultures were defined by site/source and the first positive culture was selected. Qualified blood cultures were used when there was no respiratory culture and no other infection. Hierarchical selection was used when multiple cultures were collected at the same time. Methicillin-resistant- (MRSA) or methicillin-sensitive- Staphylococcus aureus (MSSA) status was designated based on antibiotic susceptibilities. Results: A total of 3,041 patients met criteria for HAP; 434 (14.3%) had an index culture positive for MRSA and 598 (19.9%) for MSSA. Most patients (94%) qualified with a respiratory culture. The cohort was predominantly male (61%) and Caucasian (80%); mean age was 64 years. Primary diagnoses were varied; the three most common were other diseases of lung (9%), acute myocardial infarction (3.8%), and lung cancer (3.8%). MRSA patients were older, with more comorbidities vs MSSA patients. More MSSA isolates were polymicrobial (41%) vs. MRSA (22.5%) and included Gram-negative organisms (33% vs. 19%). Data generated is illustrative of methodology. Conclusion: Our algorithm identifying HAP patients within an EHR dataset potentially allows for larger sample size than other studies using clinical records from a single institution. Chart review is crucial to validate our approach and further work is needed to compare this definition with National Healthcare Safety Network definition.
Conference/Value in Health Info
2010-05, ISPOR 2010, Atlanta, GA, USA
Value in Health, Vol. 13, No. 3 (May 2010)
Code
PIN48
Topic
Clinical Outcomes, Methodological & Statistical Research, Real World Data & Information Systems
Topic Subcategory
Clinical Outcomes Assessment, Health & Insurance Records Systems, Modeling and simulation
Disease
Infectious Disease (non-vaccine), Respiratory-Related Disorders