474. Evaluation of MRSA Burden Using Administrative Data and Microbiology Lab Results
Session: Poster Session: Hospital-acquired and Transplant Infections
Friday, October 30, 2009: 12:00 AM
Room: Poster Hall A
Administrative and microbiology data collected over 4 years from an integrated healthcare system were analyzed to evaluate alternative methods to estimate the burden of S. aureus infection in hospital inpatients. One method classified admissions based on an algorithm applied to International Classification of Disease, Ninth Revision Clinical Modification (ICD9) diagnosis codes. Alternatively, admissions were classified by positive cultures obtained during hospitalization. Concordance between ICD9 codes and culture results was examined at 3 levels:
1. presence/absence of S. aureus
2. classification of infection type (bacteremia; pneumonia; skin/soft tissue; other), conditional on S. aureus being present by culture and ICD9 codes (with specimen source used as a proxy for culture infection type)
3. classification of S. aureus as MRSA/MSSA, conditional on presence of S. aureus.
Cramer’s V statistic (CvM) quantified strengths of association. Of 494,604 inpatients discharged between 2004-2007, 6764 admissions had a S. aureus ICD9 code and 7086 admissions had ≥1 S. aureus positive culture. 37% of admissions with a S. aureus ICD9 code had no positive S. aureus culture and 40% of admissions with a positive culture had no ICD9 code (CvM: 0.61). Conditional on the presence of S. aureus by both methods, infection types were concordant in 48% (CvM: 0.42) and susceptibility classification was concordant in 89% of cases (CvM: 0.79). 56% of infections were classified as MRSA by ICD9 code, 54% were classified as MRSA based on culture.
Similar numbers of records were identified through presence of S. aureus infection codes and S. aureus cultures. However, disagreement within records with respect to presence/absence of S. aureus and with respect to infection type was substantial. Yet, when S. aureus was present by both methods, susceptibility classification was usually concordant. Reasons for discordance deserve further study.
R Evans, PhD1, Rachel Gorwitz, MD, MPH2, Michael A. Jhung, MD, MPH, James Lloyd, BS1, Michael Mundorff, MBA/MHSA1, Pat Nechodom, MPH4, Matthew Samore, MD, FSHEA4 and  M. H. Samore, None..
M. A. Jhung, None..
R. S. Evans, None..
M. B. Mundorff, None..
J. F. Lloyd, None..
R. J. Gorwitz, None..
P. Nechodom, None., (1)Intermountain Healthcare, Salt Lake City, UT, (2)CDC, Atlanta, GA, (3)University of Utah School of Medicine, Salt Lake City, UT