Methods: A case-control study was conducted comparing adults with infections caused by carbapenem-non-susceptible PA to patients with carbapenem-susceptible PA at two large academic medical centers in the U.S from 2008-2013. Descriptive statistics, including medians and interquartile ranges and frequencies and per cents, were used to summarize patient characteristics. The probability of having a non-susceptible infection was modeled using mixed effects logistic regression to account for clustering within study site, which was treated as a random effect. Fixed effects included age, Charlson Comorbidity Index score, gender, year of diagnosis, origin of infection (healthcare-associated (HAI) vs. not), infection source, history of previous positive cultures, carbapenem use in the prior 30 days, and specimen collection in the ICU.
Results: A total of 1138 isolates were identified in 1101 unique subjects. Infection types included bloodstream (10%), respiratory (24%), urine (63%), intra-abdominal (IAI) (3%). On average, cases were younger (56yo vs 64yo), male (61% vs 57%) and more likely to have had prior exposure to a carbapenem within the prior 30 days (35% vs 13%) relative to controls. There were no differences in underlying comorbidity burden or proportion of HAIs. In the final multivariate model, independent predictors of carbapenem-non-susceptible PA infection included age, OR (p value) 0.98 (p<0.001) per year, IAI 7.18 (p<0.001), respiratory infection 2.54 (p=0.007) and prior carbapenem use 2.89 (p<0.001).
Conclusion: Our analysis suggests that patients with IAI and respiratory infections, and carbapenem use in the prior 30 days are at increased risk of carbapenem-non-susceptible PA infection as compared to carbapenem-susceptible PA infection. Our findings may assist in identifying these patients, allowing for adjustment of treatment strategy. However, further inquiry is warranted to substantiate these findings and help guide empiric treatment decisions in hospitalizations.
K. Holt, None
C. Hoffmann, None
G. Dumyati, None
J. Brown, None
See more of: Poster Abstract Session