Seasonality of Antibiotic Resistance and Correlation with Antibiotic

Seasonality of Antibiotic Resistance and Correlation with Antibiotic Use Lova Sun CDDEP Summer 2010

Questions • Is antibiotic resistance of different bacteria seasonal? (E coli, MRSA, VRE) – TSN Database • Do prescription levels of antibiotics which might be driving this resistance also show seasonal trends? – IMS Database • Is antibiotic prescription seasonality temporally correlated to seasonality of resistant bugs, perhaps with time lag? – Cross-correlation analysis, time-series regressions, Granger causality

Part I: E. coli Is E. Coli Antibiotic Resistance Seasonal? Resistance to Ampicillin: Winter Peak Resistance to Ciprofloxacin: Winter Peak

Are Prescriptions of these Antibiotics Seasonal? Aminopenicillin Prescriptions: Winter Peak Fluoroquinolone Prescriptions: Winter Peak

Prescription-Resistance Correlations: Detrended Data Ampicillin (1 -month lag Correlation = 0. 7816) Ciprofloxacin (Unlagged Correlation = 0. 4310)

Part II: MRSA Is MRSA a seasonal bug? Not Combined. But…

HA- and CA-MRSA Seasonality NHDS Data TSN Data HA-MRSA: Winter Peak CA-MRSA: Summer Peak

Correlation: Prescriptions and HA-MRSA Prescription Seasonality: Winter Peaks Possible Antibiotic Drivers of HA-MRSA Quinolones: 2 -month lag (Correlation 0. 5810) Macrolides: 1 -month lag (Correlation 0. 6699) Cephalosporins: 1 -month lag (Correlation 0. 6713)

Possible Antibiotic Driver of CA-MRSA Staph Penicillin Prescriptions: Summer Peak 1 -month lag (Correlation 0. 6371)

Part III: Vancomycin-Resistant Enterococci VRE Seasonality and Correlation with HA-MRSA Vancomycin Resistance: Winter Peak VRE and HA-MRSA: Correlation = 0. 6431 No time lag

Conclusions • Both antibiotic resistance and prescriptions are seasonal • After a 1 -2 month lag (presumably the time it takes for antibiotic use to select for resistant strains), antibiotic prescriptions are significantly correlated with (and perhaps drive) resistance • CA- and HA-MRSA have opposite seasonal peaks, with different antibiotics or other factors driving each type • MRSA, an indicator of vancomycin use, is correlated with VRE • Implications for hospital infection control and prescription policies

Acknowledgements • • Prof. Ramanan Laxminarayan Prof. Bryan Grenfell Eili Klein Mike Eber