What do case-control studies tell us about risk factors for antibiotic resistance? Christie Y. Jeon EPI502 Jan 22 nd, 2008
Outline Motivation Randomized Controlled Trial Understanding the outcomes OR S, OR C Variation of effect Confounding Multiplicative vs. Absolute
Motivation To identify the risk factors for infection with resistant pathogen Antibiotic use for a given infection A A R, B A R Who, where, what ? Why? Interventions Restrictions
Randomized Controlled Trial PR S = P[R] TX > 1 P[R] NoTx PR C = P[R I C] TX > 1 P[R I C] NoTx PR S < PR C
Case-control studies AB CaseColonized with resistant ControlsColonized with susceptible Colonized with susceptible OR not colonized Measure of Association OR Conditional OR Simple Example Interpretation of OR
Example – Harris et al. CID 2002 AB CaseCulture positive for imipenem-resistant P. aeruginosa ControlsCulture positive for imipenem-susceptible P. aeruginosa Randomly selected control patients from same medical services Interpretation of OR The increase in an individual’s risk of carrying imp R P. aeruginosa due to AB given an infection with susceptible P. aeruginosa(?) The increase in an individual’s risk of carrying imp R P. aeruginosa due to the particular AB Prediction : Given that an individual has recently been treated with a particular AB, the increased likelihood that the pathogen is imp R compared to not having been treated
Example – Harris et al. CID 2002 AB CaseCulture positive for imipenem-resistant P. aeruginosa ControlsCulture positive for imipenem-susceptible P. aeruginosa Randomly selected control patients from same medical services Interpretation of OR The increase in an individual’s risk of carrying imp R P. aeruginosa due to AB given an infection with susceptible P. aeruginosa(?) The increase in an individual’s risk of carrying imp R P. aeruginosa due to the particular AB Community effect : indirect effect of AB on the population as a whole by reducing the overall exposure to susceptible P. aeruginosa
Appropriate case-control design Q : “What are the risk factors for emergence of AB-resistance in pathogen X among patients previously infected with AB-susceptible pathogen X” A : Cases : individuals with resistant organism who previously had the susceptible form of the organism Controls : individuals with the susceptible pathogen and no subsequent resistant organism
Example – Harris et al. CID 2002
B A R PR c B < PR c A PR c B = P[R I C] TX > 1 P[R I C] NoTx
Tx clears some R-pathogen PR S = P[R] TX < 1 P[R] NoTx PR C = P[R I C] TX >1 P[R I C] NoTx PR S < PR C
Outbreak PR S = P[R] TX > 1 P[R] NoTx PR C = P[R I C] TX >> 1 P[R I C] NoTx PR S < PR C
Confounding Time at risk of resistant pathogen Duration at risk of resistant pathogen Comorbid illness Location of recruitment
Multiplicative vs. Absolute Odds, Rate, Risk Ratio = multiplicative measure Prediction Causal association Rate Difference = absolute measure Burden of disease Population effect
Schwaber et al AB C D E
Recovering rates from case- control studies Rate TX = # cases TX controls TX x 3 Rate NoTx = # cases No Tx controls No Tx x 3
Summary Strength and meaning of the association of antibiotic use and resistance varies by Definition of the controls Effect on the susceptible strains The nature of the antibiotic Background incidence of resistance strains Absolute measures are better measure of burden of resistance in the population