Risk factors for colistin-resistant Enterobacteriaceae in a low-endemicity setting for carbapenem resistance - a matched case-control study.

Background: Emergence of colistin resistance has been related to increased use in clinical settings, following global spread of carbapenem-resistant Gram-negative bacteria. Use of colistin in animal production may constitute a further source of spread of resistant strains to humans. We sought to determine risk factors for human colonisation or infection with colistin-resistant Escherichia coli and Klebsiella pneumoniae in a setting where colistin is mainly used for animal production. Methods: This retrospective matched case-control study was performed during a 5-year period at two university-affiliated hospitals in Basel, Switzerland. Conditional univariable logistic regression was used to calculate odds ratios (OR) for colistin resistance. All variables found to be significant in univariable analyses were included in the conditional multivariable regression model using stepwise forward and backward selection. Results: Forty-two cases (33 with colistin-resistant E. coli, 9 with colistin-resistant K. pneumoniae) and 126 matched controls were identified. Baseline characteristics, comorbidities, prior exposure to antibiotics and healthcare settings did not differ between cases and controls, except for prior exposure to carbapenems, hospitalisation and stay abroad during the prior 3 months. In multivariable analyses, only prior exposure to carbapenems remained associated with colistin resistance (OR: 5.00; 95% confidence interval (95% CI): 1.19-20.92; p = 0.028). Conclusion: In a low-endemicity setting for carbapenem resistance, prior exposure to carbapenems was the only risk factor for colonisation or infection with colistin-resistant E. coli or K. pneumoniae. Prior exposure to colistin was not significantly associated with detection of colistin resistance, which mainly occurred in the absence of concurrent carbapenem resistance.