ObjectivesThe increasing incidence of carbapenem-nonsusceptible Enterobacterales as major pathogens in healthcare associated infections (HAIs) is of paramount concern. To implement effective prevention strategies against carbapenem-nonsusceptible Enterobacterales (CnSE) HAIs, it is crucial to identify modifiable factors associated with these infections. We identified risk factors for CnSE-HAIs, and compared clinical outcomes of CnSE-HAI and carbapenem-sensitive Enterobacterales (CSE)-HAI patients.MethodsWe conducted a multi-centre parallel matched case-control study in two 1700-bedded Singapore acute-care hospitals from 2014–2016. Patients with CnSE-HAIs and CSE-HAIs were compared to a common control group without HAIs (1:1:3 ratio), matched by time-at-risk and patient ward. Carbapenem nonsusceptible was defined as non-susceptibility to either meropenem or imipenem. Presence of healthcare associated infections were defined by the criteria provided by the European Centre for Disease Prevention and Control. Outcomes of CnSE-HAI and CSE-HAI patients were compared using multivariable logistic and cox regression; the models were adjusted for infection and treatment characteristics.ResultsEighty CnSE-HAI and 80 CSE-HAI patients were matched to 240 patients without HAIs. All CRE-HAIs patients had prior antibiotic exposure, with 44 (55.0%) with prior carbapenem exposure. The most common CnSE-HAIs were intra-abdominal infections (28.8%) and pneumonia (23.8%). The most common CnSE species was Klebsiella spp. (63.8%). In the risk factor analysis, presence of drainage devices [adjusted odds ratio (aOR), 2.19; 95% CI, 1.29 – 3.70] and prior carbapenem exposure (aOR,17.09; 95% CI, 3.06 – 95.43) independently predicted CnSE-HAIs. In the crude outcomes analysis, CnSE-HAI patients had higher all-cause in-hospital mortality and longer time to discharge compared to CSE-HAI patients. After adjusting for differences in receipt of antibiotics with reported susceptibility to the Enterobacterales, there was no significant difference in all-cause in-hospital mortality between the two groups (aOR, 1.76; 95% CI, 0.86–3.58). Time to discharge remained significantly longer in patients with CnSE-HAI (adjusted hazard ratio, 0.71; 95% CI, 0.51 – 0.98) after adjusting for disease severity, receipt of antibiotics with reported susceptibility and receipt of appropriate source control.ConclusionAppropriate management of deep-seated Enterobacterales infections and reducing exposure to carbapenems may reduce risk of CnSE-HAIs in Singapore. Efforts to improve antimicrobial therapy in CnSE-HAI patients may improve patient outcomes.

Carbapenem-resistant Klebsiella pneumoniae (CRKP) is becoming increasingly problematic due to the limited effectiveness of new antimicrobials or other factors such as treatment cost. Thus, combination therapy remains a suitable treatment option. We aimed to evaluate the in vitro bactericidal activity of various antibiotic combinations against CRKP with different carbapenemase genotypes and sequence types (STs). Thirty-seven CRKP with various STs and carbapenemases were exposed to 11 antibiotic combinations (polymyxin B or tigecycline in combination with β-lactams including aztreonam, cefepime, piperacillin/tazobactam, doripenem, meropenem, and polymyxin B with tigecycline) in static time-kill studies (TKS) using clinically achievable concentrations. Out of the 407 isolate-combination pairs, only 146 (35.8%) were bactericidal (≥3 log10CFU/mL decrease from initial inoculum). Polymyxin B in combination with doripenem, meropenem, or cefepime was the most active, each demonstrating bactericidal activity in 27, 24, and 24 out of 37 isolates, respectively. Tigecycline in combination with β-lactams was rarely bactericidal. Aside from the lower frequency of bactericidal activity in the dual-carbapenemase producers, there was no apparent difference in combination activity among the strains with other carbapenemase types. In addition, bactericidal combinations were varied even in strains with similar STs, carbapenemases, and other genomic characteristics. Our findings demonstrate that the bactericidal activity of antibiotic combinations is highly strain-specific likely owing to the complex interplay of carbapenem-resistance mechanisms, i.e., carbapenemase genotype alone cannot predict in vitro bactericidal activity. The availability of WGS information can help rationalize the activity of certain combinations. Further studies should explore the use of genomic markers with phenotypic information to predict combination activity.