Background: Intra-abdominal infections (IAIs) represent a serious cause of morbimortality. A full classification, including all facets of IAIs, does not exist. Two classifications are used to subdivide IAIs: uncomplicated or complicated, considering infection extent; and community-acquired, healthcare-associated or hospital-acquired, regarding the place of acquisition. Inadequate antibiotic therapy is associated with treatment failure and increased mortality. This study was designed to determine accuracy of different classifications of IAIs to identify infections by pathogens sensitive to current treatment guidelines helping the selection of the best antibiotic therapy.
Methods: A retrospective cohort study including all adult patients discharged from hospital with a diagnosis of IAI between 1st of January and 31st of October 2016. All variables potentially associated with pre-defined outcomes: infection by a pathogen sensitive to non-pseudomonal cephalosporin or ciprofloxacin plus metronidazole (ATB 1, primary outcome), sensitive to piperacillin-tazobactam (ATB 2) and hospital mortality (secondary outcomes) were studied through logistic regression. Accuracy of the models was assessed by area under receiver operating characteristics (AUROC) curve and calibration was tested using the Hosmer-Lemeshow goodness-of-fit test.
Results: Of 1804 patients screened 154 met inclusion criteria. Sensitivity to ATB 1 was independently associated with male gender (adjusted OR=2.612) and previous invasive procedures in the last year (adjusted OR=0.424) (AUROC curve=0,65). Sensitivity to ATB 2 was independently associated with liver disease (adjusted OR=3.580) and post-operative infections (adjusted OR=2.944) (AUROC curve=0.604). Hospital mortality was independently associated with age≥70 (adjusted OR=4.677), solid tumour (adjusted OR=3.127) and sensitivity to non-pseudomonal cephalosporin or ciprofloxacin plus metronidazole (adjusted OR=0.368).
The accuracy of pre-existing classifications to identify infection by a pathogen sensitive to ATB 1 was 0.59 considering place of acquisition, 0.61 infection extent and 0.57 local of infection, for ATB 2 it was 0.66, 0.50 and 0.57, respectively.
Conclusion: None of existing classifications had a good discriminating power to identify IAIs caused by pathogens sensitive to current antibiotic treatment recommendations. A new classification, including patients’ individual characteristics like those included in the current model, might have a higher potential to distinguish IAIs by resistant pathogens allowing a better choice of empiric antibiotic therapy.
Keywords: intra-abdominal infections; classification; antibiotic therapy; hospital mortality.