Hospital-acquired, laboratory-confirmed bloodstream infections: linking national surveillance data to clinical and financial hospital data to estimate increased length of stay and healthcare costs.

This matched cohort study estimates the effect of hospital-acquired bloodstream infection (HA-BSI) on length of stay (LOS) and costs during hospitalisation of 1839 patients (age range 80 years) gathered from 19 acute hospitals in Belgium. A second objective was to evaluate the impact of the choice of matching criteria. Data from national surveillance of HA-BSI were linked to hospital administrative discharge data, with respect for the patients' right to confidentiality of their health record. Controls were identified based on a set of matching factors: hospital, All-Patient Refined Diagnosis Related Groups, age, principal diagnosis, Charlson Comorbidity Index and time to infection. The results showed that, depending on the choice of matching factors, the estimation of additional LOS decreased from 26 to 10 days, with the most critical factor being the time to infection. The additional LOS attributable to HA-BSI was 9.9 days [95% confidence interval (CI): 7.8, 11.9]. The additional cost per infection was €4900 [95% CI: €4035, €5750]; 58% of those costs were due to LOS, 10% were due to antibiotics, 10% due to other pharmaceutical products and 15% were due to billed medical acts. The main conclusion is that laboratory-confirmed HA-BSIs increase the LOS by 10 days for patients surviving the infection, and that the time to infection plays a crucial role in this estimation.