Healthcare-associated bloodstream infection trends under a provincial surveillance program.

Abstract

BACTOT, Quebec's healthcare-associated bloodstream infection (HABSI) surveillance program has been operating since 2007. In this study, we evaluated the changes in HABSI rates across 10 years of BACTOT surveillance under a Bayesian framework. A retrospective, cohort study of eligible hospitals having participated in BACTOT for at least 3 years, regardless of their entry date. Multilevel Poisson regressions were fitted independently for cases of HABSI, catheter-associated bloodstream infections (CA-BSIs), non-catheter-associated primary BSIs (NCA-BSIs), and BSIs secondary to urinary tract infections (BSI-UTIs) as the outcome and log of patient days as the offset. The log of the mean Poisson rate was decomposed as the sum of a surveillance year effect, period effect, and hospital effect. The main estimate of interest was the cohort-level rate in years 2-10 of surveillance relative to year 1. Overall, 17,479 cases and 33,029,870 patient days were recorded for the cohort of 77 hospitals. The pooled 10-year HABSI rate was 5.20 per 10,000 patient days (95% CI, 5.12-5.28). For HABSI, CA-BSI, and BSI-UTI, there was no difference between the estimated posterior rates of years 2-10 compared to year 1. The posterior means of the NCA-BSI rate ratios increased from the seventh year until the tenth year, when the rate was 29% (95% confidence interval, 1%-89%) higher than the first year rate. HABSI rates and those of the most frequent subtypes remained stable over the surveillance period. To achieve reductions in incidence, we recommend that more effort be expended in active interventions against HABSI alongside surveillance.