Defining reference range for blood culture rates – an analysis within the German hospital infection surveillance system (KISS)

Tuesday, 19 August 2014
Exhibit hall (Dena'ina Center)
André Karch , Helmholtz Centre for Infection Research, Braunschweig, Germany
Stefanie Castell , Helmholtz Centre for Infection Research, Braunschweig, Germany
Christine Geffers , Charité University Medicine Berlin, Berlin, Germany
Frank Schwab , Charité University Medicine Berlin, Berlin, Germany
Frank M Brunkhorst , Jena University Hospital, Jena, Germany
Petra Gastmeier , Charité University Medicine Berlin, Berlin, Germany
Rafael T Mikolajczyk , Helmholtz Centre for Infection Research, Braunschweig, Germany
INTRODUCTION:  

Early and appropriate blood culture diagnostics is associated with both, a reduction of sepsis mortality and a reduction of duration of antibiotic treatment. To date, 100-200 blood culture sets per 1000 patient-days are recommended as the target range. However, the empirical basis of this recommendation is not clear. A previous study observed lower sepsis rates in institutions with low blood culture rates, suggesting that some sepsis cases remain undetected. The aim of this study was to analyze the form of the association between blood culture rates and sepsis rates and to derive a reference range for blood culture rates in Intensive Care Units (ICUs).  

METHODS:

The present study is based on data from 223 ICUs taking part in the German hospital infection surveillance system, which captures all cases of primary sepsis in the participating institutions. We applied locally weighted regression in order to assess the form of association between blood culture rates and sepsis rates. Subsequently, we applied segmented Poisson regression to estimate the detection threshold.

RESULTS:  

Studied by means of nonparametric additive models, sepsis rates increased when the number of blood cultures was below 80-90 per 1000 patient days, but stabilized for higher blood culture rates. A threshold of 87 (95% confidence interval 54-120) per 1000 patient days was identified. Sixty-five percent of the ICUs participating in this study showed blood culture rates below this threshold. These ICUs were more likely to represent interdisciplinary ICUs (p=0.048) in smaller (p<0.001), non-academic hospitals (p<0.001) with a remote microbiology lab (p<0.001), a high proportion of short stay patients (p=0.003) and showed lower ventilation (p<0.001) and CVC rates (p=0.006).

CONCLUSIONS:  

We provided empirical justification for a target range for blood culture rates in ICUs. In a large majority of the studied ICUs, blood culture rates were far below this threshold.