Patient factors influencing the quality of blood culture collection in the adult emergency department

Dr. James Hughes1,2, Ms Angela Hills1, Mrs Mercedes Ray1, Ms Melisa Dixon1, Ms Jill Duncan5, A/Prof Jaimi Greenslande1,2, Dr Jacqueline Harper3, Dr Julian Williams1,4

1Royal Brisbane and Women’s Hospital, Herston, Australia, 2Queensland University of Technology, Kelvin Grove, Australia, 3Pathology Queensland, Herston, Australia, 4University of Queensland, St Lucia, Australia, 5Sunshine Coast University Hospital, Kawana , Australia

Angela Hills is a registered nurse working on the front line of health care at the Royal Brisbane and Women’s Emergency and Trauma Center. Angela places great emphasis on efficient and evidence-based care and became interested in clinical research with the hopes of improving practices in the emergency setting. Her interests include the treatment and diagnosis of sepsis and has recently joined a team looking into the quality of blood cultures in the emergency department.

Objective: To quantify suboptimal blood culture (BC) collection in the emergency department (ED) in terms of contaminated samples and single BC sets. The study will also identify patient factors that influence sub optimal collection practices. quality.

Design: This study used retrospective cohort design of all BCs collected during 2017 in a large, inner-city, adult only ED. Data from patients less than 16 years old, or those who had paediatric BCs collected were excluded. Routinely collected data were extracted from a number of administrative databases. Patient and emergency department factors that influence the proportion of contamination cultures and the proportion of single sets of BC collected were modelled using binary logistic regression.

Results: In 2017 there were 3868 sets of BCs collected on 2534 patients in the ED. Contamination was present in 2.6% (95%CI 2.1%, 3.1%) of all BCs and 55.0% (95%CI 53.1%, 57.0%) of all patients had only a single set of BC collected. The odds of contamination were higher for patients arriving by ambulance (OR 2.3, 95%CI 1.4 – 3.8, p=0.001), lower for patients with triage score of two to five (OR 0.2 – 0.4) compared to category one patients, and higher patients diagnosed with pneumonia (OR1.8, 95%CI 1.1 – 3.2, p=0.03).

Patient sex (Female: OR0.8 95%CI 0.7 – 0.9, p=0.008) , shift of arrival (Early; OR 0.7, 95%CI 0.5 – 0.9, p=0.002), month of arrival (February; OR0.5, 95%CI 0.4 – 0.8, p=0.003), triage acuity, discharge destination and diagnosis explained 13.7% of variation of the patients that had single sets of BC collected.

Conclusions:  Patient factors such as triage category, severity of illness, diagnosis and mode of arrival all impact BC collection, however their influence is relatively small. Collection techniques, individual clinician practice, and departmental culture are likely to be stronger predictors of suboptimal BC collection.


Mercedes is an experienced emergency nurse, having worked in senior positions in two of Queensland’s busiest hospitals. She is a passionate advocate for the improvement of patient care and has recently taken up the role of Clinical Nurse – Research within the Emergency and Trauma Centre at RBWH to work on the Bugs in Bottles project, a multidisciplinary research project aimed at improving the quality and use of blood cultures in the emergency department.