Choosing Blood Cultures Wisely

The Sir Charles Gairdner Hospital in Perth looked at their use of blood cultures and developed guidelines for their staff, applying to blood cultures the Choosing Wisely principle of ensuring the right amount of care – not too much and not too little. Pilot data from the WA-based Smart Sepsis project suggested that significant inefficiencies in blood culture practices exist, with both excessive sample collection in some cases and also contamination of specimens and insufficient sample collection in others.

At the Sir Charles Gairdner Hospital, we studied when and how many blood cultures are taken, and developed guidelines and a training program to improve practice and help us Choose Wisely when it comes to blood cultures.

The multidisciplinary project team comprised clinical experts from a range of specialties including Consultants from the Emergency Department, Haematology, Medical Oncology, Infectious Diseases, Junior Medical Officers and Nursing representation from Infection Prevention and Control with support from a Project Manager. The Choosing Blood Cultures Wisely Project experienced delays related to Covid-19 and took around two years to implement.

To inform our guideline, we collected blood culture pathology data from Haematology and Medical Oncology patients. This group was chosen as they have suppressed immune systems, less clinical response to infection and therefore have the highest risk of missed positive blood cultures.

We reviewed the number of blood cultures taken at presentation/ admission (best practice is two “sets” = 4 bottles) and the number of subsequent repeated blood cultures.

Results demonstrated that from June to August 2020, 243 blood cultures were taken from 189 Medical Oncology patients (with 7% testing positive) and 477 blood cultures from 161 Haematology patients (with 4.4% testing positive).

Around one third (52/168) of these patients had only one blood culture taken, which is fewer than the two cultures that is recommended.

On the other hand, ten patients had a combined 100 blood cultures taken, with no positive results after the initial two tests, Indeed, a few patients had up to 15 blood cultures taken in a single admission.

To test whether repeat cultures yielded new diagnoses, we examined all the positive cultures. Only two patients had late positive cultures (after day five), and in both situations this was associated with new physiological deterioration (hypotension, tachycardia).

Using this data as a starting point, the project team reviewed and updated the hospital blood culture guideline. Our particular focus was when and how to take blood cultures and to ensure the guideline was safe in meeting the needs and complexities of all hospital patients.

The key messages are:

Take initial blood cultures if infection is suspected. A fever alone is not predictive of positive blood cultures; therefore, cultures should be taken in patients with a fever and hypotension, increased respiratory rate or altered mental state.

Take two sets of blood cultures (4 bottles, 10mL in each bottle) from two separate sites to reduce the risk of contamination.

Only repeat blood cultures if there is a patient deterioration suggestive of a new infection or non-response to antimicrobial therapy after 72 hours since the last blood culture.

To inform clinicians of the updated guideline, education was targeted at both nurses and junior medical staff, as it is often nurses who will initiate a request for blood cultures when a patient has a fever. A poster was also published and placed in clinical areas. Furthermore, the guideline is readily available on the hospital intranet site.

We have seen a slight decrease in the number of blood cultures taken since the introduction of the program, but it is too early to determine if clinical practice change has been sustained. A formal re-evaluation using the same sample group is currently underway.

We hope that the Choosing Blood Cultures Wisely program will lead to changes in how we do blood cultures in our hospital longer term.