Using Probability of Community-Acquired Pneumonia to Tailor Antimicrobials Among Inpatients

The goal of this prospective randomized study is to improve antibiotic use among hospitalized patients with suspected pneumonia. An alert was built into the electronic health record to guide use of diagnostic testing based on probability of bacterial pneumonia. Patients with test results suggesting viral infection will be randomized to either: (1) receive a structured communication from the antimicrobial stewardship team to de-escalate antibiotics or (2) usual care.

Low-risk patients with viral pneumonia do not benefit from and may be harmed by antibiotic therapy. In this study, an alert will appear in the electronic health record of patients undergoing molecular diagnostic testing for respiratory symptoms that provides options for diagnostic testing based on pre-test probability of bacterial infection. Patients with test results suggesting possible viral infection will be randomized to either usual care or to receive test results along with structured guidance from antimicrobial stewardship to consider discontinuing or de-escalating antibiotics. This guidance, which will include an explicit calculation of the post-test probability of bacterial infection based on considering risk factors, vital signs, symptoms, and available imaging, will be communicated to the primary care team via direct electronic message and a summary note in the patient's chart. The final decision on whether to continue antibiotic therapy will be up to the primary team. The primary outcome of interest will be in-hospital antibiotic use. Safety outcomes will include length of stay, readmissions, hospital-free days, and mortality.