Diagnostic Stewardship for Ventilator Associated Pneumonia

The primary efficacy outcome is change in proportion of unnecessary antibiotic therapy for presumed VAP in the intervention (modified reporting) period compared to control (standard reporting) period.

Antibiotic use for VAP relative to the index respiratory culture will be classified as follows:

• Not on empiric antibiotic(s), new antibiotic started based on index respiratory culture result
• Not on empiric antibiotic(s), no new antibiotics initiated based on culture result
• On empiric antibiotic(s), changed based on index respiratory culture, completed course for VAP
• On empiric antibiotic(s), continued without change following index respiratory culture, completed course for VAP
• On empiric antibiotic(s), antibiotics discontinued based on culture results
• On empiric antibiotics(s), but this was not directed towards VAP Antibiotic use in categories a, c, and d will together be considered "unnecessary antibiotic therapy for VAP treatment".

To understand the impact on antibiotic use in ICU patients, we will measure the duration of antibiotic therapy for study patients, unit-level total antibiotic days of therapy (DOTs) standardized to patient census, and unit-level respiratory antibiotic days of therapy (DOTs) standardized to patient census.

Measurement of this outcome will help understand how modification of reporting would influence the likelihood of a patient receiving a diagnosis of VAP. The clinical diagnosis of VAP is an intermediate outcome that precedes the endpoint of antibiotic use. Because there are no consensus definitions for VAP, we will use the treating clinician's documentation of pneumonia or VAP and treatment for those conditions to capture this outcome. Similarly, clinical diagnosis of tracheobronchitis within 7 days after the index culture will also be recorded.

We will perform a sensitivity analysis of the impact on antibiotic use outcomes after excluding patients diagnosed with tracheobronchitis within 7 days of index culture. The hypothesis is that the intervention will have limited impact on those diagnosed and treated for tracheobronchitis, and thus, we may see a larger difference between intervention and control when excluding patients with diagnoses of tracheobronchitis.

Ventilator-free days is defined as the number of calendar days within 28 days after the index respiratory culture on which the patient was not mechanically ventilated. Any patient dying within 28 days of the index respiratory culture collection will be assigned zero ventilator-free days. Ventilator-free days will be compared between the intervention and control periods for all study ICUs combined, and by hospital.

Frequency of requests for complete reports in the intervention period - these data are applicable in the intervention period only. This will be a primarily descriptive analysis to help us understand if the intervention worked as intended i.e., how frequently did clinicians still want the full report with organism identification despite the nudge to consider asymptomatic colonization.

Death, bacteremia, and septic shock (from any cause) within 7 days of the index culture order will be recorded as potential significant adverse events from a delayed or missed true VAP diagnosis or delay in initiation of appropriate antimicrobial therapy. Two-person adjudication will be used to determine whether the above events were attributed to the intervention, during the intervention period. Rates of these events will be compared between intervention and control periods individually for each event, and as a combined adverse event outcome.

The clinical outcome of a participant will be ranked from most to least desirable outcome, to obtain a 5-level ordinal outcome. We will conduct a series of pairwise comparisons of each intervention patient to each control patient using two rules: When comparing patients with different overall clinical outcomes, the patient with a better overall clinical outcome based on pre-specified ordered ranks is determined to have the better outcome. When comparing two patients who have the same rank based on overall clinical outcome, the patient with a shorter course of antibiotics has the better outcome. The DOOR/RADAR probability represents the probability that a randomly selected participant who received the intervention (modified reporting) has a better overall outcome than a randomly selected participant who received standard reporting (control), accounting for antibiotic duration of therapy.