Physician Variation in Time to Antimicrobial Treatment for Septic Patients Presenting to the Emergency Department

Abstract

Objectives: Delayed initiation of appropriate antimicrobials is linked to higher sepsis mortality. We investigated interphysician variation in septic patients' door-to-antimicrobial time.

Design: Retrospective cohort study.

Setting: Emergency department of an academic medical center.

Subjects: Adult patients treated with antimicrobials in the emergency department between 2009 and 2015 for fluid-refractory severe sepsis or septic shock. Patients who were transferred, received antimicrobials prior to emergency department arrival, or were treated by an attending physician who cared for less than five study patients were excluded.

Interventions: None.

Measurements and main results: We employed multivariable linear regression to evaluate the association between treating attending physician and door-to-antimicrobial time after adjustment for illness severity (Acute Physiology and Chronic Health Evaluation II score), patient age, prehospital or arrival hypotension, admission from a long-term care facility, mode of arrival, weekend or nighttime admission, source of infection, and trainee involvement in care. Among 421 eligible patients, 74% received antimicrobials within 3 hours of emergency department arrival. After covariate adjustment, attending physicians' (n = 40) median door-to-antimicrobial times varied significantly, ranging from 71 to 359 minutes (p = 0.002). The percentage of each physician's patients whose antimicrobials began within 3 hours of emergency department arrival ranged from 0% to 100%. Overall, 12% of variability in antimicrobial timing was explained by the attending physician compared with 4% attributable to illness severity as measured by the Acute Physiology and Chronic Health Evaluation II score (p < 0.001). Some but not all physicians started antimicrobials later for patients who were normotensive on presentation (p = 0.017) or who had a source of infection other than pneumonia (p = 0.006). The adjusted odds of in-hospital mortality increased by 20% for each 1 hour increase in door-to-antimicrobial time (p = 0.046).

Conclusions: Among patients with severe sepsis or septic shock receiving antimicrobials in the emergency department, door-to-antimicrobial times varied five-fold among treating physicians. Given the association between antimicrobial delay and mortality, interventions to reduce physician variation in antimicrobial initiation are likely indicated.

Figure 1

Adjusted time from emergency department admission to antimicrobial initiation by physician. The solid horizontal line indicates the physician’s adjusted median time to antimicrobials, bars represent the interquartile range, and vertical lines represent the total range. The horizontal dotted line is the adjusted median door-to-antimicrobial time for the overall cohort.

Figure 2

Adjusted percentage of physicians’ patients receiving antimicrobials within 1, 2, or 3 hours of emergency department (ED) arrival. Symbols sizes are proportional to the number of eligible patients seen by each physician.

Figure 3

Adjusted time from emergency department admission to antimicrobial initiation by physician and presence or absence of (A) hypotension on or prior to emergency department arrival (13 physicians, 209 patients) or (B) pneumonia versus other source of infection (8 physicians, 144 patients). Physicians were rank ordered by the median door-to-antimicrobial time (solid horizontal line) for patients without hypotension or pneumonia, respectively. Bars represent the interquartile range, and vertical lines represent the total range. The horizontal dotted line is the adjusted median door-to-antimicrobial time for the overall cohort. The influence of hypotension (p=0.017 for interaction) and pneumonia (p=0.006 for interaction) varied significantly by physician.