Antimicrobial Exposure and the Risk of Delirium in Critically Ill Patients

Jessica J. Grahl; Joanna L. Stollings; Shayan Rakhit; Anna K. Person; Li Wang; Jennifer L. Thompson; Pratik P. Pandharipande; E. Wesley Ely; Mayur B. Patel


Crit Care. 2018;22(337) 

In This Article


Of the 521 patients enrolled at a single center within the BRAIN-ICU Study, 418 patients met our inclusion criteria. Exclusion criteria and exposure to antimicrobials in this ICU cohort with delirium assessments are outlined in Figure 1. Baseline demographics are presented in Table 1. Patients had a median age of 58 (IQR 47–68), and 171 patients (41% of the cohort) were admitted to the surgical ICU. The population was predominantly mechanically ventilated at enrollment (n= 350, 84%), median modified SOFA score was 7 (IQR 5–9), and Charlson Comorbidity Index was 2 (IQR 1–4). Many patients received benzodiazepines (n = 309, 74%) and opiates (n = 367, 88%). Delirium occurred in 308 (74%) patients during their ICU stay with a median duration of delirium of 3 (IQR 2–6) days. Antimicrobial exposure to any agent was common in around three fourths of subjects (n = 318, 76%), including the classes of beta-lactams (n = 223, 53%), fluoroquinolones (n = 138, 33%), macrolides (n = 29, 7%), and other antimicrobials (n = 285, 68%). One quarter (n = 100, 24%) of subjects did not receive any antimicrobials in their ICU stay. Patients were exposed to a wide variety of non-mutually exclusive antimicrobial agents, which are presented in Additional file 1: Table S1.

Figure 1.

Cohort eligibility for antimicrobial exposures among critically ill patients with delirium assessments

Among antimicrobial classes, only first- through third-generation cephalosporins demonstrated an association with delirium in the main statistical model (Table 2) (logistic regression with cluster sandwich covariance estimator model OR = 2.2, 95% CI 1.28–3.79, P = 0.004). This was corroborated on the sensitivity analyses restricted to ICU time accounting for daily sepsis and daily mechanical ventilation in Additional file 1: Table S2. Even after we adjusted for ICU type (that is, surgical versus medical) in our main statistical model and sensitivity analysis, our findings remained consistent. The classes of cefepime, penicillins, carbapenems, fluoroquinolones, and macrolides did not have any consistent associations with delirium across our models. After we accounted for total antimicrobial exposure in the second statistical model (Table 3), only the other antimicrobial class demonstrated an association between delirium and total days of exposure (proportional odds logistic regression model, OR = 3.14, 95% CI 2.27–4.35, P<0.001). The beta-lactams, fluoroquinolones, and macrolides did not have an association between total antimicrobial exposure and days of delirium.

The most consistent remaining risk factors (main and sensitivity analyses) ranked from strongest to weakest were as follows: delirium on previous day, mechanical ventilation, sepsis, and age (Table 2 and Additional file 1: Table S2). Although other antimicrobial agents, most analgesic and sedative agents, and antipsychotics were associated with risk of delirium, the association was not significant after we adjusted for daily sepsis and daily mechanical ventilation in the ICU model (Additional file 1: Table S2).