In nonwhite patients hospitalized with cirrhosis, biliary involvement, recent health care exposure, and cultures taken more than 48 hours after admission all independently predicted that bacteremia would be caused by multidrug-resistant organisms (MDROs), according to a medical record review at CHI St. Luke’s Medical Center, an 850-bed tertiary care center in Houston.

“These variables along with severity of infection and liver disease may help clinicians identify patients who will benefit most from broader-spectrum empiric antimicrobial therapy,” wrote the investigators, led by Jennifer Addo Smith, PharmD, of St. Luke’s, in the Journal of Clinical Gastroenterology .

But local epidemiology remains important. “Although a gram-positive agent (e.g., vancomycin) and a carbapenem-sparing gram-negative agent (e.g., ceftriaxone, cefepime) are reasonable empiric agents at our center, other centers with different resistance patterns may warrant different empiric therapy. Given the low prevalence of VRE [vancomycin-resistant Enterococcus] in this study … and E. faecium in other studies (4%-7%), an empiric agent active against VRE does not seem to be routinely required,” they said.

The team looked into the issue because there hasn’t been much investigation in the United States of the role of multidrug resistant organisms in bacteremia among patients hospitalized with cirrhosis.

Thirty patients in the study had bacteremia caused by MDROs while 60 had bacteremia from non-MDROs, giving a 33% prevalence of MDRO bacteremia, which was consistent with previous, mostly European studies.

Enterobacteriaceae (43%), Staphylococcus aureus (18%), Streptococcus spp. (11%), Enterococcus spp. (10%), and nonfermenting gram-negative bacilli (6%) were the main causes of bacteremia overall.

Among the 30 MDRO cases, methicillin-resistant S. aureus was isolated in seven (23%); methicillin-resistant coagulase-negative Staphylococci in four (13%); fluoroquinolone-resistant Enterobacteriaceae in nine (30%); extended spectrum beta-lactamase–producing Enterobacteriaceae in three (10%), and VRE in two (7%). No carbapenemase-producing gram-negative bacteria were identified.

The predictors of MDRO bacteremia emerged on multivariate analysis and included biliary cirrhosis (adjusted odds ratio, 11.75; 95% confidence interval, 2.08-66.32); recent health care exposure (aOR, 9.81; 95% CI, 2.15-44.88); blood cultures obtained 48 hours after hospital admission (aOR, 6.02; 95% CI, 1.70-21.40).

Blood cultures past 48 hours and recent health care exposure – generally hospitalization within the past 90 days – were likely surrogates for nosocomial infection. The association with nonwhites, particularly Hispanic patients, has been reported before. “Reasons for this disparity are unclear but could be partially because of more knowledge deficits regarding antibiotics and more self-medication with antibiotics obtained without prescriptions among Hispanic patients,” the authors said.

The link with biliary cirrhosis also is unclear. “Compared with other cirrhotic patients, perhaps patients with PBC [primary biliary cholangitis] have had more cumulative antimicrobial exposure because of [their] higher risk for UTIs [urinary tract infections] and therefore are at increased risk for MDROs,” they wrote.

The median age in the study was 59 years. Half of the patients were white; 46% were women. Hepatitis C was the most common cause of cirrhosis, followed by alcohol.

MDRO was defined in the study as bacteria not susceptible to at least one antibiotic in at least three antimicrobial categories; 90 cirrhosis patients without bacteremia served as controls.

The funding source was not reported. Dr. Addo Smith had no disclosures.

SOURCE: Smith JA et al. J Clin Gastroenterol. 2017 Nov 23. doi: 10.1097/MCG.0000000000000964.