By Richard R. Watkins, MD, MS, FACP, FIDSA, FISAC

Professor of Internal Medicine, Northeast Ohio Medical University; Division of Infectious Diseases, Cleveland Clinic Akron General, Akron, OH

Dr. Watkins reports no financial relationships relevant to this field of study.

SYNOPSIS: A retrospective cohort study revealed broad-spectrum antibiotics were unnecessarily prescribed to patients with community-onset sepsis and were associated with worse outcomes and higher mortality.

SOURCE: Rhee C, Kadri SS, Dekker JP, et al. Prevalence of antibiotic-resistant pathogens in culture-proven sepsis and outcomes associated with inadequate and broad-spectrum empiric antibiotic use. JAMA Netw Open 2020;3:e202899.

In the management of sepsis, early administration of empiric antibiotic therapy is a critical step. However, overtreatment (i.e., unnecessary broad-spectrum therapy) can produce adverse reactions, lead to potential selection for antibiotic-resistant bacteria, Clostridioides difficile infection (CDI), and cost more. Inappropriately narrow therapy can lead to higher mortality. Rhee et al examined the outcomes associated with overtreatment and undertreatment in patients with culture-positive community-onset sepsis.

The authors conducted a retrospective cohort analysis using data from a diverse group of U.S. hospitals. Included patients were ≥ age 20 years, had received a diagnosis of community-acquired sepsis, and had positive cultures for potentially pathogenic organisms by hospital day 2. Patients were excluded if they had hospital-acquired sepsis or were transferred from long-term care or rehabilitation facilities, hospice, or other hospitals. The authors used ICD-9-CM discharge codes to diagnosis sepsis. The investigators considered patients to have received inadequate empiric therapy if at least one pathogen isolated from any clinical culture was not susceptible to all antibiotics administered during the initial two days of admission. They considered patients to have received unnecessarily broad empiric therapy if they were prescribed anti-methicillin-resistant Staphylococcus aureus (MRSA) antibiotics, anti-vancomycin-resistant enterococci (VRE) antibiotics, anti-Pseudomonas β-lactams or carbapenems but none of the organisms targeted by these antibiotics were recovered. The primary outcome was in-hospital mortality.

The cohort included 17,430 patients. The median age was 69 years, and 55.9% were women. Urinary tract infection was the most common source of sepsis (48.9%), followed by pulmonary (32.9%), intra-abdominal (13.6%), and skin and soft tissue (10.3%) infections. The most frequently identified pathogens were Escherichia coli (33.7%), Staphylococcus aureus (21.3%), Streptococcus spp. (13.5%), Klebsiella spp. (12.9%), and Enterococcus spp. (11.1%). Antibiotic-resistant organisms included MRSA (11.7%) and ceftriaxone-resistant gram-negative organisms (CTX-RO) (13.1%). Of these, 66.3% had P. aeruginosa, VRE (2.1%), extended spectrum beta-lactamases (ESBLs) (0.8%), and carbapenem-resistant Enterobacteriaceae (CRE) (0.5%). The prevalence of at least one resistant gram-positive organism (MRSA or VRE) was 13.6%, while 13.2% of the cohort had at least one resistant gram-negative organism. Patients with resistant organisms were more likely to have a great number of comorbidities, have a pulmonary infection, require vasopressors or mechanical ventilation, be admitted to the intensive care unit, and die in the hospital.

The most commonly prescribed antibiotic was vancomycin, received by 41.7% of patients, followed by a fluoroquinolone (40.1%), piperacillin-tazobactam (33.9%), and ceftriaxone (29.8%). Overall, 11,797 patients received antibiotic therapy directed against the antimicrobial-resistant (AMR) pathogens listed earlier. Of these patients, 3,447 had at least one of these organisms isolated in culture. The empiric antibiotics were active in 12,398 out of 15,183 cases of sepsis in which antibiotic-pathogen susceptibility combinations could be examined. A multivariable analysis determined inadequate therapy was significantly associated with higher mortality (adjusted odds ratio [aOR], 1.19; 95% confidence interval [CI], 1.03-1.37; P = 0.02). This held true after adjusting for baseline characteristics, severity of illness, and adequacy of therapy. Unnecessarily broad antibiotic therapy was associated with higher mortality in patients without shock (aOR, 1.22; 95% CI, 1.06-1.49; P = 0.007), as well as a greater risk for CDI (aOR, 1.26; 95% CI, 1.01-1.57; P = 0.04). There also was a trend toward more acute kidney injury that did not reach statistical significance. Vital sign data during the first two days of hospitalization were reported missing in 47.8% of cases, while lactic acid levels (an important marker for sepsis and septic shock) were missing in 40.9%.


Clinicians are well aware of the mantra to recognize and treat sepsis early. But pragmatically, it is challenging to estimate accurately which patients might have an infection because of an AMR organism. Although risk factors (e.g., prior AMR infection, recent antibiotic use, recent hospitalization, and underlying comorbidities) may be helpful, they are not foolproof. Thus, a trade-off exists between choosing empiric antibiotics with a spectrum that is too broad vs. one that is too narrow. Rhee et al provided some useful information for dealing with this common clinical scenario, yet highlights limitations in the diagnostic process. Most patients with community-onset sepsis do not have AMR pathogens, so treating them with broad-spectrum antibiotics is inappropriate. That said, identifying the small fraction who do is limited by the widespread lack of rapid diagnostic testing. Better prediction models to determine the risk of an AMR infection that is assessed in many patients with sepsis also could help inform empiric antibiotic decisions.

A major limitation of the study is that in approximately 30% to 50% of sepsis cases, cultures were negative or the exact source of sepsis was not clearly determined. The reliance on culture data in this study almost certainly led to an overestimate of resistant pathogens in the cohort. Another limitation is related to the observation that more severely ill patients received broad-spectrum antibiotics, which, in fact, may have been because of residual confounding in this group related to the wide range of organ dysfunction. Finally, limitations in the data set mean the investigators did not have information on patient allergies, potential complications of antibiotics, or whether patients were hospitalized recently.

Most patients with community-acquired sepsis do not have resistant organisms, but inadequate therapy in those who do leads to higher mortality. This is another reminder that we need faster and more accurate diagnostic tests to optimize the management of patients with sepsis.