Prevalence of Antibiotic-Resistant Pathogens in Culture-Proven Sepsis and Outcomes Associated With Inadequate and Broad-Spectrum Empiric Antibiotic Use | Critical Care Medicine | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Study Cohort Flowchart
Study Cohort Flowchart

ICD-10 indicates International Statistical Classification of Diseases and Related Health Problems, Tenth Revision.

Figure 2.  Prevalence of Pathogens in Culture-Positive Community-Onset Sepsis
Prevalence of Pathogens in Culture-Positive Community-Onset Sepsis

The reported prevalence of each pathogen is relative to 17 430 patients with culture-positive community-onset sepsis in the cohort from any clinical culture site. Only pathogens isolated within the first 2 days of hospitalization were analyzed. The same pathogen isolated from different sites from the same patient was counted as 1 pathogen. Of 2278 patients with ceftriaxone-resistant gram-negative organism (CTX-RO), 1510 (66.3%) had Pseudomonas aeruginosa. CRE indicates carbapenem-resistant Enterobacteriaceae; E coli, Escherichia coli; ESBL, extended-spectrum β-lactamase producing gram-negative organism; MRSA, methicillin-resistant Staphylococcus aureus; and VRE, vancomycin-resistant Enterococcus.

Figure 3.  Proportion of Culture-Positive Sepsis Patients Treated With Broad-Spectrum Antibiotics in Whom Targeted Resistant Organisms Were Subsequently Recovered
Proportion of Culture-Positive Sepsis Patients Treated With Broad-Spectrum Antibiotics in Whom Targeted Resistant Organisms Were Subsequently Recovered

The dark bars indicate the proportion of 17 430 patients with culture-positive sepsis on admission who received empiric antibiotics directed at specific resistant organisms. Anti–methicillin-resistant Staphylococcus aureus (MRSA) antibiotics include vancomycin, linezolid, and daptomycin; anti–ceftriaxone-resistant gram-negative organism (CTX-RO) antibiotics (ie, anti-Pseudomonal β-lactams) include ceftazidime, cefepime, piperacillin-tazobactam, aztreonam, imipenem, meropenem, and doripenem; anti–vancomycin-resistant Enterococcus (VRE) antibiotics include linezolid or daptomycin; and anti–extended-spectrum β-lactamase (ESBL) producing gram-negative organism antibiotics include carbapenems (ie, imipenem, meropenem, doripenem, or ertapenem). The light bars indicate the proportion of patients treated with antibiotics directed at resistant organisms who had that organism recovered from any clinical site within the first 2 days of hospitalization.

Table 1.  Characteristics of 17 430 Patients With Culture-Positive Sepsis
Characteristics of 17 430 Patients With Culture-Positive Sepsis
Table 2.  Outcomes Associated With Inadequate and Unnecessarily Broad Empiric Antibiotic Therapya
Outcomes Associated With Inadequate and Unnecessarily Broad Empiric Antibiotic Therapya
1.
Rhee  C, Jones  TM, Hamad  Y,  et al; Centers for Disease Control and Prevention (CDC) Prevention Epicenters Program.  Prevalence, underlying causes, and preventability of sepsis-associated mortality in US acute care hospitals.   JAMA Netw Open. 2019;2(2):e187571. doi:10.1001/jamanetworkopen.2018.7571PubMedGoogle Scholar
2.
Kumar  A, Roberts  D, Wood  KE,  et al.  Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock.   Crit Care Med. 2006;34(6):1589-1596. doi:10.1097/01.CCM.0000217961.75225.E9PubMedGoogle ScholarCrossref
3.
Kalil  AC, Johnson  DW, Lisco  SJ, Sun  J.  Early goal-directed therapy for sepsis: a novel solution for discordant survival outcomes in clinical trials.   Crit Care Med. 2017;45(4):607-614. doi:10.1097/CCM.0000000000002235PubMedGoogle ScholarCrossref
4.
Seymour  CW, Gesten  F, Prescott  HC,  et al.  Time to treatment and mortality during mandated emergency care for sepsis.   N Engl J Med. 2017;376(23):2235-2244. doi:10.1056/NEJMoa1703058PubMedGoogle ScholarCrossref
5.
Vazquez-Guillamet  C, Scolari  M, Zilberberg  MD, Shorr  AF, Micek  ST, Kollef  M.  Using the number needed to treat to assess appropriate antimicrobial therapy as a determinant of outcome in severe sepsis and septic shock.   Crit Care Med. 2014;42(11):2342-2349. doi:10.1097/CCM.0000000000000516PubMedGoogle ScholarCrossref
6.
Aaronson  EL, Filbin  MR, Brown  DF, Tobin  K, Mort  EA.  New mandated Centers for Medicare & Medicaid Services requirements for sepsis reporting: caution from the field.   J Emerg Med. 2017;52(1):109-116. doi:10.1016/j.jemermed.2016.08.009PubMedGoogle ScholarCrossref
7.
Rhodes  A, Evans  LE, Alhazzani  W,  et al.  Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock: 2016.   Crit Care Med. 2017;45(3):486-552. doi:10.1097/CCM.0000000000002255PubMedGoogle ScholarCrossref
8.
Huttner  A, Harbarth  S, Carlet  J,  et al.  Antimicrobial resistance: a global view from the 2013 World Healthcare-Associated Infections Forum.   Antimicrob Resist Infect Control. 2013;2:31. doi:10.1186/2047-2994-2-31PubMedGoogle ScholarCrossref
9.
Hiensch  R, Poeran  J, Saunders-Hao  P,  et al.  Impact of an electronic sepsis initiative on antibiotic use and health care facility-onset Clostridium difficile infection rates.   Am J Infect Control. 2017;45(10):1091-1100. doi:10.1016/j.ajic.2017.04.005PubMedGoogle ScholarCrossref
10.
Polgreen  PM, Chen  YY, Cavanaugh  JE,  et al.  An outbreak of severe Clostridium difficile-associated disease possibly related to inappropriate antimicrobial therapy for community-acquired pneumonia.   Infect Control Hosp Epidemiol. 2007;28(2):212-214. doi:10.1086/512174PubMedGoogle ScholarCrossref
11.
Seetharaman  S, Wilson  C, Landrum  M,  et al.  Does use of electronic alerts for systemic inflammatory response syndrome (SIRS) to identify patients with sepsis improve mortality?   Am J Med. 2019;132(7):862-868. doi:10.1016/j.amjmed.2019.01.032PubMedGoogle ScholarCrossref
12.
Teshome  BF, Vouri  SM, Hampton  N, Kollef  MH, Micek  ST.  Duration of exposure to antipseudomonal β-lactam antibiotics in the critically ill and development of new resistance.   Pharmacotherapy. 2019;39(3):261-270. doi:10.1002/phar.2201PubMedGoogle ScholarCrossref
13.
Hranjec  T, Rosenberger  LH, Swenson  B,  et al.  Aggressive versus conservative initiation of antimicrobial treatment in critically ill surgical patients with suspected intensive-care-unit-acquired infection: a quasi-experimental, before and after observational cohort study.   Lancet Infect Dis. 2012;12(10):774-780. doi:10.1016/S1473-3099(12)70151-2PubMedGoogle ScholarCrossref
14.
Kett  DH, Cano  E, Quartin  AA,  et al; Improving Medicine through Pathway Assessment of Critical Therapy of Hospital-Acquired Pneumonia (IMPACT-HAP) Investigators.  Implementation of guidelines for management of possible multidrug-resistant pneumonia in intensive care: an observational, multicentre cohort study.   Lancet Infect Dis. 2011;11(3):181-189. doi:10.1016/S1473-3099(10)70314-5PubMedGoogle ScholarCrossref
15.
Webb  BJ, Sorensen  J, Jephson  A, Mecham  I, Dean  NC.  Broad-spectrum antibiotic use and poor outcomes in community-onset pneumonia: a cohort study.   Eur Respir J. 2019;54(1):1900057. doi:10.1183/13993003.00057-2019PubMedGoogle Scholar
16.
Rhee  C, Dantes  R, Epstein  L,  et al; CDC Prevention Epicenter Program.  Incidence and trends of sepsis in US hospitals using clinical vs claims data, 2009-2014.   JAMA. 2017;318(13):1241-1249. doi:10.1001/jama.2017.13836PubMedGoogle ScholarCrossref
17.
Choudhry  SA, Li  J, Davis  D, Erdmann  C, Sikka  R, Sutariya  B.  A public-private partnership develops and externally validates a 30-day hospital readmission risk prediction model.   Online J Public Health Inform. 2013;5(2):219. doi:10.5210/ojphi.v5i2.4726PubMedGoogle ScholarCrossref
18.
Goyal  A, Spertus  JA, Gosch  K,  et al.  Serum potassium levels and mortality in acute myocardial infarction.   JAMA. 2012;307(2):157-164. doi:10.1001/jama.2011.1967PubMedGoogle ScholarCrossref
19.
Lagu  T, Pekow  PS, Shieh  MS,  et al.  Validation and comparison of seven mortality prediction models for hospitalized patients with acute decompensated heart failure.   Circ Heart Fail. 2016;9(8):e002912. doi:10.1161/CIRCHEARTFAILURE.115.002912PubMedGoogle Scholar
20.
Petrick  JL, Nguyen  T, Cook  MB.  Temporal trends of esophageal disorders by age in the Cerner Health Facts database.   Ann Epidemiol. 2016;26(2):151-154.e4.PubMedGoogle ScholarCrossref
21.
Vandenbroucke  JP, von Elm  E, Altman  DG,  et al; STROBE Initiative.  Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration.   PLoS Med. 2007;4(10):e297. doi:10.1371/journal.pmed.0040297PubMedGoogle Scholar
22.
Centers for Disease Control and Prevention. Hospital toolkit for adult sepsis surveillance. Accessed February 16th, 2020. https://www.cdc.gov/sepsis/pdfs/Sepsis-Surveillance-Toolkit-Mar-2018_508.pdf
23.
Tabah  A, Koulenti  D, Laupland  K,  et al.  Characteristics and determinants of outcome of hospital-acquired bloodstream infections in intensive care units: the EUROBACT International Cohort Study.   Intensive Care Med. 2012;38(12):1930-1945. doi:10.1007/s00134-012-2695-9PubMedGoogle ScholarCrossref
24.
Kadri  SS, Adjemian  J, Lai  YL,  et al; National Institutes of Health Antimicrobial Resistance Outcomes Research Initiative (NIH–ARORI).  Difficult-to-treat resistance in gram-negative bacteremia at 173 US hospitals: retrospective cohort analysis of prevalence, predictors, and outcome of resistance to all first-line agents.   Clin Infect Dis. 2018;67(12):1803-1814. doi:10.1093/cid/ciy378PubMedGoogle Scholar
25.
Kalil  AC, Metersky  ML, Klompas  M,  et al.  Executive summary: management of adults with hospital-acquired and ventilator-associated pneumonia: 2016 Clinical Practice Guidelines by the Infectious Diseases Society of America and the American Thoracic Society.   Clin Infect Dis. 2016;63(5):575-582. doi:10.1093/cid/ciw504PubMedGoogle ScholarCrossref
26.
Cooper  AJ, Keller  S, Chan  C,  et al; CDC Prevention Epicenters Program.  Improvements in sepsis-associated mortality in hospitalized cancer vs non-cancer patients: a 12-year analysis using clinical data.   Ann Am Thorac Soc. 2019. doi:10.1513/AnnalsATS.201909-655OCPubMedGoogle Scholar
27.
Kadri  SS, Hohmann  SF, Orav  EJ,  et al.  Tracking colistin-treated patients to monitor the incidence and outcome of carbapenem-resistant Gram-negative infections.   Clin Infect Dis. 2015;60(1):79-87. doi:10.1093/cid/ciu741PubMedGoogle ScholarCrossref
28.
Moore  BJ, White  S, Washington  R, Coenen  N, Elixhauser  A.  Identifying increased risk of readmission and in-hospital mortality using hospital administrative data: the AHRQ Elixhauser Comorbidity Index.   Med Care. 2017;55(7):698-705. doi:10.1097/MLR.0000000000000735PubMedGoogle ScholarCrossref
29.
Cassini  A, Högberg  LD, Plachouras  D,  et al; Burden of AMR Collaborative Group.  Attributable deaths and disability-adjusted life-years caused by infections with antibiotic-resistant bacteria in the EU and the European Economic Area in 2015: a population-level modelling analysis.   Lancet Infect Dis. 2019;19(1):56-66. doi:10.1016/S1473-3099(18)30605-4PubMedGoogle ScholarCrossref
30.
Centers for Disease Control and Prevention. Antibiotic resistance threats in the United States, 2019. Accessed February 16, 2020. https://www.cdc.gov/drugresistance/pdf/threats-report/2019-ar-threats-report-508.pdf
31.
Phua  J, Ngerng  W, See  K,  et al.  Characteristics and outcomes of culture-negative versus culture-positive severe sepsis.   Crit Care. 2013;17(5):R202. doi:10.1186/cc12896PubMedGoogle ScholarCrossref
32.
Gupta  S, Sakhuja  A, Kumar  G, McGrath  E, Nanchal  RS, Kashani  KB.  Culture-negative severe sepsis: nationwide trends and outcomes.   Chest. 2016;150(6):1251-1259. doi:10.1016/j.chest.2016.08.1460PubMedGoogle ScholarCrossref
33.
Jain  S, Self  WH, Wunderink  RG,  et al; CDC EPIC Study Team.  Community-acquired pneumonia requiring hospitalization among U.S. adults.   N Engl J Med. 2015;373(5):415-427. doi:10.1056/NEJMoa1500245PubMedGoogle ScholarCrossref
34.
Klein Klouwenberg  PM, Cremer  OL, van Vught  LA,  et al.  Likelihood of infection in patients with presumed sepsis at the time of intensive care unit admission: a cohort study.   Crit Care. 2015;19:319. doi:10.1186/s13054-015-1035-1PubMedGoogle ScholarCrossref
35.
Bhattacharyya  RP, Bandyopadhyay  N, Ma  P,  et al.  Simultaneous detection of genotype and phenotype enables rapid and accurate antibiotic susceptibility determination.   Nat Med. 2019;25(12):1858-1864. doi:10.1038/s41591-019-0650-9PubMedGoogle ScholarCrossref
36.
Kothari  A, Morgan  M, Haake  DA.  Emerging technologies for rapid identification of bloodstream pathogens.   Clin Infect Dis. 2014;59(2):272-278. doi:10.1093/cid/ciu292PubMedGoogle ScholarCrossref
37.
Maxson  T, Blancett  CD, Graham  AS, Stefan  CP, Minogue  TD.  Rapid antibiotic susceptibility testing from blood culture bottles with species agnostic real-time polymerase chain reaction.   PLoS One. 2018;13(12):e0209042. doi:10.1371/journal.pone.0209042PubMedGoogle Scholar
38.
Maurer  FP, Christner  M, Hentschke  M, Rohde  H.  Advances in rapid identification and susceptibility testing of bacteria in the clinical microbiology laboratory: implications for patient care and antimicrobial stewardship programs.   Infect Dis Rep. 2017;9(1):6839. doi:10.4081/idr.2017.6839PubMedGoogle ScholarCrossref
39.
Vazquez-Guillamet  MC, Vazquez  R, Micek  ST, Kollef  MH.  Predicting resistance to piperacillin-tazobactam, cefepime and meropenem in septic patients with bloodstream infection due to gram-negative bacteria.   Clin Infect Dis. 2017;65(10):1607-1614. doi:10.1093/cid/cix612PubMedGoogle ScholarCrossref
40.
Paul  M, Shani  V, Muchtar  E, Kariv  G, Robenshtok  E, Leibovici  L.  Systematic review and meta-analysis of the efficacy of appropriate empiric antibiotic therapy for sepsis.   Antimicrob Agents Chemother. 2010;54(11):4851-4863. doi:10.1128/AAC.00627-10PubMedGoogle ScholarCrossref
41.
Zilberberg  MD, Shorr  AF, Micek  ST, Vazquez-Guillamet  C, Kollef  MH.  Multi-drug resistance, inappropriate initial antibiotic therapy and mortality in gram-negative severe sepsis and septic shock: a retrospective cohort study.   Crit Care. 2014;18(6):596. doi:10.1186/s13054-014-0596-8PubMedGoogle ScholarCrossref
42.
Depuydt  PO, Vandijck  DM, Bekaert  MA,  et al.  Determinants and impact of multidrug antibiotic resistance in pathogens causing ventilator-associated-pneumonia.   Crit Care. 2008;12(6):R142. doi:10.1186/cc7119PubMedGoogle ScholarCrossref
43.
Zilberberg  MD, Nathanson  BH, Sulham  K, Fan  W, Shorr  AF.  A novel algorithm to analyze epidemiology and outcomes of carbapenem resistance among patients with hospital-acquired and ventilator-associated pneumonia: a retrospective cohort study.   Chest. 2019;155(6):1119-1130. doi:10.1016/j.chest.2018.12.024PubMedGoogle ScholarCrossref
44.
Klompas  M, Rhee  C.  The CMS sepsis mandate: right disease, wrong measure.   Ann Intern Med. 2016;165(7):517-518. doi:10.7326/M16-0588PubMedGoogle ScholarCrossref
45.
Tamma  PD, Avdic  E, Li  DX, Dzintars  K, Cosgrove  SE.  Association of adverse events with antibiotic use in hospitalized patients.   JAMA Intern Med. 2017;177(9):1308-1315. doi:10.1001/jamainternmed.2017.1938PubMedGoogle ScholarCrossref
46.
Carignan  A, Allard  C, Pépin  J, Cossette  B, Nault  V, Valiquette  L.  Risk of Clostridium difficile infection after perioperative antibacterial prophylaxis before and during an outbreak of infection due to a hypervirulent strain.   Clin Infect Dis. 2008;46(12):1838-1843. doi:10.1086/588291PubMedGoogle ScholarCrossref
47.
Brown  KA, Khanafer  N, Daneman  N, Fisman  DN.  Meta-analysis of antibiotics and the risk of community-associated Clostridium difficile infection.   Antimicrob Agents Chemother. 2013;57(5):2326-2332. doi:10.1128/AAC.02176-12PubMedGoogle ScholarCrossref
48.
Branch-Elliman  W, O’Brien  W, Strymish  J, Itani  K, Wyatt  C, Gupta  K.  Association of duration and type of surgical prophylaxis with antimicrobial-associated adverse events.   JAMA Surg. 2019;154(7):590-598. doi:10.1001/jamasurg.2019.0569PubMedGoogle ScholarCrossref
49.
Hammond  DA, Smith  MN, Li  C, Hayes  SM, Lusardi  K, Bookstaver  PB.  Systematic review and meta-analysis of acute kidney injury associated with concomitant vancomycin and piperacillin/tazobactam.   Clin Infect Dis. 2017;64(5):666-674.PubMedGoogle Scholar
50.
Jacobs  MC, Haak  BW, Hugenholtz  F, Wiersinga  WJ.  Gut microbiota and host defense in critical illness.   Curr Opin Crit Care. 2017;23(4):257-263. doi:10.1097/MCC.0000000000000424PubMedGoogle ScholarCrossref
51.
Bhalodi  AA, van Engelen  TSR, Virk  HS, Wiersinga  WJ.  Impact of antimicrobial therapy on the gut microbiome.   J Antimicrob Chemother. 2019;74(Supplement_1):i6-i15.PubMedGoogle ScholarCrossref
52.
Koo  HL, Van  JN, Zhao  M,  et al.  Real-time polymerase chain reaction detection of asymptomatic Clostridium difficile colonization and rising C. difficile-associated disease rates.   Infect Control Hosp Epidemiol. 2014;35(6):667-673. doi:10.1086/676433PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Infectious Diseases
    April 16, 2020

    Prevalencia de patógenos resistentes a los antibióticos en sepsis y resultados probados en cultivo asociados con el uso de antibióticos empíricos inadecuados y de amplio espectro

    Author Affiliations
    • 1Department of Population Medicine, Harvard Pilgrim Health Care Institute, Harvard Medical School, Boston, Massachusetts
    • 2Division of Infectious Diseases, Brigham and Women’s Hospital, Boston, Massachusetts
    • 3Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, Maryland
    • 4Laboratory of Clinical Immunology and Microbiology, National Institutes of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland
    • 5Commonwealth Informatics, Waltham, Massachusetts
    JAMA Netw Open. 2020;3(4):e202899. doi:10.1001/jamanetworkopen.2020.2899
    Puntos claveEnglish 中文 (chinese)

    Pregunta  ¿Cuál es la prevalencia de la resistencia a los antibióticos en la sepsis de aparición comunitaria? ¿Existe algún riesgo asociado con la recepción de antibióticos empíricos de amplio espectro?

    Conclusiones  En este estudio de cohortes de 17 430 adultos con sepsis con cultivo positivo ingresados en 104 hospitales de EE. UU., el 67 % recibió antibióticos empíricos de amplio espectro, pero los organismos grampositivos resistentes se aislaron en solo el 13.6 % de los pacientes y los organismos gramnegativos resistentes en el 13.2 %. Tanto el tratamiento deficiente (falta de cobertura de los organismos) como el tratamiento excesivo (organismos resistentes dirigidos pero no aislados) se asociaron con una mayor mortalidad después del ajuste detallado del riesgo.

    Significado  En este estudio, con frecuencia se administraron antibióticos de amplio espectro a pacientes con sepsis de aparición comunitaria sin organismos resistentes, y estas terapias se asociaron con peores resultados.

    Abstract

    Importance  Broad-spectrum antibiotics are recommended for all patients with suspected sepsis to minimize the risk of undertreatment. However, little is known regarding the net prevalence of antibiotic-resistant pathogens across all patients with community-onset sepsis or the outcomes associated with unnecessarily broad empiric treatment.

    Objective  To elucidate the epidemiology of antibiotic-resistant pathogens and the outcomes associated with both undertreatment and overtreatment in patients with culture-positive community-onset sepsis.

    Design, Setting, and Participants  This cohort study included 17 430 adults admitted to 104 US hospitals between January 2009 and December 2015 with sepsis and positive clinical cultures within 2 days of admission. Data analysis took place from January 2018 to December 2019.

    Exposures  Inadequate empiric antibiotic therapy (ie, ≥1 pathogen nonsusceptible to all antibiotics administered on the first or second day of treatment) and unnecessarily broad empiric therapy (ie, active against methicillin-resistant Staphylococcus aureus [MRSA]; vancomycin-resistant Enterococcus [VRE]; ceftriaxone-resistant gram-negative [CTX-RO] organisms, including Pseudomonas aeruginosa; or extended-spectrum β-lactamase [ESBL] gram-negative organisms when none of these were isolated).

    Main Outcomes and Measures  Prevalence and empiric treatment rates for antibiotic-resistant organisms and associations of inadequate and unnecessarily broad empiric therapy with in-hospital mortality were assessed, adjusting for baseline characteristics and severity of illness.

    Results  Of 17 430 patients with culture-positive community-onset sepsis (median [interquartile range] age, 69 [57-81] years; 9737 [55.9%] women), 2865 (16.4%) died in the hospital. The most common culture-positive sites were urine (9077 [52.1%]), blood (6968 [40.0%]), and the respiratory tract (2912 [16.7%]). The most common pathogens were Escherichia coli (5873 [33.7%]), S aureus (3706 [21.3%]), and Streptococcus species (2361 [13.5%]). Among 15 183 cases in which all antibiotic-pathogen susceptibility combinations could be calculated, most (12 398 [81.6%]) received adequate empiric antibiotics. Empiric therapy targeted resistant organisms in 11 683 of 17 430 cases (67.0%; primarily vancomycin and anti-Pseudomonal β-lactams), but resistant organisms were uncommon (MRSA, 2045 [11.7%]; CTX-RO, 2278 [13.1%]; VRE, 360 [2.1%]; ESBLs, 133 [0.8%]). The net prevalence for at least 1 resistant gram-positive organism (ie, MRSA or VRE) was 13.6% (2376 patients), and for at least 1 resistant gram-negative organism (ie, CTX-RO, ESBL, or CRE), it was 13.2% (2297 patients). Both inadequate and unnecessarily broad empiric antibiotics were associated with higher mortality after detailed risk adjustment (inadequate empiric antibiotics: odds ratio, 1.19; 95% CI, 1.03-1.37; P = .02; unnecessarily broad empiric antibiotics: odds ratio, 1.22; 95% CI, 1.06-1.40; P = .007).

    Conclusions and Relevance  In this study, most patients with community-onset sepsis did not have resistant pathogens, yet broad-spectrum antibiotics were frequently administered. Both inadequate and unnecessarily broad empiric antibiotics were associated with higher mortality. These findings underscore the need for better tests to rapidly identify patients with resistant pathogens and for more judicious use of broad-spectrum antibiotics for empiric sepsis treatment.

    Introduction

    Sepsis, the syndrome of life-threatening organ dysfunction complicating severe infection, is a leading cause of death in hospitalized patients.1 Early active antibiotic therapy is associated with better outcomes.2-5 Therefore, national quality measures and international guidelines recommend immediate empiric broad-spectrum antibiotics for all patients with suspected sepsis.6,7

    However, it is unclear how many patients presenting with sepsis require coverage for methicillin-resistant Staphylococcus aureus (MRSA), Pseudomonas aeruginosa, and other potentially resistant pathogens. Most existing epidemiologic studies present rates of resistance to selected antimicrobials per pathogen rather than quantifying overall resistance rates across all patients and syndromes associated with sepsis. These data are needed to inform the rational use of antibiotics, given that overuse of broad-spectrum therapy may also confer harm by selecting for antibiotic-resistant bacteria, increasing the risk of adverse events, such as Clostridioides difficile infections, and raising costs.8-12 Overtreatment has also been associated with higher mortality rates in some populations.13-15 Therefore, we sought to elucidate the epidemiology of antibiotic-resistant pathogens in patients with culture-positive community-onset sepsis and the risks of both inadequate and unnecessarily broad antibiotic treatments in US hospitals.

    Methods
    Study Design, Data Source, and Population

    We conducted a retrospective cohort study using Cerner HealthFacts, a deidentified data set that includes detailed electronic clinical data from diverse US hospitals.16-20 We included all patients aged at least 20 years who were admitted between January 2009 and September 2015, excluding those with missing discharge dispositions or International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) rather than ICD-9-CM codes.16 Data analysis took place between January 2018 and December 2019 and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.21 This study was approved by the institutional review board at Harvard Pilgrim Health Care Institute with a waiver of informed consent given that the patients in this study had high mortality rates, making collection of consent infeasible.

    Identifying Patients With Culture-Positive Sepsis

    We identified sepsis hospitalizations using the US Centers for Disease Control and Prevention Adult Sepsis Event surveillance criteria, which defines sepsis as concurrent evidence of presumed serious infection (ie, blood culture order and ≥4 consecutive days of new antibiotics) and organ dysfunction (ie, initiation of vasopressors or mechanical ventilation, lactate level ≥2.0 mmol/L [to convert to milligram per deciliter, divide by 0.111], doubling in baseline creatinine level or ≥50% decrease in estimated glomerular filtration rate, doubling in total bilirubin level to ≥2.0 mg/dL [to convert to micromoles per liter, multiply by 17.104], or ≥50% decrease in platelet count to <100 × 103/μL [to convert to × 109/L, multiply by 1.0]).22 We focused on culture-positive community-onset sepsis, as defined by a blood culture draw, first antibiotic administration, organ dysfunction, and clinical cultures obtained and subsequently testing positive for potentially pathogenic organisms, all on hospital day 2 or earlier. We excluded culture-negative patients because of the difficulty determining antibiotic appropriateness in this population using electronic data alone. We further excluded patients transferred from hospitals, rehabilitation or long-term facilities, and hospice as well as patients with hospital-onset sepsis because the epidemiology, antibiotic resistance patterns, and recommended treatments differ substantially in health care–acquired vs community-acquired infections.23-25 Infectious syndromes were classified using ICD-9-CM discharge codes (eAppendix 1 in the Supplement).26

    Culture Sites and Pathogens of Interest

    We included clinical cultures from the following anatomic sites: blood, respiratory, urine, deep tissue, central nervous system fluid, body fluid, and superficial tissue. We focused on pathogens commonly encountered in routine practice. Gram-negative organisms included Acinetobacter species, Citrobacter species, Enterobacter species, Escherichia coli, Klebsiella species, Proteus species, P aeruginosa, and Serratia species. Gram-positive organisms included S aureus, Streptococcus species, and Enterococcus species. Surveillance cultures and cultures positive for other organisms within the first 2 days of hospitalization were excluded, including coagulase-negative Staphylococcus species (because it can be difficult to distinguish contaminants vs true infections) and Enterococcus species isolated from respiratory samples.

    We assessed the prevalence of the following resistant organisms: MRSA, vancomycin-resistant Enterococcus (VRE), extended spectrum β-lactamase (ESBL) gram-negative organisms (ie, resistant to all β-lactams except carbapenems), carbapenem-resistant Enterobacteriaceae (CRE; defined as resistant to imipenem, meropenem, doripenem, or ertapenem), and ceftriaxone-resistant gram negatives (CTX-RO; including P aeruginosa and ESBLs). We grouped P aeruginosa with other CTX-RO organisms because they share the need for treatment with β-lactams with anti-Pseudomonal coverage, which are an important class of broad-spectrum antibiotics.

    Antibiotic Susceptibilities and Adequacy of Therapy

    We assessed each potential antibiotic-pathogen combination using antibiotic susceptibilities derived from in vitro reports generated by each institution. Intermediate susceptibilities were treated as nonsusceptible. In some cases, susceptibilities to specific antibiotics administered were not explicitly listed but could be assumed using knowledge of the spectrum of activity for each antibiotic-pathogen combination. For example, many antibiotics are intrinsically inactive against certain species (eg, ceftriaxone vs P aeruginosa) and are therefore not included in susceptibility reports. Alternatively, a gram-negative organism susceptible to ceftriaxone may not have susceptibilities reported to all higher-generation cephalosporins (ie, cefepime), but these agents can be safely used. We created rules to impute susceptibilities for each antibiotic-pathogen combination (eAppendix 2 in the Supplement).

    We considered patients to have received inadequate empiric therapy if at least 1 pathogen isolated from any clinical culture site was not susceptible to all antibiotics administered on the first and second day of treatment. We considered patients to have received unnecessarily broad empiric therapy if they received adequate empiric therapy and anti-MRSA antibiotics (ie, vancomycin, linezolid, or daptomycin), anti-VRE antibiotics (ie, linezolid or daptomycin), anti-Pseudomonal β-lactams (ie, ceftazidime, cefepime, piperacillin-tazobactam, aztreonam, imipenem, meropenem, or doripenem), or carbapenems (ie, imipenem, meropenem, doripenem, or ertapenem), but none of the organisms targeted by these antibiotics (ie, MRSA, VRE, CTX-RO, or ESBL) were recovered. We did not consider CRE in the analysis of unnecessarily broad therapy given the infrequency of empiric CRE treatment.27

    Outcomes

    We examined associations of both inadequate and unnecessarily broad empiric regimens with in-hospital mortality. The analysis of unnecessarily broad antibiotics was limited to patients who received adequate antibiotics because an empiric regimen would generally not be considered unnecessarily broad if the spectrum was inadequate. Secondary outcomes included hospital-onset acute kidney injury, defined as an increase in creatinine level by at least 0.5 mg/dL (to convert to micrograms per liter, multiply by 1) at any point during hospitalization relative to the initial value on presentation and C difficile infections, which we identified using the ICD-9-CM code 008.45 because C difficile test results were unavailable in our data set.

    Statistical Analysis

    We fit logistic regression models using generalized estimating equations to account for clustering within hospitals. Model covariates included admission year, hospital characteristics (bed size, region, teaching status), patient demographic characteristics (age, sex, race), total burden of comorbidities (Agency for Healthcare Research and Quality [AHRQ] Elixhauser Comorbidity Index28), microbiologic characteristics (site of positive clinical culture, pathogen, and presence of antibiotic resistance [MRSA, VRE, CTX-RO, ESBL, or CRE]), infectious syndrome (by ICD-9-CM codes), intensive care unit (ICU) care on admission, and physiological markers of severity of illness on admission (number of vasopressors, need for mechanical ventilation, and worst values for temperature, systolic blood pressure, respiratory rate, Glasgow Coma Scale score, serum lactate level, creatinine level, anion gap, total bilirubin level, aspartate aminotransferase level, white blood cell count, hematocrit level, platelet count, and serum albumin level). We used univariate logistic regression to assess associations of covariates with each outcome. Covariates with univariate P < .10 were included in the multivariable models. The 2247 cases (12.9%) in which susceptibility or resistance to all administered antibiotics could not be reliably imputed were excluded from this multivariable analysis (eAppendix 3 in the Supplement).

    Multiple imputation was used in our primary analysis to assign values for missing severity-of-illness physiological variables (eAppendix 3 in the Supplement). Several sensitivity analyses were conducted to handle missing data in different ways. First, we imputed all missing severity-of-illness covariates with median values. Second, we limited analyses to patients with nonmissing vital signs (ie, temperature, blood pressure, and respiratory rate) and used multiple imputation to account for other missing covariates. Third, we limited analyses to patients with complete data for all covariates.

    For our assessment of inadequate antibiotic therapy and mortality, we conducted an additional sensitivity analysis restricted to positive blood cultures because these organisms unequivocally represent true pathogens, whereas organisms isolated from most other sites can sometimes be colonizers rather than pathogens. We also conducted subgroup analyses restricted to patients with septic shock, as defined by the need for vasopressors on admission.

    All tests of significance used a 2-sided P < .05. Analyses were conducted using SAS version 9.4 (SAS Institute).

    Results
    Patient Characteristics and Pathogen Epidemiology

    The cohort included 17 430 patients with culture-positive community-onset sepsis (median [interquartile range {IQR}] age, 69 [57-81] years; 9737 [55.9%] women) (Figure 1). Of these, 5609 (32.2%) had septic shock requiring vasopressors, 8001 (45.9%) were admitted to the ICU, and 2865 (16.4%) died in the hospital (Table 1). Patients with septic shock had higher mortality than those without shock (1733 of 5345 [32.4%] vs 1137 of 12 098 [9.4%]; P < .001).

    Urinary tract infection was the most common infectious diagnosis (8515 [48.9%]), followed by pulmonary (5728 [32.9%]), intra-abdominal (2373 [13.6%]), and skin or soft tissue (1787 [10.3%]) infections. The most common positive culture sites were urine (9077 [52.1%]), blood (6968 [40.0%]), and the respiratory tract (2912, [16.7%]). The top pathogens were E coli (5873 [33.7%]), S aureus (3706 [21.3%]), Streptococcus species (2361 [13.5%]), Klebsiella (2254 [12.9%]), and Enterococcus (1928 [11.1%]) (Figure 2). Empiric therapy targeted resistant organisms in 11 683 of 17 430 cases (67.0%). Drug-resistant pathogens were relatively uncommon (MRSA, 2045 [11.7%]; CTX-RO, 2278 [13.1%], of whom 1510 patients [66.3%] had P aeruginosa; VRE, 360 [2.1%]; ESBLs, 133 [0.8%]; and CRE, 83 [0.5%]). The net prevalence of at least 1 resistant gram-positive organism (ie MRSA or VRE) was 13.6% (2376 patients); at least 1 resistant gram negative organism (ie, CTX-RO, ESBL, CRE), 13.2% (2297 patients); and any of these organisms, 25.7% (4474 patients). The prevalence of pathogens across blood, respiratory, and urine cultures is shown in eFigure 1 in the Supplement.

    The characteristics of patients with vs without resistant organisms (4474 [25.7%] vs 12 956 [74.3%]) are shown in Table 1. Patients with resistant organisms were more likely to have more comorbidities (median [IQR] AHRQ Elixhauser Comorbidity Index score, 12 [4-20] vs 11 [3-19]; P < .001), to have a pulmonary infection (1851 [41.4%] vs 3877 [29.9%]; P < .001), to have positive respiratory cultures (1339 [29.9%] vs 1573 [12.1%]; P < .001), to require vasopressors (1612 [36.0%] vs 3997 [30.9%]; P < .001) or mechanical ventilation (1264 [28.3%] vs 2489 [19.2%]; P < .001), to require ICU admission (2243 [50.1%] vs 5758 [44.4%]; P < .001), and to die in the hospital (888 [19.9%] vs 1977 [15.3%]; crude odds ratio [OR], 1.38; 95% CI, 1.26-1.50).

    The prevalence of resistant organisms was higher in the 5609 patients with septic shock vs 11 821 patients without shock (1612 [28.7%] vs 2862 [24.2%]; P < .001), including higher rates of MRSA (785 [14.0%] vs 1260 [10.7%]; P < .001) and CTX-RO (801 [14.3%] vs 1477 [12.5%]; P = .001) but not ESBL (49 [0.9%] vs 84 [0.7%]; P = .25), VRE (117 [2.1%] vs 243 [2.1%]; P = .90), or CRE (52 [0.9%] vs 31 [0.3%]; P = .31) (eFigure 2 in the Supplement).

    Empiric Antibiotics and Rates of Overtreatment for Resistant Pathogens

    Vancomycin was the most commonly prescribed empiric antibiotic (7262 [41.7%]), followed by an anti-Pseudomonal fluoroquinolone (ciprofloxacin or levofloxacin, 6997 [40.1%]), piperacillin-tazobactam (5911 [33.9%]), ceftriaxone (5187 [29.8%]), and third- or fourth-generation cephalosporins (ceftazidime or cefepime, 2300 [13.2%]) (eFigure 3 in the Supplement). Anti-MRSA treatment (ie, vancomycin, linezolid, or daptomycin) was given to 7936 of 17 430 patients (45.5%), of whom 1310 (16.5%) had positive cultures for MRSA. Anti-Pseudomonal β-lactams were given to 9031 patients (51.8%), of whom 1367 (15.1%) had positive cultures for CTX-ROs. Anti-VRE treatment was given to 1040 patients (6.0%), of whom 58 (5.6%) had positive cultures for VRE. Carbapenems were given to 1408 patients (8.1%), of whom 19 (1.4%) had positive cultures for ESBL. Overall, 11 797 patients (67.7%) received anti-MRSA, anti-Pseudomonal, anti-VRE, and/or anti-ESBL treatment, of whom 3447 (29.2%) had at least 1 of these organisms isolated (Figure 3). In addition, 4090 patients (23.5%) received double coverage for gram-negative organisms; 3260 patients (8.7%) received anti-Pseudomonal β-lactams with ciprofloxacin or levofloxacin, and 830 patients (4.8%) received anti-Pseudomonal β-lactams with amikacin, gentamicin, or tobramycin.

    Associations Between Empiric Therapy, Antibiotic Resistance, and Outcomes

    The crude and adjusted associations between empiric therapy patterns and outcomes are summarized in Table 2. Empiric antibiotics were active against all isolated pathogens in 12 398 of 15 183 (81.6%) sepsis cases in which all antibiotic-pathogen susceptibility combinations could be calculated. Compared with 12 398 patients who received adequate therapy, 2785 patients who received inadequate empiric antibiotic therapy were older (median [IQR] age, 71 [60-83] years vs 68 [56-80] years; P < .001) and had a higher burden of comorbidities (median [IQR] AHRQ Elixhauser Comorbidity Index score, 13 [5-21] vs 11 [2-19]; P < .001), but the groups had similar rates of organ dysfunction (eg, renal dysfunction: 1480 [53.1%] vs 6465 [52.2%]; P = .34), ICU admission (1248 [44.8%] vs 5710 [46.1%]; P = .23), and in-hospital mortality (488 [17.5%] vs 2011 [16.3%]; P = .09) (eTable 1 in the Supplement). However, on multivariable analysis, inadequate therapy was significantly associated with higher mortality (adjusted OR, 1.19; 95% CI, 1.03-1.37; P = .02).

    Inadequate therapy was much more likely in patients with resistant pathogens (MRSA, VRE, CTX-RO, ESBL, or CRE) vs nonresistant pathogens (1544 of 3811 [40.5%] vs 1241 of 11 372 [10.9%]; P < .001). Although patients with antibiotic-resistant organisms had higher crude hospital mortality rates, there was no difference after adjusting for baseline characteristics, severity of illness, and adequacy of therapy (adjusted OR, 1.04; 95% CI, 0.83-1.30; P = .75). There was also no association between antibiotic-resistant organisms and mortality when only considering positive blood cultures (adjusted OR, 1.10; 95% CI, 0.82-1.46; P = .54).

    Inadequate antibiotic therapy was not significantly associated with hospital death in the subgroup of patients with septic shock (OR, 1.10; 95% CI, 0.87-1.38; P = .44), but there was an association between inadequate antibiotics and hospital death when considering patients with sepsis who had positive blood cultures alone (adjusted OR, 1.40; 95% CI, 1.07-1.84; P = .02). The risk of C difficile was similar in patients who received inadequate vs adequate therapy (adjusted OR 1.19; 95% CI, 0.98-1.45; P = .09) as was the risk of hospital-onset acute kidney injury (adjusted OR, 1.02; 95% CI, 0.90-1.16; P = .72).

    Patients who received adequate but unnecessarily broad empiric antibiotics were younger and had a similar burden of comorbidities compared with those who did not receive unnecessarily broad therapy (median [IQR] age, 67 [55-79] years vs 71 [58-82] years; P < .001; median [IQR] AHRQ Elixhauser Comorbidity Index score, 11 [2-19] vs 11 [3-18]; P = .11) but were more severely ill on admission, with higher rates of vasopressor use (3310 [39.4%] vs 858 [21.5%]; P < .001), mechanical ventilation (1987 [23.6%] vs 601 [15.1%]; P < .001), ICU care (4276 [50.9%] vs 1434 [35.9%]; P < .001), and crude mortality (1575 [18.7%] vs 436 [10.9%]; P < .001) (eTable 2 in the Supplement). The association between unnecessarily broad empiric antibiotics and higher mortality persisted after risk adjustment (adjusted OR, 1.22; 95% CI, 1.06-1.49; P = .007). On subgroup analysis, the association between unnecessarily broad therapy and higher mortality was only seen in patients with sepsis and without shock (adjusted OR, 1.33; 95% CI, 1.09-1.60; P = .005) but not in patients with septic shock (adjusted OR, 1.12; 95% CI, 0.90-1.40; P = .32). The risk of C difficile among patients with sepsis was higher with unnecessarily broad therapy (adjusted OR, 1.26; 95% CI, 1.01-1.57; P = .04), but there was no association between unnecessarily broad therapy and hospital-onset acute kidney injury (adjusted OR, 1.12; 95% CI, 1.00-1.26; P = .05). The median (IQR) duration of treatment in the unnecessarily broad group was 3 (1-5) days for vancomycin, 4 (2-6) days for anti-Pseudomonal β-lactams, and 4 (2-6) days for carbapenems.

    The full univariate and multivariable models for mortality are shown in eTable 3 in the Supplement. The distribution of severity-of-illness covariates is shown in eTable 4 in the Supplement. The frequencies of missing data within the first 2 days of hospitalization are shown in eFigure 4 in the Supplement and were highest among laboratory data for lactate levels (7123 of 17 430 [40.9%]) but low for general chemistry (140 [0.8%]) and complete blood cell count variables (376 [2.2%]); vital sign data was missing in as many of 8309 cases (47.8%), and Glasgow Coma Scale scores were missing in 8872 cases (50.9%). Sensitivity analyses handling missing data in different ways, including limiting to patients with nonmissing data, yielded similar point estimates as the primary analysis (eTable 5 in the Supplement).

    Discussion

    Prior studies have estimated the national and global burden of antimicrobial resistance,29,30 but our study is among the first to estimate the net prevalence of antibiotic resistance across all culture sites in patients with culture-positive community-onset sepsis. We found that approximately 1 in 8 patients had resistant gram-positive organisms (primarily MRSA and rarely VRE) and 1 in 8 had resistant gram-negative organisms (primarily ceftriaxone-resistant gram-negative organisms and rarely ESBL or CRE). More than two-thirds of patients received broad-spectrum therapy directed at resistant organisms, but MRSA was only isolated in 1 in 6 patients treated with vancomycin or linezolid, P aeruginosa or other ceftriaxone-resistant gram-negative organisms in 1 in 6 patients treated with anti-Pseudomonal agents, VRE in 1 in 16 patients treated with linezolid or daptomycin, and ESBLs in 1 in 70 patients treated with carbapenems. Both inadequate and unnecessarily broad empiric therapy were associated with higher mortality after detailed risk adjustment.

    Our findings almost certainly overestimate the prevalence of resistant pathogens across the entire spectrum of patients treated for possible sepsis given the following: (1) we limited our analysis to bacterial, culture-positive sepsis, and between 30% and 50% of all patients with sepsis are culture negative31,32; (2) viruses are often implicated in severe pneumonia (the most common cause of sepsis), and (3) there are many noninfectious mimickers of sepsis that are treated as sepsis.33,34 All told, the net fraction of patients with sepsis who would benefit from broad-spectrum therapy, including agents active against both MRSA and Pseudomonas, is small. This may be an acceptable trade-off given the increased risk of death associated with inadequate therapy, but it underscores the need for rapid tests to more efficiently identify the small fraction of patients who truly need broad spectrum therapy.35 Alternatively, predictive models may soon allow clinicians to effectively select and tailor antibiotic regimens at the point of care.36-39

    In our cohort, inadequate therapy was associated with a 20% to 40% higher odds of death depending on whether all cultures or only blood cultures were analyzed. Our estimates are in the range of the results of a meta-analysis of 48 studies that reported a pooled odds ratio of 1.6 for death associated with inadequate therapy in patients with sepsis.40 We also found that patients with antibiotic-resistant pathogens received inadequate empiric therapy 4 times as often compared with patients with nonresistant organisms; patients with resistant pathogens who received inadequate therapy had higher mortality rates. However, we did not find an association between antibiotic-resistant organisms and mortality after adjusting for baseline and clinical characteristics as well as adequacy of empiric antibiotics. This suggests that the higher crude mortality rates in patients with resistant organisms could be mediated by their higher comorbidity burden, greater severity of illness, and inadequate antibiotic therapy rather than intrinsic virulence of resistant organisms.41-43

    While clinicians, guidelines, and policies understandably emphasize broad-spectrum antibiotics to ensure adequate empiric treatment,44 our findings suggest that the risk of inadequate therapy needs to be weighed against the risks of unnecessarily broad empiric antibiotics. Among patients who received adequate therapy, overtreatment was associated with a 20% increase in the odds of death. Other studies have also reported that more aggressive antibiotic regimens may be associated with higher mortality rates in critically ill patients.13-15 In our cohort, we found an association between overtreatment and mortality only among patients without shock. This may be because the morbidity of acute severe illness in patients with septic shock outweighs the possible morbidity of excessively broad antibiotic therapy, whereas in less critically ill patients, the morbidity of excessively broad antibiotics may be more significant. Another possibility is residual confounding among patients without shock because of the wide array of infections and organ dysfunction in this group that may make it more difficult to adequately adjust for all gradations in illness. The possibility of residual confounding is supported by the observation that more severely ill patients were more likely to receive broad-spectrum antibiotics.

    There are several other potential explanations for the association between overtreatment and higher mortality. As many as 20% of hospitalized patients who receive antibiotics experience adverse effects.45 Even a single dose of antibiotics can increase the risk of C difficile; this risk is higher with broad- vs narrow-spectrum antibiotics.46,47 We found unnecessarily broad empiric therapy was associated with a 26% increased risk of C difficile infection in our study. Unnecessary antibiotics may also increase the risk of acute kidney injury.48 We observed a trend toward more acute kidney injury in patients treated with unnecessarily broad antibiotics. In particular, the combination of vancomycin and piperacillin-tazobactam, a common regimen during the study period, is associated with renal toxicity.49 Broad-spectrum antibiotics also disrupt the gut microbiome, an increasingly recognized modulator of the immune system and outcomes in sepsis.50,51 Lastly, broad-spectrum antibiotics may increase the risk of resistant hospital-acquired infections.

    Limitations

    Our study has important limitations. First, we used a convenience sample of hospitals, which may limit generalizability. Antibiotic resistance rates vary substantially by region, hospital, and even within a facility. Second, our data did not allow us to calculate the time to antibiotics on the scale of hours, an important predictor of patient outcomes in some studies.2,4 Third, our primary analysis included pathogens isolated from all clinical cultures, but not all of these may be pathogenic. However, an analysis using only organisms isolated from blood demonstrated similar results with respect to the association between inadequate antibiotics and mortality. Fourth, we excluded patients with atypical pathogens owing to the complexity of determining adequate vs excessive treatment in this population. Fifth, C difficile assay results were unavailable in our data sets, so we had to use ICD-9-CM codes for this outcome; this prevented us from knowing whether C difficile infections developed while in the hospital or were present on admission and to what degree there was misclassification of true infections vs colonization.52 Sixth, we excluded patients transferred from other hospitals or health care facilities to focus on community-onset infections, but our data did not allow us to identify patients who might have been recently hospitalized. Seventh, our data sets did not allow us to examine the full array of potential complications of antibiotics, such as hepatitis, cytopenias, and drug eruptions. Data on patients’ allergies were also unavailable to us; therefore, we could not account for broad-spectrum antibiotics administered because patients were allergic to narrower agents (such as vancomycin or carbapenems for patients with β-lactam allergies). Furthermore, we only included patients with culture-positive sepsis, but a substantial fraction of patients with sepsis are culture negative.31,32 Determining the consequences of unnecessarily broad antibiotic therapy in the culture-negative population is an important topic for future research.

    Conclusions

    In this study of a large US cohort, we found that most patients with culture-positive community-onset sepsis did not have resistant organisms; however, empiric, broad-spectrum antibiotics targeting these organisms were frequently prescribed. Both inadequate and unnecessarily broad empiric therapy were associated with higher mortality. These findings underscore the need for better diagnostic tests to rapidly identify resistant pathogens and an increased focus on judicious use of broad-spectrum antibiotics for the empiric treatment of sepsis.

    Back to top
    Article Information

    Accepted for Publication: February 19, 2020.

    Published: April 16, 2020. doi:10.1001/jamanetworkopen.2020.2899

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Rhee C et al. JAMA Network Open.

    Corresponding Author: Chanu Rhee, MD, MPH, Department of Population Medicine, Harvard Pilgrim Health Care Institute, Harvard Medical School, 401 Park Dr, Ste 401, Boston, MA 02215 (crhee@bwh.harvard.edu).

    Author Contributions: Dr Rhee had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Rhee, Dekker, Danner, Klompas.

    Acquisition, analysis, or interpretation of data: Rhee, Kadri, Dekker, Chen, Fram, Zhang, Wang, Klompas.

    Drafting of the manuscript: Rhee, Wang, Klompas.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Rhee, Zhang, Wang.

    Obtained funding: Rhee, Kadri, Klompas.

    Administrative, technical, or material support: Chen, Fram.

    Supervision: Klompas.

    Conflict of Interest Disclosures: Dr Rhee reported receiving personal fees from UpToDate outside the submitted work. Dr Chen reported having a consulting service contract with Harvard Pilgrim Health Care Institute during the conduct of the study. Mr Fram reported having a consulting service contract with Harvard Pilgrim Healthcare Institute during the conduct of the study. Dr Klompas reported receiving personal fees from UpToDate outside the submitted work. No other disclosures were reported.

    Funding/Support: This work was funded by grant U54CK000484 from the Centers for Disease Control and Prevention, grant K08HS025008 from the Agency for Healthcare Research and Quality to Dr Rhee, and intramural funds from the National Institutes of Health Clinical Center and National Institute of Allergy and Infectious Diseases to Drs Kadri, Dekker, and Danner.

    Role of the Funder/Sponsor: Members of the National Institutes of Health Clinical Center who are coauthors on this study contributed to the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The Centers for Disease Control and Prevention and Agency for Healthcare Research and Quality had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    References
    1.
    Rhee  C, Jones  TM, Hamad  Y,  et al; Centers for Disease Control and Prevention (CDC) Prevention Epicenters Program.  Prevalence, underlying causes, and preventability of sepsis-associated mortality in US acute care hospitals.   JAMA Netw Open. 2019;2(2):e187571. doi:10.1001/jamanetworkopen.2018.7571PubMedGoogle Scholar
    2.
    Kumar  A, Roberts  D, Wood  KE,  et al.  Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock.   Crit Care Med. 2006;34(6):1589-1596. doi:10.1097/01.CCM.0000217961.75225.E9PubMedGoogle ScholarCrossref
    3.
    Kalil  AC, Johnson  DW, Lisco  SJ, Sun  J.  Early goal-directed therapy for sepsis: a novel solution for discordant survival outcomes in clinical trials.   Crit Care Med. 2017;45(4):607-614. doi:10.1097/CCM.0000000000002235PubMedGoogle ScholarCrossref
    4.
    Seymour  CW, Gesten  F, Prescott  HC,  et al.  Time to treatment and mortality during mandated emergency care for sepsis.   N Engl J Med. 2017;376(23):2235-2244. doi:10.1056/NEJMoa1703058PubMedGoogle ScholarCrossref
    5.
    Vazquez-Guillamet  C, Scolari  M, Zilberberg  MD, Shorr  AF, Micek  ST, Kollef  M.  Using the number needed to treat to assess appropriate antimicrobial therapy as a determinant of outcome in severe sepsis and septic shock.   Crit Care Med. 2014;42(11):2342-2349. doi:10.1097/CCM.0000000000000516PubMedGoogle ScholarCrossref
    6.
    Aaronson  EL, Filbin  MR, Brown  DF, Tobin  K, Mort  EA.  New mandated Centers for Medicare & Medicaid Services requirements for sepsis reporting: caution from the field.   J Emerg Med. 2017;52(1):109-116. doi:10.1016/j.jemermed.2016.08.009PubMedGoogle ScholarCrossref
    7.
    Rhodes  A, Evans  LE, Alhazzani  W,  et al.  Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock: 2016.   Crit Care Med. 2017;45(3):486-552. doi:10.1097/CCM.0000000000002255PubMedGoogle ScholarCrossref
    8.
    Huttner  A, Harbarth  S, Carlet  J,  et al.  Antimicrobial resistance: a global view from the 2013 World Healthcare-Associated Infections Forum.   Antimicrob Resist Infect Control. 2013;2:31. doi:10.1186/2047-2994-2-31PubMedGoogle ScholarCrossref
    9.
    Hiensch  R, Poeran  J, Saunders-Hao  P,  et al.  Impact of an electronic sepsis initiative on antibiotic use and health care facility-onset Clostridium difficile infection rates.   Am J Infect Control. 2017;45(10):1091-1100. doi:10.1016/j.ajic.2017.04.005PubMedGoogle ScholarCrossref
    10.
    Polgreen  PM, Chen  YY, Cavanaugh  JE,  et al.  An outbreak of severe Clostridium difficile-associated disease possibly related to inappropriate antimicrobial therapy for community-acquired pneumonia.   Infect Control Hosp Epidemiol. 2007;28(2):212-214. doi:10.1086/512174PubMedGoogle ScholarCrossref
    11.
    Seetharaman  S, Wilson  C, Landrum  M,  et al.  Does use of electronic alerts for systemic inflammatory response syndrome (SIRS) to identify patients with sepsis improve mortality?   Am J Med. 2019;132(7):862-868. doi:10.1016/j.amjmed.2019.01.032PubMedGoogle ScholarCrossref
    12.
    Teshome  BF, Vouri  SM, Hampton  N, Kollef  MH, Micek  ST.  Duration of exposure to antipseudomonal β-lactam antibiotics in the critically ill and development of new resistance.   Pharmacotherapy. 2019;39(3):261-270. doi:10.1002/phar.2201PubMedGoogle ScholarCrossref
    13.
    Hranjec  T, Rosenberger  LH, Swenson  B,  et al.  Aggressive versus conservative initiation of antimicrobial treatment in critically ill surgical patients with suspected intensive-care-unit-acquired infection: a quasi-experimental, before and after observational cohort study.   Lancet Infect Dis. 2012;12(10):774-780. doi:10.1016/S1473-3099(12)70151-2PubMedGoogle ScholarCrossref
    14.
    Kett  DH, Cano  E, Quartin  AA,  et al; Improving Medicine through Pathway Assessment of Critical Therapy of Hospital-Acquired Pneumonia (IMPACT-HAP) Investigators.  Implementation of guidelines for management of possible multidrug-resistant pneumonia in intensive care: an observational, multicentre cohort study.   Lancet Infect Dis. 2011;11(3):181-189. doi:10.1016/S1473-3099(10)70314-5PubMedGoogle ScholarCrossref
    15.
    Webb  BJ, Sorensen  J, Jephson  A, Mecham  I, Dean  NC.  Broad-spectrum antibiotic use and poor outcomes in community-onset pneumonia: a cohort study.   Eur Respir J. 2019;54(1):1900057. doi:10.1183/13993003.00057-2019PubMedGoogle Scholar
    16.
    Rhee  C, Dantes  R, Epstein  L,  et al; CDC Prevention Epicenter Program.  Incidence and trends of sepsis in US hospitals using clinical vs claims data, 2009-2014.   JAMA. 2017;318(13):1241-1249. doi:10.1001/jama.2017.13836PubMedGoogle ScholarCrossref
    17.
    Choudhry  SA, Li  J, Davis  D, Erdmann  C, Sikka  R, Sutariya  B.  A public-private partnership develops and externally validates a 30-day hospital readmission risk prediction model.   Online J Public Health Inform. 2013;5(2):219. doi:10.5210/ojphi.v5i2.4726PubMedGoogle ScholarCrossref
    18.
    Goyal  A, Spertus  JA, Gosch  K,  et al.  Serum potassium levels and mortality in acute myocardial infarction.   JAMA. 2012;307(2):157-164. doi:10.1001/jama.2011.1967PubMedGoogle ScholarCrossref
    19.
    Lagu  T, Pekow  PS, Shieh  MS,  et al.  Validation and comparison of seven mortality prediction models for hospitalized patients with acute decompensated heart failure.   Circ Heart Fail. 2016;9(8):e002912. doi:10.1161/CIRCHEARTFAILURE.115.002912PubMedGoogle Scholar
    20.
    Petrick  JL, Nguyen  T, Cook  MB.  Temporal trends of esophageal disorders by age in the Cerner Health Facts database.   Ann Epidemiol. 2016;26(2):151-154.e4.PubMedGoogle ScholarCrossref
    21.
    Vandenbroucke  JP, von Elm  E, Altman  DG,  et al; STROBE Initiative.  Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration.   PLoS Med. 2007;4(10):e297. doi:10.1371/journal.pmed.0040297PubMedGoogle Scholar
    22.
    Centers for Disease Control and Prevention. Hospital toolkit for adult sepsis surveillance. Accessed February 16th, 2020. https://www.cdc.gov/sepsis/pdfs/Sepsis-Surveillance-Toolkit-Mar-2018_508.pdf
    23.
    Tabah  A, Koulenti  D, Laupland  K,  et al.  Characteristics and determinants of outcome of hospital-acquired bloodstream infections in intensive care units: the EUROBACT International Cohort Study.   Intensive Care Med. 2012;38(12):1930-1945. doi:10.1007/s00134-012-2695-9PubMedGoogle ScholarCrossref
    24.
    Kadri  SS, Adjemian  J, Lai  YL,  et al; National Institutes of Health Antimicrobial Resistance Outcomes Research Initiative (NIH–ARORI).  Difficult-to-treat resistance in gram-negative bacteremia at 173 US hospitals: retrospective cohort analysis of prevalence, predictors, and outcome of resistance to all first-line agents.   Clin Infect Dis. 2018;67(12):1803-1814. doi:10.1093/cid/ciy378PubMedGoogle Scholar
    25.
    Kalil  AC, Metersky  ML, Klompas  M,  et al.  Executive summary: management of adults with hospital-acquired and ventilator-associated pneumonia: 2016 Clinical Practice Guidelines by the Infectious Diseases Society of America and the American Thoracic Society.   Clin Infect Dis. 2016;63(5):575-582. doi:10.1093/cid/ciw504PubMedGoogle ScholarCrossref
    26.
    Cooper  AJ, Keller  S, Chan  C,  et al; CDC Prevention Epicenters Program.  Improvements in sepsis-associated mortality in hospitalized cancer vs non-cancer patients: a 12-year analysis using clinical data.   Ann Am Thorac Soc. 2019. doi:10.1513/AnnalsATS.201909-655OCPubMedGoogle Scholar
    27.
    Kadri  SS, Hohmann  SF, Orav  EJ,  et al.  Tracking colistin-treated patients to monitor the incidence and outcome of carbapenem-resistant Gram-negative infections.   Clin Infect Dis. 2015;60(1):79-87. doi:10.1093/cid/ciu741PubMedGoogle ScholarCrossref
    28.
    Moore  BJ, White  S, Washington  R, Coenen  N, Elixhauser  A.  Identifying increased risk of readmission and in-hospital mortality using hospital administrative data: the AHRQ Elixhauser Comorbidity Index.   Med Care. 2017;55(7):698-705. doi:10.1097/MLR.0000000000000735PubMedGoogle ScholarCrossref
    29.
    Cassini  A, Högberg  LD, Plachouras  D,  et al; Burden of AMR Collaborative Group.  Attributable deaths and disability-adjusted life-years caused by infections with antibiotic-resistant bacteria in the EU and the European Economic Area in 2015: a population-level modelling analysis.   Lancet Infect Dis. 2019;19(1):56-66. doi:10.1016/S1473-3099(18)30605-4PubMedGoogle ScholarCrossref
    30.
    Centers for Disease Control and Prevention. Antibiotic resistance threats in the United States, 2019. Accessed February 16, 2020. https://www.cdc.gov/drugresistance/pdf/threats-report/2019-ar-threats-report-508.pdf
    31.
    Phua  J, Ngerng  W, See  K,  et al.  Characteristics and outcomes of culture-negative versus culture-positive severe sepsis.   Crit Care. 2013;17(5):R202. doi:10.1186/cc12896PubMedGoogle ScholarCrossref
    32.
    Gupta  S, Sakhuja  A, Kumar  G, McGrath  E, Nanchal  RS, Kashani  KB.  Culture-negative severe sepsis: nationwide trends and outcomes.   Chest. 2016;150(6):1251-1259. doi:10.1016/j.chest.2016.08.1460PubMedGoogle ScholarCrossref
    33.
    Jain  S, Self  WH, Wunderink  RG,  et al; CDC EPIC Study Team.  Community-acquired pneumonia requiring hospitalization among U.S. adults.   N Engl J Med. 2015;373(5):415-427. doi:10.1056/NEJMoa1500245PubMedGoogle ScholarCrossref
    34.
    Klein Klouwenberg  PM, Cremer  OL, van Vught  LA,  et al.  Likelihood of infection in patients with presumed sepsis at the time of intensive care unit admission: a cohort study.   Crit Care. 2015;19:319. doi:10.1186/s13054-015-1035-1PubMedGoogle ScholarCrossref
    35.
    Bhattacharyya  RP, Bandyopadhyay  N, Ma  P,  et al.  Simultaneous detection of genotype and phenotype enables rapid and accurate antibiotic susceptibility determination.   Nat Med. 2019;25(12):1858-1864. doi:10.1038/s41591-019-0650-9PubMedGoogle ScholarCrossref
    36.
    Kothari  A, Morgan  M, Haake  DA.  Emerging technologies for rapid identification of bloodstream pathogens.   Clin Infect Dis. 2014;59(2):272-278. doi:10.1093/cid/ciu292PubMedGoogle ScholarCrossref
    37.
    Maxson  T, Blancett  CD, Graham  AS, Stefan  CP, Minogue  TD.  Rapid antibiotic susceptibility testing from blood culture bottles with species agnostic real-time polymerase chain reaction.   PLoS One. 2018;13(12):e0209042. doi:10.1371/journal.pone.0209042PubMedGoogle Scholar
    38.
    Maurer  FP, Christner  M, Hentschke  M, Rohde  H.  Advances in rapid identification and susceptibility testing of bacteria in the clinical microbiology laboratory: implications for patient care and antimicrobial stewardship programs.   Infect Dis Rep. 2017;9(1):6839. doi:10.4081/idr.2017.6839PubMedGoogle ScholarCrossref
    39.
    Vazquez-Guillamet  MC, Vazquez  R, Micek  ST, Kollef  MH.  Predicting resistance to piperacillin-tazobactam, cefepime and meropenem in septic patients with bloodstream infection due to gram-negative bacteria.   Clin Infect Dis. 2017;65(10):1607-1614. doi:10.1093/cid/cix612PubMedGoogle ScholarCrossref
    40.
    Paul  M, Shani  V, Muchtar  E, Kariv  G, Robenshtok  E, Leibovici  L.  Systematic review and meta-analysis of the efficacy of appropriate empiric antibiotic therapy for sepsis.   Antimicrob Agents Chemother. 2010;54(11):4851-4863. doi:10.1128/AAC.00627-10PubMedGoogle ScholarCrossref
    41.
    Zilberberg  MD, Shorr  AF, Micek  ST, Vazquez-Guillamet  C, Kollef  MH.  Multi-drug resistance, inappropriate initial antibiotic therapy and mortality in gram-negative severe sepsis and septic shock: a retrospective cohort study.   Crit Care. 2014;18(6):596. doi:10.1186/s13054-014-0596-8PubMedGoogle ScholarCrossref
    42.
    Depuydt  PO, Vandijck  DM, Bekaert  MA,  et al.  Determinants and impact of multidrug antibiotic resistance in pathogens causing ventilator-associated-pneumonia.   Crit Care. 2008;12(6):R142. doi:10.1186/cc7119PubMedGoogle ScholarCrossref
    43.
    Zilberberg  MD, Nathanson  BH, Sulham  K, Fan  W, Shorr  AF.  A novel algorithm to analyze epidemiology and outcomes of carbapenem resistance among patients with hospital-acquired and ventilator-associated pneumonia: a retrospective cohort study.   Chest. 2019;155(6):1119-1130. doi:10.1016/j.chest.2018.12.024PubMedGoogle ScholarCrossref
    44.
    Klompas  M, Rhee  C.  The CMS sepsis mandate: right disease, wrong measure.   Ann Intern Med. 2016;165(7):517-518. doi:10.7326/M16-0588PubMedGoogle ScholarCrossref
    45.
    Tamma  PD, Avdic  E, Li  DX, Dzintars  K, Cosgrove  SE.  Association of adverse events with antibiotic use in hospitalized patients.   JAMA Intern Med. 2017;177(9):1308-1315. doi:10.1001/jamainternmed.2017.1938PubMedGoogle ScholarCrossref
    46.
    Carignan  A, Allard  C, Pépin  J, Cossette  B, Nault  V, Valiquette  L.  Risk of Clostridium difficile infection after perioperative antibacterial prophylaxis before and during an outbreak of infection due to a hypervirulent strain.   Clin Infect Dis. 2008;46(12):1838-1843. doi:10.1086/588291PubMedGoogle ScholarCrossref
    47.
    Brown  KA, Khanafer  N, Daneman  N, Fisman  DN.  Meta-analysis of antibiotics and the risk of community-associated Clostridium difficile infection.   Antimicrob Agents Chemother. 2013;57(5):2326-2332. doi:10.1128/AAC.02176-12PubMedGoogle ScholarCrossref
    48.
    Branch-Elliman  W, O’Brien  W, Strymish  J, Itani  K, Wyatt  C, Gupta  K.  Association of duration and type of surgical prophylaxis with antimicrobial-associated adverse events.   JAMA Surg. 2019;154(7):590-598. doi:10.1001/jamasurg.2019.0569PubMedGoogle ScholarCrossref
    49.
    Hammond  DA, Smith  MN, Li  C, Hayes  SM, Lusardi  K, Bookstaver  PB.  Systematic review and meta-analysis of acute kidney injury associated with concomitant vancomycin and piperacillin/tazobactam.   Clin Infect Dis. 2017;64(5):666-674.PubMedGoogle Scholar
    50.
    Jacobs  MC, Haak  BW, Hugenholtz  F, Wiersinga  WJ.  Gut microbiota and host defense in critical illness.   Curr Opin Crit Care. 2017;23(4):257-263. doi:10.1097/MCC.0000000000000424PubMedGoogle ScholarCrossref
    51.
    Bhalodi  AA, van Engelen  TSR, Virk  HS, Wiersinga  WJ.  Impact of antimicrobial therapy on the gut microbiome.   J Antimicrob Chemother. 2019;74(Supplement_1):i6-i15.PubMedGoogle ScholarCrossref
    52.
    Koo  HL, Van  JN, Zhao  M,  et al.  Real-time polymerase chain reaction detection of asymptomatic Clostridium difficile colonization and rising C. difficile-associated disease rates.   Infect Control Hosp Epidemiol. 2014;35(6):667-673. doi:10.1086/676433PubMedGoogle ScholarCrossref
    ×