Underlying Causes and Preventability of Hospital-Onset Sepsis Events
Introduction
Health care–associated infection (HAI) surveillance in US hospitals currently focuses on just 6 infections that constitute only a fraction of all serious HAIs.1 Hospital-onset sepsis surveillance using the US Centers for Disease Control and Prevention electronic Adult Sepsis Event criteria could expand the completeness and efficiency of HAI surveillance.2,3 Little is known, however, about the underlying causes and preventability of hospital-onset sepsis events, which are key considerations in assessing their potential use as hospital quality and safety metrics. We conducted detailed medical record reviews of hospital-onset sepsis events to address these knowledge gaps.
Methods
We performed a retrospective cohort study of adults age 18 years or older admitted between January 2021 and December 2023 to 9 Mass General Brigham acute care hospitals (2 academic and 7 community) in New England. We identified patients with hospital-onset sepsis using Centers for Disease Control and Prevention’s Adult Sepsis Event criteria (incorporating recently proposed updates),4 -6 which requires concurrent evidence of presumed serious infection and acute organ dysfunction starting on hospital day 4 or later (eTable 1 in Supplement 1). The study was approved by the Mass General Brigham institutional review board, which waived the need for direct patient consent, and was reported following the STROBE reporting guidelines.
We randomly selected 250 cases for detailed record reviews by 2 physicians (F.M. and S.S.) to determine likelihood and source of infection (eTable 2 in Supplement 1). We then selected 100 infection-related cases for preventability assessments considering both patients’ intrinsic risks and extrinsic health care–related factors per the consensus of 3 physicians (F.M., M.K., and C.R.) using a 6-point Likert scale (eTable 3 in Supplement 1).7 Cases scored 1 to 3 were classified as potentially preventable, and those rated 4 to 6 as not likely preventable (eAppendix and eFigure in Supplement 1).
Characteristics of potentially preventable and not likely preventable cases were compared using Fisher exact and Wilcoxon rank-sum tests. The 95% CIs for proportions were calculated using the Wilson method. Statistical significance was defined as P < .05 on 2-sided tests. Analyses were conducted using R statistical software version 2024.04.2+764 (R Project for Statistical Computing).
Results
Among 558 037 admissions, 5212 (0.9%) had hospital-onset sepsis. Of 250 events reviewed (133 male patients [53%]; median [IQR] age, 72 [61-80] years), 198 (79%; 95% CI, 74%-84%) were deemed attributable to infections and 52 (21%; 95% CI, 16%-26%) to noninfectious conditions (Table). Crude in-hospital mortality was 93 of 250 events (37%). Respiratory infections predominated (111 of 198 events [56%]), primarily nonventilator hospital-acquired pneumonia (NV-HAP; 88 of 198 events [44%]), followed by primary bloodstream (33 of 198; 17%]) and intra-abdominal infections (29 of 198 events [15%]). Only 25 of 198 events (13%) met surveillance criteria for currently reportable HAIs, and only 44 of 198 events (22%) had positive blood cultures with noncommensal organisms (hospital-onset bacteremia).
Table. Characteristics of Patients With Hospital-Onset Sepsis Events View LargeDownload (opens in new tab)Go to Figure in ArticleCharacteristicPatients, No. (%)Overall hospital-onset sepsis cohort (N = 250)aPreventability cohort (n = 100)Potentially preventable (n = 31)Not likely preventable (n = 69)Age, median (IQR), y72 (61-80)72 (64-80)72 (57-78)73 (64-81)Sex Male133 (53)46 (46)15 (48)31 (45) Female117 (47)54 (54)16 (52)38 (55)Hospital length of stay, median (IQR), d21 (13-36)21 (13-36)19 (14-29)22 (13-37)Location before admission Home207 (83)86 (86)27 (87)59 (86) Health care facility43 (17)14 (14)4 (13)10 (14)In-hospital death93 (37)37 (37)9 (29)28 (41)Discharge to hospice27 (11)12 (12)4 (13)8 (12)Inpatient time to hospital-onset sepsis, median (IQR), d8 (5-13)8 (6-14)10 (2– 14)8 (14-14)Agency for Healthcare Research and Quality Elixhauser Mortality Index, median (IQR)21 (10-35)23 (11-33)21 (5-33)24 (12-34)Infection likelihood Definite66 (26)28 (28)14 (45)14 (20) Probable36 (14)19 (19)6 (19)13 (19) Possible96 (38)53 (53)11 (35)42 (61) No infection52 (21)000Primary pathogen category Bacterial181 (91)92 (92)28 (90)64 (93) Viral10 (5)5 (5)2 (7)3 (4) Fungal6 (3)2 (2)02 (3) Parasitic1 (0)1 (1)1 (3)0Infection source Respiratory111 (56)56 (56)15 (47)41 (53) Abdominal or genitourinary42 (21)20 (20)10 (32)10 (14) Endovascular41 (21)18 (18)3 (9)15 (19) Head and neck or central nervous system3 (2)2 (2)02 (3) Musculoskeletal or skin soft tissue16 (8)7 (7)3 (9)4 (6) Other or unknown14 (7)7 (7)1 (3)6 (9)Reportable health care–associated infectionbClostridium difficile colitis2 (1)000 Central line–associated bloodstream infection10 (4)6 (6)2 (7)4 (6) Catheter-associated urinary tract infection6 (2)2 (2)2 (7)0 Methicillin-resistant Staphylococcus aureus bacteremia3 (1)3 (3)1 (3)2 (3) Colorectal surgical site infection4 (2)4 (4)04 (6) Hysterectomy site infection0000Sepsis organ dysfunctionc Cardiovascular164 (66)71 (71)23 (74)48 (70) Respiratory63 (25)25 (25)11 (35)14 (20) Lactic acidosis126 (50)51 (51)16 (52)35 (51) Kidney20 (8)10 (10)5 (16)5 (7) Hepatic19 (8)10 (10)2 (6)8 (12) Coagulation26 (10)14 (14)5 (16)9 (13)
Among 100 infection-related cases evaluated for preventability, 31 (31%; 95% CI, 23%-41%) were judged potentially preventable (Likert scores: 1, n = 0; 2, n = 10; 3, n = 21; 4, n = 26; 5, n = 35; and 6, n = 8). The most common opportunities (n = 58) included suboptimal management of recognized infections before progression to sepsis (22 of 58 events [38%]), lack of NV-HAP prevention (22 of 58 events [38%]), and gaps in general medical care (6 of 58 events [10%]) (Figure). Lapses in traditional device-related infection prevention bundles were uncommon (5 of 58 events [9%]).
Figure. Bar Graphs of Opportunities for Improvement Among Potentially Preventable Hospital-Onset Sepsis Events View LargeDownload (opens in new tab)Go to Figure in ArticleGraphs show lapses in clinical care identified among patients with potentially preventable sepsis cases.
Specific missed opportunities in early infection management included delays in subspecialty consultation (7 of 58 events [12%]), source control (6 of 58 events [10%]), and microbiologic investigation (3 of 58 events [5%]). NV-HAP prevention gaps included delays in dysphagia evaluation (5 of 58 events [9%]), missed oral care (5 of 58 events [9%]), lack of aspiration precautions (4 of 58 events [7%]), oversedation (3 of 58 events [5%]), inadequate ileus and emesis management (2 of 58 events [3%]), and nosocomial respiratory virus transmission (2 of 58 events [3%]). General medical care gaps included inadequate treatment of constipation leading to urinary retention or stercoral colitis catalyzing sepsis (3 of 58 events [5%]) and suboptimal management of hypervolemia leading to progressive respiratory failure (1 of 58 events [2%]).
Among the 52 noninfectious hospital-onset sepsis events, common precipitants included bleeding (7 of 52 events [14%]), cardiac ischemia or arrhythmia (7 of 52 events [14%]), and hypovolemia (6 of 52 events [12%]). All these events were treated with antibiotics, most for 4 or more days, despite no infection.
Discussion
In this cohort study, we evaluated 250 hospital-onset sepsis events across 9 hospitals and found that most were attributable to respiratory, bloodstream, and intra-abdominal infections; only 13% met current US criteria for reportable HAIs. Among the 100 infection-related sepsis events reviewed for preventability, approximately one-third were deemed potentially preventable. Principal opportunities included optimizing management of recognized infections to prevent progression to sepsis, instituting measures to prevent NV-HAP, and more aggressive control of emesis, constipation, urinary retention, and volume status.
Our estimate that only one-third of hospital-onset sepsis events were potentially preventable is on the lower end of preventability estimates for HAIs, possibly reflecting the heterogeneous pathways leading to hospital-onset sepsis and the complexity of affected patients.7,8 Notably, one-fifth of hospital-onset sepsis events were attributable to noninfectious conditions, highlighting that hospital-onset sepsis surveillance can potentially inform antimicrobial stewardship and inpatient quality improvement programs beyond sepsis alone.9 Limitations of our study include small sample size owing to the resource-intensive preventability adjudication process, subjectivity in retrospective preventability assessment, and risk of cognitive bias despite structured consensus adjudication.10
Hospital-onset sepsis events identify serious complications of hospitalization that are largely missed by current HAI surveillance programs yet are highly morbid and often preventable. Hospital-onset sepsis surveillance could provide a more comprehensive framework for identifying preventable harms and expanding opportunities for quality improvement.
Back to top
Article Information
Accepted for Publication: July 23, 2026.Published: September 17, 2026. doi:10.1001/jamanetworkopen.2026.34324Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License, which does not permit alteration or commercial use, including those for text and data mining, AI training, and similar technologies. © 2026 Manzoor F et al. JAMA Network Open.Corresponding Author: Fizza Manzoor, MD, Department of Population Medicine, Harvard Pilgrim Health Care Institute, 401 Park Dr, Ste 401, Boston, MA 02215 (fmanzoor@mgh.harvard.edu).Author Contributions: Dr Manzoor 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: Manzoor, Klompas, Rhee.Acquisition, analysis, or interpretation of data: All authors.Drafting of the manuscript: Manzoor, Klompas, Rhee.Critical review of the manuscript for important intellectual content: All authors.Statistical analysis: Manzoor, Chan.Obtained funding: Klompas, Rhee.Administrative, technical, or material support: Seetharaman.Supervision: Klompas, Rhee.Conflict of Interest Disclosures: Dr Klompas reported receiving grants from Centers for Disease Control and Prevention (CDC) and personal fees from UpToDate outside the submitted work. Dr Rhee reported receiving grants from CDC and personal fees from UpToDate and DynaMed outside the submitted work. No other disclosures were reported.Funding/Support: This work was supported by the Agency for Healthcare Research and Quality (1R01HS0269660-01) and the Thomas O. Pyle Fellowship in the Department of Population Medicine at Harvard Pilgrim Health Care Institute.Role of the Funder/Sponsor: The funders 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.Data Sharing Statement: See Supplement 2.
References
1.National Healthcare Safety Network. CDC/NHSN surveillance definitions for specific types of infections. January 2026. Accessed August 4, 2026. https://www.cdc.gov/nhsn/pdfs/pscmanual/17pscnosinfdef_current.pdf2.Ginestra
JC, Coz Yataco
AO, Dugar
SP, Dettmer
MR. Hospital-onset sepsis warrants expanded investigation and consideration as a unique clinical entity. Chest. 2024;165(6):1421-1430. doi:10.1016/j.chest.2024.01.028PubMedGoogle ScholarCrossref3.Manzoor
F, Rhee
C, Klompas
M. Advancing surveillance of health care-associated infections: targeting hospital-onset sepsis. N Engl J Med. 2025;393(15):1452-1454. doi:10.1056/NEJMp2507138PubMedGoogle ScholarCrossref4.Millham
L, Gupta
S, Klompas
M, Chan
CA, Dantes
RB, Rhee
C. Evaluating potential modifications to the Centers for Disease Control and Prevention’s adult sepsis event definition: impact on sepsis incidence, outcomes, and clinical validity. Infect Control Hosp Epidemiol. 2025;46(9):880-887. doi:10.1017/ice.2025.10218PubMedGoogle ScholarCrossref5.Rhee
C, Zhang
Z, Kadri
SS,
et al. Sepsis surveillance using Adult Sepsis Events Simplified eSOFA criteria versus Sepsis-3 Sequential Organ Failure Assessment criteria. Crit Care Med. 2019;47(3):307-314. doi:10.1097/CCM.0000000000003521PubMedGoogle ScholarCrossref6.Centers for Disease Control and Prevention. Hospital toolkit for adult sepsis surveillance. May 2018. Accessed January 19, 2026. https://www.cdc.gov/sepsis/media/pdfs/Sepsis-Surveillance-Toolkit-Aug-2018-508.pdf7.Dantes
RB, Rock
C, Milstone
AM,
et al. Preventability of hospital onset bacteremia and fungemia: a pilot study of a potential healthcare-associated infection outcome measure. Infect Control Hosp Epidemiol. 2019;40(3):358-361. doi:10.1017/ice.2018.339PubMedGoogle ScholarCrossref8.Leekha
S, Robinson
GL, Jacob
JT,
et al; CDC Prevention Epicenters Program. Evaluation of hospital-onset bacteraemia and fungaemia in the USA as a potential healthcare quality measure: a cross-sectional study. BMJ Qual Saf. 2024;33(8):487-498. doi:10.1136/bmjqs-2023-016831PubMedGoogle ScholarCrossref9.US Centers for Disease Control and Prevention. Hospital sepsis management and practices: findings from the 2024 NHSN annual survey. US Department of Health and Human Services. 2025. Accessed February 24, 2026. https://www.cdc.gov/sepsis/media/pdfs/hospital-2024-annual-survey-508.pdf10.Hayward
RA, Hofer
TP. Estimating hospital deaths due to medical errors: preventability is in the eye of the reviewer. JAMA. 2001;286(4):415-420. doi:10.1001/jama.286.4.415ArticlePubMedGoogle ScholarCrossref
jamanetwork.com