NEWS2 versus shock status to stratify antibiotic urgency in patients with suspected sepsis: a retrospective, multicentre cohort study.
The Surviving Sepsis Campaign guidelines stratify antibiotic urgency in suspected sepsis by shock status, recommending treatment within 1 h for patients with shock and within 3 h for possible sepsis without shock. UK National Institute for Health and Care Excellence (NICE) guidelines use National Early Warning Score 2 (NEWS2) risk categories, with a window of 1 h for the group at highest risk of mortality (aggregate score ≥7). We aimed to compare these strategies for identifying patients in whom antibiotic delays are associated with mortality.
We conducted a retrospective cohort study between May 31, 2015, and Feb 28, 2024, of adults (age ≥18 years) admitted via the emergency departments of nine hospitals within the Mass General Brigham health-care system (MA, USA) and treated for suspected infection, defined as blood culture sampling and intravenous antibiotic administration within 6 h of emergency department arrival. Exclusion criteria included transition to palliative care or death within 6 h of arrival to emergency department, transfer from acute care hospitals, psychiatric or obstetric admissions, missing vital signs or key laboratory test results within 12 h of arrival, oral or intravenous antibiotic receipt before arrival, insufficient data to calculate NEWS2 scores, and positive SARS-CoV-2 PCR test results within 2 days of arrival. We estimated hourly associations between time to antibiotic administration and hospital mortality using multivariable logistic regression with generalised estimating equations, stratified by shock (defined as persistent hypotension and lactate >2 mmol/L) and by laboratory-upgraded NEWS2 scores (NEWS2+), which were calculated per UK NICE guidance by shifting patients up one category if they had neutropenia, lactate concentration higher than 2 mmol/L, or laboratory evidence of new organ dysfunction within 3 h of arrival.
115 464 encounters with suspected infection were identified, of which 15 797 were ineligible based on exclusion criteria. The final analysis cohort included 71 593 patients who contributed 99 667 encounters (52 835 [53%] male, 46 828 [47%] female). 4867 (4·9%) encounters had shock. NEWS2+ scores were 0 in 9672 (9·7%) encounters, 1-4 in 37 310 (37·4%) encounters, 5-6 in 26 134 (26·2%) encounters, and ≥7 in 26 551 (26·6%) encounters. Each additional hour to antibiotic administration was associated with higher mortality in encounters with shock (adjusted odds ratio [OR] 1·15, 95% CI 1·07-1·25; 1·5% absolute increase per hour, 95% CI 0·5-2·4), but not in those without shock (1·03, 1·00-1·05). Delays were also associated with higher mortality in encounters with NEWS2+ ≥7 (1·07, 1·04-1·11; 0·5% absolute increase per hour, 95% CI 0·3-0·8), including 22 902 (86·3%) of 26 551 encounters without shock (1·05, 1·01-1·09; 0·4% absolute increase per hour, 95% CI 0·1-0·6%). No association was observed in encounters with NEWS2+ scores less than 7.
The NICE-defined high-risk NEWS2 category (scores ≥7) identified patients in whom each additional hour to antibiotic administration was associated with increased mortality, including many without shock, whereas no association was observed for lower NEWS2 scores. These findings suggest that NEWS2 might better target urgent antibiotic therapy beyond shock-based assessment alone.
US Agency for Healthcare Research and Quality and US Centers for Disease Control and Prevention.
We conducted a retrospective cohort study between May 31, 2015, and Feb 28, 2024, of adults (age ≥18 years) admitted via the emergency departments of nine hospitals within the Mass General Brigham health-care system (MA, USA) and treated for suspected infection, defined as blood culture sampling and intravenous antibiotic administration within 6 h of emergency department arrival. Exclusion criteria included transition to palliative care or death within 6 h of arrival to emergency department, transfer from acute care hospitals, psychiatric or obstetric admissions, missing vital signs or key laboratory test results within 12 h of arrival, oral or intravenous antibiotic receipt before arrival, insufficient data to calculate NEWS2 scores, and positive SARS-CoV-2 PCR test results within 2 days of arrival. We estimated hourly associations between time to antibiotic administration and hospital mortality using multivariable logistic regression with generalised estimating equations, stratified by shock (defined as persistent hypotension and lactate >2 mmol/L) and by laboratory-upgraded NEWS2 scores (NEWS2+), which were calculated per UK NICE guidance by shifting patients up one category if they had neutropenia, lactate concentration higher than 2 mmol/L, or laboratory evidence of new organ dysfunction within 3 h of arrival.
115 464 encounters with suspected infection were identified, of which 15 797 were ineligible based on exclusion criteria. The final analysis cohort included 71 593 patients who contributed 99 667 encounters (52 835 [53%] male, 46 828 [47%] female). 4867 (4·9%) encounters had shock. NEWS2+ scores were 0 in 9672 (9·7%) encounters, 1-4 in 37 310 (37·4%) encounters, 5-6 in 26 134 (26·2%) encounters, and ≥7 in 26 551 (26·6%) encounters. Each additional hour to antibiotic administration was associated with higher mortality in encounters with shock (adjusted odds ratio [OR] 1·15, 95% CI 1·07-1·25; 1·5% absolute increase per hour, 95% CI 0·5-2·4), but not in those without shock (1·03, 1·00-1·05). Delays were also associated with higher mortality in encounters with NEWS2+ ≥7 (1·07, 1·04-1·11; 0·5% absolute increase per hour, 95% CI 0·3-0·8), including 22 902 (86·3%) of 26 551 encounters without shock (1·05, 1·01-1·09; 0·4% absolute increase per hour, 95% CI 0·1-0·6%). No association was observed in encounters with NEWS2+ scores less than 7.
The NICE-defined high-risk NEWS2 category (scores ≥7) identified patients in whom each additional hour to antibiotic administration was associated with increased mortality, including many without shock, whereas no association was observed for lower NEWS2 scores. These findings suggest that NEWS2 might better target urgent antibiotic therapy beyond shock-based assessment alone.
US Agency for Healthcare Research and Quality and US Centers for Disease Control and Prevention.
Authors
Rhee Rhee, Singer Singer, Pak Pak, McKenna McKenna, Agan Agan, DelloStritto DelloStritto, Shappell Shappell, Filbin Filbin, Dutta Dutta, Kadri Kadri, Septimus Septimus, Klompas Klompas
View on Pubmed