Early Management of Sepsis

Prompt recognition and appropriate early management of sepsis reduces mortality. There are 2 major goals of early sepsis management, which should be pursued urgently and in parallel—(1) infection control and (2) hemodynamic resuscitation. Treating infection involves timely antimicrobial therapy, and for some infections, source control. Hemodynamic resuscitation should start with a rapid bolus of balanced crystalloid followed by personalized approaches that tailor fluid and vasopressor therapy to evolving patient physiology. In patients who do not improve with initial antibiotic therapy and resuscitation, it is essential to reassess adequacy of infection control and resuscitation.

Key points

•
Early sepsis management has 2 major goals: (1) infection control and (2) hemodynamic resuscitation.
•
Infection management should focus on prompt and appropriate antimicrobial therapy and source control. Clinicians should re-evalaute for adequacy of infection control if patients do not improve.
•
Patients with hypoperfusion benefit from a fluid bolus, followed by personalized fluid resuscitation guided by dynamic assessments of fluid responsiveness.
•
Patients with persistent hypotension despite fluid resuscitation should be started on norepinephrine, with vasopressin and corticosteroids added for worsening shock.
•
Noninvasive hemodynamic management, with peripheral vasopressors and noninvasive blood pressure measurements, is feasible for many patients.

Abbreviations

AngII	angiotensin II
CKD	chronic kidney disease
CO	cardiac output
CRT	capillary refill time
CVC	central venous catheter
ED	emergency department
HFrEF	heart failure with reduced ejection fraction
IR	interventional radiology
IV	intravenous
LR	Lactated Ringer’s
MAP	mean arterial pressure
NS	normal saline
SBP	systolic blood pressure
SSC	Surviving Sepsis Campaign
SV	stroke volume

Introduction

Sepsis is a leading cause of morbidity and mortality, resulting in over 11 million deaths globally in 2017 and accounting for 1 in every 2 to 3 hospital deaths in the United States. ^,^, Sepsis is defined as life-threatening acute organ dysfunction due to a dysregulated host response to infection. In practice, sepsis diagnosis is based on presence of both infection and acute organ dysfunction. Prompt recognition and appropriate early management are associated with reduced mortality. There are 2 major goals of early sepsis management, which should be pursued urgently and in parallel: (1) treatment of the infection and (2) hemodynamic resuscitation. Achieving these goals helps halt the damaging host response and prevent progressive organ dysfunction. In this article, we review key aspects of early sepsis management related to both infection management and resuscitation.

Scope

While early sepsis recognition is essential for timely and effective treatment, approaches to sepsis diagnosis are outside the scope of this article.

Case

A 62-year-old woman with hypertension, heart failure with reduced ejection fraction (HFrEF), and chronic kidney disease (CKD) presents to the emergency department (ED) with 3 days of pain with urination, 2 days of bloody urine, and 1 day of fevers, chills, and left flank pain. On presentation, blood pressure is 84/52 mm Hg, heart rate is 110 beats per minute, and respiratory rate is 24 breaths per minute. On examination, she is confused and has trouble following commands. Based on the clinical presentation consistent with urinary infection and acute organ dysfunction (altered mental status, hypotension), you diagnose sepsis. What should you do next?

Treating Infection

Antimicrobial therapy

Early initiation of appropriate antimicrobial therapy is a core tenet of sepsis management, which has consistently been linked to decreased mortality. ^,^, The Surviving Sepsis Campaign (SSC) guidelines recommend administering antimicrobial therapy within 1 hour of recognition for patients with definite or probable sepsis, as well as in patients with shock and possible sepsis. ^, For patients with possible sepsis without shock, the guidelines suggest a rapid assessment of infectious and noninfectious causes of illness, with antibiotic administration within 3 hours if concern for sepsis persists. While the 1- and 3-hour thresholds are helpful to guide management and drive performance improvement, antibiotic timing is a continuous measure without a clear threshold effect. The absolute increase in mortality with each hour of delay is greater in patients with shock, ^, though earlier antibiotic administration is associated with reduced mortality for all patients with sepsis. Clinicians should therefore evaluate possible sepsis urgently, and start antibiotic therapy as soon as possible after sepsis is deemed likely, ideally within 1 hour ( Fig. 1 ).

Alt text begins. A flowchart depicts the approach to infection management in patients with suspected sepsis, showing initial diagnostic evaluation, initial management, and subsequent evaluation and management. Alt text provided by Elsevier. — Fig. 1

Timely antibiotic initiation is a complex, multidisciplinary process—the clinician must order the antibiotics, the pharmacist must prepare the drugs (or have it available on the unit), and the nurse must see the order, get the medication, ensure available intravenous (IV) access, and start the infusion. Even with this complex process, antibiotic delays are often driven by delayed sepsis recognition and antibiotic ordering rather than delays between ordering and administration. Indeed, in a large study of Veteran Affairs hospitals, the median time from ED antibiotic order to administration was 23 minutes (interquartile range 15 to 34 minutes). Therefore, initiatives aimed at decreasing time-to-antibiotics should focus on sepsis recognition and expedited antibiotic ordering, in addition to streamlining postorder processes and ensuring that new antibiotic orders default to immediate administration.

While some experts have raised concern that efforts to shorten time to antibiotic administration in sepsis may worsen antimicrobial stewardship, such effects have not been demonstrated. In a study of over 1 million patients at 152 hospitals, as time to antimicrobial administration for sepsis decreased over time, there were concurrent reductions in overall antimicrobial use and broad-spectrum antibiotics. Furthermore, in a large multicenter study of physician variation in door-to-antibiotic time, faster antibiotic prescribing was not associated with overtreatment. These findings indicate that timely antibiotic initiation can be balanced with antimicrobial stewardship.

Blood culture timing

Blood cultures should be drawn before initiating antibiotic therapy ^,^, since cultures collected after antibiotic initiation have lower yield and reduced ability to detect blood stream infection. However, the mortality benefit of prompt antibiotic therapy outweighs the potential benefit of increased blood culture yield, especially given culture-negative sepsis is common. Therefore, in the event that blood cultures cannot be drawn in a timely manner (eg, due to limited resources or difficult access), clinicians should proceed with antibiotic therapy.

Antibiotic selection

Prompt antibiotic therapy can reduce mortality only if it is effective—ie, the therapy adequately covers the inciting pathogen. Given the inciting pathogen is rarely known at time of initial treatment, clinicians should aim to maximize the likelihood that antibiotic therapy will be effective by selecting empiric antibiotics to cover likely potential pathogens based on (1) suspected site of infection, (2) patient-specific risk factors for resistant or atypical pathogens, (3) the patient’s prior culture results, and (4) local epidemiology of resistance.

In addition to ensuring that the selected empiric antibiotic therapy covers likely potential pathogens, it is also important to avoid or limit coverage of unlikely pathogens. In particular, there is mounting experimental and observational human evidence showing the detrimental effects of empiric anaerobic coverage. ^, Antianaerobic antibiotics deplete the healthy gut microbiome and are associated with increased 90-day mortality—likely mediated by increased risk for secondary or recurrent infection. ^, In the ACORN trial comparing empiric antibiotics, mortality did not differ by randomization to cefepime (anaerobe-sparing) versus piperacillin-tazobactam (anaerobe-depleting). However, this trial was not designed to test the impact of anaerobic coverage, since many patients in the cefepime arm received cotreatment with metronidazole. Furthermore, the trial examined outcomes to only 14 days, before mortality benefit with anaerobe-sparing regimens would be expected. ^, In general, unnecessary antianaerobic coverage should be avoided for sites of infection where anaerobic pathogens are unlikely, such as respiratory or urinary tract.

Other antimicrobial therapy

While we have focused on antibiotics, sepsis can be caused by pathogens other than bacteria, including fungi, particularly Candida species, and viruses. In retrospective analyses of patients with Candida infections, delays in time-to-antifungal therapy have been associated with higher mortality. ^, However, prospective trials have not found benefit of empiric antifungal therapy. The EMPIRICUS trial randomized 260 patients with empirical antifungal treatment in ICUS-acquired sepsis to addition of empiric micafungin versus placebo and found no difference in fungal infection–free survival to day 28. However, few patients ended up having confirmed fungal infection (5%), which limited the power of this trial to detect potential benefit of empiric micafungin and highlights the challenge of predicting fungal infection in practice. Empiric antifungal therapy should be considered on a case-by-case basis in high-risk patients, particularly those not showing clinical improvement with initial antibiotic therapy, but should not be used across the board for empiric treatment of hospital-onset sepsis. ^, Given the rise of azole-resistant Candida species, echinocandins (eg, micafungin or anidulafungin) are the preferred empiric antifungal agents in many regions.

Viruses can also cause sepsis. Antiviral therapy has been associated with reduced mortality in some viral infections (eg, oseltamivir in influenza). ^, However, empiric antiviral therapy is rarely indicated since rapid diagnostic testing can identify viral infections quickly and guide initiation of antiviral therapy in a timely manner. Instead, a common clinical challenge for patients with positive viral testing is determining whether to treat with empiric antibiotic therapy for possible concomitant bacterial infection. Given there is little room for error in omitting antibiotics in critically ill patients with potential bacterial coinfection, empiric antibiotic therapy should be considered in patients with positive viral testing and shock or severe respiratory failure if there is any concern for concomitant bacterial infection. ^, However, antibiotic therapy should be reassessed daily and discontinued if subsequent testing (eg, serial procalcitonin) suggest bacterial coinfection is unlikely.

Source control

Even with appropriate antibiotic therapy, some infections require source control interventions for timely cure. Source control encompasses surgical and procedural interventions to drain or remove the nidus of infection (eg, appendectomy or removal of infected hardware) or correct anatomic blockages that impede infection clearance (eg, endoscopic retrograde cholangiopancreatography for biliary stricture). Many infections require source control, including intrabdominal infections (eg, appendicitis), infections proximal to an obstruction (eg, obstructive kidney stone, cholangitis), infected devices (eg, infected central lines), empyema, and abscesses. Delayed source control in these scenarios is associated with increased risk of mortality, particularly among patients with shock. ^, For example, in a multicenter study of nearly 5000 patients with sepsis who underwent source control procedures, later source control (after 6 hours) was associated with higher mortality than early source control, particularly for abdominal and soft tissue infections. Given the importance of source control, the initial diagnostic workup for infection should seek to clarify the site of infection and assess whether source control should be prioritized after initial stabilization of the patient.

Re-evaluation: missed bug? Missed source?

Most patients with sepsis begin to show signs of clinical improvement after resuscitation and initiation of appropriate antimicrobial therapy. When a patient does not stabilize or improve during the initial hours of treatment, it is essential to consider whether the initial infection control measures were sufficient. For patients with worsening clinical trajectories or persistent clinical instability (eg, ongoing shock without improvement), clinicians should consider the following questions: (a) Was the initial antimicrobial therapy appropriate? Or should it be broadened to cover additional pathogens (eg, methicillin-resistant Staph aureus ) or expanded to include double coverage for potential resistant gram-negative bacteria? (b) Is there an additional site of infection beyond what has been identified? (c) Is there an indication for source control? Or, has the need for source control been adequately ruled out?

Back to the Case (Infection management) : We ordered blood cultures, urine culture, and urinalysis to evaluate the suspected source of infection urinary tract infection (UTI ) and completed a history, physical examination, and chart review with particular attention to prior culture data and risk factors for resistant organisms, which were not present. We started empiric antibiotic therapy with Ceftriaxone to cover likely urinary pathogens. We started antibiotics as quickly as possible by ordering them promptly and coordinating with pharmacy and nursing to ensure timely initiation (goal <1 hour). The patient had worsening shock despite initial treatment, which prompted re-evaluation and a computed tomography scan ( CT scan ) of the abdomen. CT revealed an obstructing left kidney stone with proximal hydronephrosis and evidence of associated left pyelonephritis. Interventional radiology (IR) was consulted emergently for the placement of a percutaneous nephrostomy tube. The patient’s shock improved within a few hours of this source control procedure.

Restoring Organ Perfusion (Resuscitation)

In sepsis, systemic inflammation can lead to vasodilation, which decreases preload, causes hypotension, and eventually impairs organ perfusion. Patients with sepsis also often have concomitant hypovolemia from dehydration and fluid losses (eg, diarrhea, insensible losses from fever). Patients with sepsis-induced hypoperfusion require prompt resuscitation to restore organ perfusion and prevent progressive organ damage. Initial clinical evaluation should focus on identifying evidence of hypotension and hypoperfusion (eg, delayed capillary refill), while initial laboratory evaluation should include lactate measurement to evaluate for occult hypoperfusion ( Fig. 2 ).

Alt text begins. A flowchart shows the approach to hemodynamic resuscitation in sepsis, outlining initial diagnostic evaluation, initial management paths based on patient status, and ongoing management steps. Alt text provided by Elsevier. — Fig. 2

Case : Let’s come back to the beginning of our case to focus on the patient’s hemodynamic resuscitation. In the case, the patient is presenting with sepsis due to a urinary source and has an initial blood pressure of 84/52 mm Hg (mean arterial pressure [MAP] 63 mm Hg). How should we resuscitate this patient?

Fluid resuscitation

Initial fluid resuscitation

IV fluids are a key component of sepsis resuscitation, as they help correct intravascular volume depletion, restore preload, and improve organ perfusion. Fluid resuscitation is indicated for patients with sepsis who have hypotension (eg, systolic blood pressure [SBP]<90 mm Hg or MAP<65 mm Hg) or lactate greater than 4 mmol/L. Fluid resuscitation should also be considered in patients with relative hypotension (ie, SBP below known baseline), intermediate lactate elevation (lactate 2–4 mmol/L), and other measures of end-organ dysfunction (eg, abnormal capillary refill time [CRT]).

The goal of resuscitation is to improve blood pressure and restore organ perfusion as soon as possible. The optimal fluid volume to support this goal varies across patients. Both under-resuscitation and over-resuscitation are associated with worse outcomes, ^,^, so clinical status should be monitored closely to guide ongoing resuscitation.

The 2026 SSC guidelines suggest an initial bolus of 30 mL/kg body weight (or adjusted body weight if obese ) as the default practice, which translates to between 1.5 and 4L for most adult patients. This pragmatic recommendation is based on observational studies ^,^,^,^, and pre-enrollment fluid volumes in major trials ^,^, —not trials directly comparing 30 mL/kg versus other initial fluid volumes. Observational studies suggest optimal outcomes with an initial volume around 30 mL/kg, with both lower and higher volumes (typically >5L or >70 mL/kg) associated with worse outcomes. ^,^,^,^,^, However, by issuing a conditional (rather than strong) recommendation on initial fluid volume, the SSC guidelines acknowledge the importance of considering the clinical context when selecting initial fluid bolus volume. For patients at very high risk of harm from fluid overload or a clinical evaluation concerning for mixed shock, a smaller initial bolus and sooner transition to personalized resuscitation guided by serial dynamic assessments may be warranted (see Fig. 2 ).

The initial fluid bolus should be given rapidly to restore perfusion as quickly as possible. Stuttering resuscitation approaches—whereby smaller volumes are given intermittently over longer time periods—are suboptimal and can delay time to achieving intravascular repletion ( Fig. 3 ). Likewise, aborted resuscitation approaches—where fluid is capped at an arbitrary volume due to risk for fluid overload—are also suboptimal because clinical history (eg, history of HFrEF or CKD) and physical examination features (eg, edema) are unreliable in determining fluid responsiveness. Instead, patients with hypoperfusion but high risk for fluid-related harms should be transitioned more quickly to personalized resuscitation approaches guided by serial assessment of fluid responsiveness, as discussed in the next section.

Alt text begins. A diagram illustrates the approaches to initial fluid resuscitation, comparing guideline-recommended methods with suboptimal methods like stuttering and aborted resuscitation. Alt text provided by Elsevier. — Fig. 3

Ongoing fluid resuscitation

After initial fluid resuscitation with a weight-based fluid bolus, clinicians should use dynamic assessments of fluid responsiveness to guide ongoing resuscitation.

Prior trials comparing fluid-liberal versus fluid-restrictive approaches to ongoing resuscitation yielded neutral results, suggesting that neither approach is optimal as a general strategy ( Table 1 ). The CLOVERS trial randomized 1563 patients with sepsis-induced hypotension to restrictive fluids (paired with early vasopressors) versus liberal fluids (up to 5 L) in the first 24 hours of presentation. The CLASSIC trial randomized 1554 patients with septic shock to a restrictive approach to ICU fluid management (no maintenance fluids, boluses for only limited clinical criteria) versus usual care for the duration of ICU admission. Both studies found no difference in mortality.

Table 1

Summary of landmark trials in early sepsis resuscitation

Trial	Population	Intervention	Comparator	Outcome	Takeaway
Fluid resuscitation
CLASSIC, 2022	1554 patients with septic shock in the ICU [Europe]	Restrictive fluid strategy throughout ICU admission (no maintenance fluid, strict criteria for boluses)	Usual care	90-d mortality: 42.3% vs 42.1%, P =.96	Restrictive strategies to managing fluid in the ICU do not reduce mortality. Patients received a median of 3L of fluid before enrollment.
CLOVERS, 2023	1563 patients with sepsis-induced hypotension, mostly enrolled in ED [United States]	Restrictive fluid, early vasopressors in first 24 h	Liberal fluid (up to 5 L) in first 24 h	Death before discharge home by day 90: 14.0% vs 14.9%, P =.61	Generalized restrictive vs liberal strategies for ongoing fluid resuscitation in the first 24 h result in similar outcomes. Patients received a median 2L of IV fluid before enrollment.
FRESH, 2020	124 patients with septic shock in the ICU [United States]	SV-guided fluid resuscitation (personalized approach)	Usual care	72-h fluid balance: 0.65 vs 2.02 L RRT: 5.1% vs 17.5%, P =.04. Mechanical ventilation: 17.7% vs 34.1%, P =.04	Personalizing ongoing fluid resuscitation based on SV assessments may decrease fluid overload and improve patient outcomes. Patients received a median of 2.5 L fluid before enrollment.
Fluid choice
SMART, 2018	7942 patients in the ICU [United States]	Balanced fluid	NS	Major adverse kidney event: 14.3% vs 15.4%, P =.04	Balanced fluid may improve patient outcomes. In secondary analyses, larger effects were seen in sicker patients, patients with sepsis, and when balanced fluid was used from the onset of resuscitation (starting in the ED). ^,
PLUS, 2022	5037 patients in the ICU [Australia/New Zealand]	Balanced fluid	NS	90-d mortality : 21.8% vs 22.0%, P =.90	No difference in mortality with fluid choice, though all patients were enrolled in the ICU, potentially missing the effects of starting balanced fluid in the ED that was seen in secondary analyses of SMART.
BaSICs, 2023	11,052 patients in the ICU [Brazil]	Balanced fluid	NS	90-d mortality : 26.4% vs 27.2%, P =.47	No difference in mortality with fluid choice, though secondary analysis suggest benefit with balanced fluids for patients with sepsis and when balanced fluid was used from the onset of resuscitation (starting in the ED).
Vasopressor selection
SOAP II, 2010	1679 patients with septic, cardiogenic, or hemorrhagic shock [Europe]	Norepinephrine	Dopamine	28-d mortality: 48.5% vs 52.5%, P =.10. Results were statically significant favoring norepinephrine in patients with cardiogenic shock. Arrhythmias: 12.4% vs 24.1%, P <.001	First-line norepinephrine carries less risk of arrhythmias than dopamine and may be associated with lower mortality. Based on this trial, norepinephrine became widely adopted as the preferred first-line vasopressor in shock, including septic shock.
VASST, 2008	778 patients with septic shock in the ICU [Canada, Australia, United States]	Norepinephrine + Vasopressin	Norepinephrine	28-d mortality : 35.4% vs 39.3%, P =.26. Results were statistically significant in patients with less severe shock (norepinephrine doses of 5–14 mcg/min): 26.5% vs 35.7%, P =.05.	Based on the VASST trial, many clinicians favor earlier vasopressin initiation in patients with less severe shock.
Adjunctive therapies—corticosteroids
ADRENAL, 2018	3800 patients with septic shock on mechanical ventilation and vasopressors at any dose for ≥4 h [Europe, Australia, Saudi Arabia]	Hydrocortisone (continuous, 200 mg/day)	Placebo	90-d mortality: 27.9% vs 28.8%, P =.50. Resolution of shock: 3 vs 4 d, P <.001.	Taken together, these trials suggest that corticosteroids reduce the duration of septic shock and in patients on higher doses of vasopressors (≥0.25 mcg/kg/min norepinephrine, as in APROCCHSS), may also improve mortality.
APROCCHSS, 2018	1241 patients with septic shock on vasopressor dose ≥0.25 mcg/kg/min of norepinephrine or equivalent for ≥6 h [France]. This trial included a factorial design with activated protein C.	Hydrocortisone + Fludrocortisone	Placebo	90-d mortality: 43.0% vs 49.1%, P =.03. Vasopressor-free days: 17 vs 15, P <.001
Resuscitation Targets
SEPSISPAM, 2014	776 patients with septic shock [France]	MAP target: 65–70 mm Hg	MAP target: 80–85 mm Hg	28-d mortality : 34.0% vs 36.6%, P =.57 Atrial fibrillation: 2.8% vs 6.7%, P =.02	MAP ≥65 mm Hg became the accepted target for resuscitation.
The 65 Trial, 2020	2600 patients ≥ 65 year old with vasodilatory hypotension [UK]	MAP target: 60–65 mm Hg	MAP target: ≥65 mm Hg	90-d mortality: 41.0% vs 43.8%, P =.15	Lower MAP targets (60–65 mm Hg) are safe in older patients.
ANDROMEDA-SHOCK, 2019	424 patients with septic shock [South America]	Resuscitation protocol aimed at normalizing CRT	Resuscitation protocol aimed at normalizing lactate	28-d mortality: 34.9% vs 43.4%, P =.06. Among patients with lower APACHE II, HR 0.61 (0.39–0.96).	Targeting resuscitation to normalizing peripheral perfusion (CRT) may improve outcomes compared to targeting lactate clearance.
ANDROMEDA-SHOCK 2, 2025	1467 patients with septic shock [Americas, Europe, Asia]	Personalized resuscitation protocol targeted to CRT	Usual care	WIN ratio of death, duration of vital support, and hospital length of stay: 1.16 (95% CI: 1.02–1.33), favoring intervention	Personalized hemodynamic resuscitation targeted to normalizing peripheral perfusion (CRT) may improve patient outcomes.
Hemodynamic monitoring
EVERDAC, 2025	1010 ICU patients on vasopressors [France]	Invasive strategy: insertion of arterial catheter within 4 h of randomization	Noninvasive strategy: delayed insertion of arterial catheter. Arterial catheter only placed for key safety criteria .	28-d mortality: 36.9% vs 34.3%, P =.006 for noninferiority	A noninvasive strategy involving placing an arterial catheter only for certain safety criteria is noninferior to an early invasive strategy where arterial catheters are routinely placed in all patients with shock.

Abbreviations: CI, confidence interval; ED, emergency department; EGDT, early goal-directed therapy; HR, hazard ratio; RRT, renal replacement therapy; SEPSISPAM, Assessment of Two Levels of Arterial Pressure on Survival in Patients With Septic Shock.

By contrast, trials evaluating personalized approaches to resuscitation that include serial assessments of fluid responsiveness have resulted in improved outcomes. ^, Dynamic assessments of fluid responsiveness are superior to static measures (eg, central venous pressure, inferior venous cava collapsibility), which reflect the patient’s volume status but not their potential to respond to fluid. Dynamic assessments involve 2 steps: (1) a fluid challenge, which can be performed with either passive leg raise or small 250 to 500 mL test bolus and (2) measurement of change in cardiac output (CO) or stroke volume (SV) with the fluid challenge. There are several validated tools for measuring change in CO and SV, which range from noninvasive technologies like bioreactance (Starling [Baxter]; formerly NICOM [Cheetah Medical]) to technologies that require central venous catheters [CVCs] and/or arterial catheters (transpulmonary thermodilution, PiCCO [Getinge]; pulse contour analysis, Flotrac [Edwards]) or a clinician trained in ultrasound. ^, A recent meta-analysis of several smaller studies found fluid resuscitation guided by dynamic assessments performed using any of these validated tools was associated with reduced mortality. For example, the FRESH trial randomized 124 patients with sepsis-induced hypotension to SV-guided fluid resuscitation, using bioreactance-based serial dynamic assessments, versus usual care (see Table 1 ). Patients randomized to SV-guided resuscitation had less intubation (17.7% vs 34.1%, P =.04), less dialysis (5.1% vs 17.5%), and numeric decreases in mortality and ICU length of stay. Further supporting the value of serial dynamic assessments, the ANDROMEDA-SHOCK 2 trial, discussed below in the Resuscitation Targets section, found that a hemodynamic-guided resuscitation protocol that included dynamic assessments of fluid responsiveness improved patient outcomes compared to usual care.

In the FRESH trial, SV-guided fluid resuscitation was started after initial fluid resuscitation (approximately 2.5 L or around 25 mL/kg). However, this strategy can be used earlier, particularly in patients at high risk of fluid-related harms. Earlier use of personalized fluid strategies may be particularly helpful in patients with hospital-onset sepsis, who have different characteristics and for whom there are little trial data to guide resuscitation (many landmark sepsis trials have focused on patients presenting to the ED with community-onset sepsis. )

Type of fluid

Balanced solutions, such as Lactated Ringer’s (LR), should be used for resuscitation based on accruing data indicating a likely mortality benefit with balanced solutions versus 0.9% normal saline (NS). In the cluster-randomized SMART trial, critically ill patients randomized to balanced solutions had a lower incidence of major adverse kidney events (the primary composite outcome) (see Table 1 ). In the prespecified subgroup of 1641 patients with sepsis, patients randomized to balanced solutions had lower 30-day mortality (26.3% vs 31.1%, P =.01), which was consistent in both unadjusted and adjusted analyses. The effect was strongest (and only detectable) when balanced solutions were used throughout resuscitation—suggesting that resuscitation should start with LR and not merely convert to LR in the event of metabolic derangements with NS. Secondary analyses of the individually randomized BaSICs trial comparing NS versus LR and the CLOVERS trial of fluid resuscitation strategies likewise suggest the importance of using LR throughout resuscitation. ^,^,^,

Subsequent large-scale implementation studies have confirmed the feasibility and benefit of using LR in real-world practice. ^, LR is also associated better outcomes in other conditions beyond sepsis, such as diabetic ketoacidosis and sickle cell vaso-occlusive episodes—suggesting that LR can be the default IV fluid for most conditions, simplifying implementation. ^, The only exception is for patients with traumatic brain injury, where balanced solutions are associated with harm, likely due to their hypotonicity and risk for increased intracranial pressures. Pragmatic algorithms for addressing drug incompatibilities with LR can help to improve implementation ( Fig. 4 ).