BMJ Best Practice

Please note that formulations/routes and doses may differ between drug names and brands, drug formularies, or locations. Treatment recommendations are specific to patient groups: see disclaimer

INITIAL

in hospital: sepsis highly suspected and unknown or unclear source of bacterial infection

1st line – broad-spectrum intravenous antibiotics

Start treatment immediately if a senior clinical decision-maker makes a diagnosis of suspected sepsis, based on acute deterioration (e.g., National Early Warning Score 2 [NEWS2] score of 5 or more, or a similar trigger using another validated scoring system) in a patient with known or likely infection.[41][42]

In the UK, a senior clinical decision-maker is CT3/ST3 or higher, or a trained nurse with prescribing rights in acute care.[41]

For any acutely ill and deteriorating patient with a suspected or known bacterial infection and suspected sepsis, above all else prioritise (if needed):[3][43][45][46]

Securing their airway
Correcting hypoxaemia
Establishing venous access for the early administration of antibiotics and fluids.

Where there is evidence of a bacterial infection and a strong suspicion of sepsis (based on acute deterioration [e.g., NEWS2 score of 5 or more, or a similar trigger using another validated scoring system]), give broad-spectrum intravenous antibiotics within 1 hour of the risk of sepsis being recognised.[3][41][43][46] Do this before a pathogen is identified but after blood cultures have been taken.[3][41][43][46]

The Surviving Sepsis Campaign international guideline recommends empirical combination therapy (using at least two antibiotics of different antimicrobial classes) aimed at the most likely bacterial pathogen(s) for the initial management of septic shock.[43]

Follow local policy and consider discussing with microbiology/infectious disease colleagues to determine the most appropriate choice.

Use a 'start smart then focus' approach.[41][136]



More info: Antimicrobial resistance

NHS England recommends following a 'start smart then focus’ approach for antibiotic use in people with sepsis.[41] This is derived from Public Health England guidance, which outlines an evidence-based approach to improving antimicrobial prescribing and stewardship in hospital settings.[136] The prevalence of antimicrobial resistance (AMR) has risen alarmingly over the last 50 years and no new classes of antibiotics have been developed in decades. By 2050 it is estimated that AMR will kill 10 million people per year, more than cancer and diabetes combined.[143] The relationship between antibiotic exposure and antibiotic resistance is unambiguous not only at the population level but also in individual patients.[144][145]

Start smart – in the context of sepsis:[136]

Do not start antimicrobial therapy unless there is clear evidence of infection
Take a thorough drug allergy history
Initiate prompt effective antibiotic treatment within 1 hour of diagnosis (or as soon as possible) in patients with sepsis or life-threatening infections. Avoid inappropriate use of broad-spectrum antibiotics
Comply with local antimicrobial prescribing guidance
Document clinical indication (and disease severity if appropriate), drug name, dose, and route on drug chart and in clinical notes
Include review/stop date or duration
Obtain cultures prior to starting therapy where possible (but do not delay therapy).

Then focus – in the context of sepsis:[136]

Review the clinical diagnosis and the continuing need for antibiotics at 48 to 72 hours* and document in a clear plan of action – the ‘antimicrobial prescribing decision’
The ‘antimicrobial prescribing decision’ options are:
1. Stop antibiotics if there is no evidence of infection
2. Switch antibiotics from intravenous to oral
3. Change antibiotics – ideally to a narrower spectrum, or broader if required
4. Continue and document next review date or stop date
It is essential that the review and subsequent decision is clearly documented in the clinical notes and on the drug chart where possible (e.g., ‘stop antibiotic’).

*In clinical practice, daily prompting about de-escalation is encouraged.

Target the presumed site of infection.[3][43]

If there is no clinical evidence to suggest a specific site of infection but a senior clinical decision-maker strongly suspects the presence of a bacterial infection, still give empirical broad-spectrum intravenous antibiotics.[3][41] Choose an empirical antibiotic based on:[146][147]

Local antibiotic protocols and resistance patterns
- Consult microbiology/infectious disease colleagues to determine the most appropriate choice
The likely causative organism
The patient’s immune function.

Practical tip

Check local policies regarding repeat cultures, which are particularly indicated if there are persistent or repeated fever spikes or if you identify a potential new site of infection. Observations from studies to date support taking as many as four blood culture sets over a 24-hour period for >99% test sensitivity.[148]

Practical tip

If a patient has a mild allergy (e.g., rash) to an unknown antibiotic, you should still give empirical broad-spectrum antibiotics if indicated to prevent delay in the treatment of sepsis, which is likely to worsen outcome. If the antibiotic is known and is part of the empirical protocol for your hospital, discuss with potential alternatives with a microbiologist.



Controversy: 1-hour antibiotic targets

There is widespread agreement that appropriate and timely recognition of sepsis and subsequent resuscitation are key approaches to the management of the severely ill patient with sepsis. However, guideline-derived antibiotic delivery goals (as outlined by the UK National Institute for Health and Care Excellence [NICE], the Surviving Sepsis Campaign [SSC], and the UK Sepsis Trust) have been challenged owing to gaps in the evidence and concerns about over-treatment of the individual patient and the subsequent effect on antimicrobial resistance.[57][58]

The 1-hour antibiotic targets outlined by NICE,[3] the SSC,[43] and the UK Sepsis Trust (Sepsis Six)[46] are derived from data that appear to draw a direct correlation between each hour of delayed treatment of the patient with sepsis and an increased risk of further deterioration or death.[93][149][150] However, examining some of these data closely shows that 1-hour antibiotic targets may not be possible or necessarily advantageous for all patients with eventual sepsis diagnoses.[58]

A retrospective cohort study of hypotensive inpatients with sepsis, published in 2006, first described a potential link between timing of antibiotics and outcomes.[149]
- The authors noted 79.9% survival if septic patients received in vitro-active antibiotics within 1 hour of the onset of hypotension, with a subsequent 7.6% survival rate decrement with each additional hour to treatment.
- Looking closely at the data, however, there was a lower survival rate (52%) among patients who received antibiotics before hypotension compared with those who received them within the first few hours of the onset of septic shock.
- Commentators note that ascribing a biological effect to antibiotics for either the improvement or the decreased survival is endemic to observational or natural experiment designs, which are prone to confounding and bias in their interpretation of results.[58]
More recently, another retrospective cohort study of 3929 patients with severe sepsis described an 8% hourly incremental increased risk of progression to septic shock with longer time to antibiotics.[150]
- There was little change in rate of worsening (progression to septic shock rate) until after 5 hours; by then, approximately 75% of patients had received antibiotics. The remaining 25% of the cohort (who were treated later than the first patient group) differed from those treated earlier. Notably, the later-treated group had more comorbidities.
- As noted in commentary on the study, this is in line with what’s commonly seen in clinical practice: sepsis is harder to recognise in people with comorbidities, and patients with sepsis and comorbid illness have a worse prognosis in general.[58]
Another recent trial randomised 2672 patients with suspected sepsis (>95% without shock) to receive antibiotics in ambulances or ‘quickly’ in an emergency department. The median 96-minute-earlier administration was not linked to improved outcomes, regardless of illness severity.[93]

In intensive care settings only, consider prolonged infusion when giving beta-lactam antibiotics to patients with sepsis (apart from those with kidney-related complications).[151] Note that prolonged infusion times are not licensed, as most manufacturers advise infusion of beta-lactam antibiotics over 15 to 60 minutes.



Evidence: Prolonged antibiotic infusion

Intravenous antibiotics, administered over 3 hours, are linked to lower death rates in sepsis.[151] Prolonged infusion should be easy to apply in the intensive care setting, without the need for additional training or equipment.

A systematic review and meta-analysis pooled the results of 22 randomised controlled trials involving 1876 adults with sepsis. The trials compared prolonged versus short-term administration of any antipseudomonal beta-lactam. Carbapenems were studied in nine trials, penicillins in nine trials, and cephalosporins in eight trials.[151]
- Prolonged infusion was associated with lower all-cause mortality than short-term infusion, with 13.6% deaths compared with 19.8% (risk ratio [RR] 0.70, 95% CI 0.56 to 0.87; 17 studies, 1597 participants).
- There was no significant difference between prolonged and short-term infusion for clinical cure or improvement (RR 1.06, 95% CI 0.96 to 1.17; 11 studies, 1219 participants).
- There was no difference in reported adverse events between the groups (RR 0.88, 95% CI 0.71 to 1.09; 7 studies, 980 participants).
- Two trials had no incidence of antibiotic resistance, and two trials had no difference in resistance between the two methods of antibiotic administration (RR 0.60, 95% CI 0.15 to 2.38).

Narrow choice of antibiotic as soon as a pathogen has been identified and sensitivities are available.[43][137] Assess the need to de-escalate antimicrobial therapy daily.[43]

Studies have shown that daily prompting about antimicrobial de-escalation is effective and may be associated with improved outcomes.[193][194]

Continue broad-spectrum coverage to include all common pathogens if the source is unknown or unclear.[3]

Bear in mind that a definite source of infection cannot be found in 20% to 30% of people with sepsis.[9]

Use the shortest effective course of antibiotics.[195]

Unnecessarily prolonged antibiotic treatment is associated with resistance. See More info: Antimicrobial resistance above.

Consult local microbiology guidance for other specific recommendations on de-escalation.

Most protocols will recommend switching from intravenous to oral antibiotics as soon as possible.

According to the Surviving Sepsis Campaign (SSC), most serious infections associated with sepsis and septic shock will need 7 to 10 days of antibiotic treatment.[43] However, in practice, shorter courses of antibiotics are often appropriate.[196] The optimal duration of antibiotic treatment in patients with sepsis remains contentious, with concerns regarding not only under-treatment but also the potential encouragement of antibiotic resistance. Consider seeking advice from microbiology/infectious disease colleagues.

The SSC guideline suggests considering shorter courses in some patients, particularly those with rapid clinical resolution following effective source control of intra-abdominal or urinary sepsis and those with anatomically uncomplicated pyelonephritis.[43]
Longer courses of treatment may be appropriate in patients who have:[43]
- A slow clinical response
- Undrainable foci of infection
- Bacteraemia with Staphylococcus aureus
- Some fungal and viral infections
- Immunological deficiencies, including neutropenia.

Baseline serum procalcitonin is increasingly being used in critical care settings to guide decisions on how long to continue antibiotic therapy.[43][105][106][107]

Procalcitonin is a peptide precursor of calcitonin, which is responsible for calcium homeostasis.
It is currently excluded from key guidelines, but increasingly used in practice.

Plus – monitor comorbidities during hospital stay

Plus – reassess and monitor

Treatment recommended for ALL patients in selected patient group

Ensure frequent and ongoing monitoring.[3]

Standard monitoring of vital signs, pulse oximetry, level of consciousness, and urinary output is important for any patient with suspected sepsis.
The UK National Institute for Health and Care Excellence (NICE) recommends continuous or half-hourly monitoring (depending on setting) for any patient considered to be at high risk of deterioration (defined in the NICE guideline as meeting one or more of its high-risk criteria for severe illness or death from sepsis).[3]
See the Risk stratification subsection of Diagnosis recommendations for more information.

Use a track-and-trigger scoring system such as the National Early Warning Score 2 (NEWS2) to identify any signs of deterioration.[3] Your monitoring should include:

Vital signs: heart rate, blood pressure, oxygen saturations, respiratory rate, and temperature
- Measure blood pressure via an arterial line if the patient does not respond to initial treatment or needs vasoactive drugs. It provides precise, continuous monitoring, and access for arterial blood sampling
Hourly urine output[3][43][46]
Lactate
- The lactate level should decrease if the patient is clinically improving
- Frequency of repeat lactate measurement depends on the cause of sepsis and treatment given.

Measure serum lactate, on a blood gas, to monitor response to treatment.[3][43][46]

Lactate is a marker of stress and may be a marker of a worse prognosis (as a reflection of the degree of stress). Raised serum lactate highlights the possibility of tissue hypoperfusion and may be present in many conditions.[77][78]
Lactate may normalise quickly after fluid resuscitation. Patients whose lactate levels fail to normalise after adequate fluids are the group that fare worst.
Lactate >4 mmol/L (>36 mg/dL) is associated with worse outcomes.
- One study found in-hospital mortality rates as follows:[79]
  - Lactate <2 mmol/L (<18 mg/dL): 15%
  - Lactate 2.1 to 3.9 mmol/L (19 to 35mg/dL): 25%
  - Lactate >4 mmol/L (>36 mg/dL): 38%.

Sepsis guidelines from NICE and NHS England recommend escalating treatment depending on lactate level.[3][41] Alert critical care immediately if the patient is acutely unwell and has persistent lactate > 4 mmol/L (36 mg/dL)[3] despite fluid resuscitation.

Do not be falsely reassured by a normal lactate (<2 mmol/L [<18 mg/dL]).

This does not rule out the patient being acutely unwell or at risk of deterioration or death due to organ dysfunction. You must take into account the full clinical picture of the individual patient in front of you including their NEWS2 score.

Practical tip

Lactate is typically measured using a blood gas analyser, although laboratory analysis can also be performed.

Traditionally, arterial blood gas has been recommended as the ideal means of measuring lactate accurately. However, in the emergency department setting it is more practical and quicker to use venous blood gas, which is recommended by NICE.[3] Evidence suggests good agreement at lactate levels <2 mmol/L (<18 mg/dL) with small disparities at higher lactate levels.[80][81][82]



Evidence: Lactate clearance

The best available evidence supports lactate clearance (the rate at which lactate is cleared over a period of 6 hours) as being as useful as more invasive tests, such as central venous oxygen saturation (ScvO₂), in determining a patient’s response to treatment.[197][198][199][200][201][202]

In one study that looked at patients with septic shock who were treated to normalise central venous and mean arterial pressure, additional management to normalise lactate clearance compared with additional management to normalise ScvO₂ did not result in significantly different in-hospital mortality.[198]
- Of 300 patients enrolled, 150 were assigned to each group and patients were well matched by demographics, comorbidities, and physiological features. There were no differences in treatments administered during the initial 72 hours of hospitalisation.
- Thirty-four patients (23%) in the ScvO₂ group died while in hospital (95% CI 17% to 30%) compared with 25 (17%, 95% CI 11% to 24%) in the lactate clearance group. This observed difference between mortality rates did not reach the predefined -10% threshold (intent-to-treat analysis: 95% CI for the 6% difference, -3% to 15%). There were no differences in treatment-related adverse events between the groups.
Several trials have assessed the diagnostic accuracy of percentage lactate clearance over 0 to 6 hours. It is worth noting that these studies provide very low-quality evidence, owing mainly to a presumed lack of blinding of treating physicians to the patient’s lactate status.
- The studies’ findings agree that lactate clearance early in the hospital course is associated with decreased mortality rate. Patients with higher lactate clearance after 6 hours of emergency department intervention had improved outcomes compared with those with lower lactate clearance.[197][199][200][201][202]

In the UK, use physiological track-and-trigger systems to monitor all adult patients in acute hospital settings.[3]

Consider using a validated scale such as the Glasgow Coma Scale or AVPU ('Alert, responds to Voice, responds to Pain, Unresponsive') scale to monitor the mental state of a patient with suspected sepsis.[ Glasgow Coma Scale ][3]

Practical tip

AVPU should raise concerns if the assessment shows the patient is anything other than 'alert'.

Any patient with sepsis may be at significant risk of severe illness or death so it is vital to consider escalation of care to senior colleagues and/or healthcare facilities where increased and more advanced monitoring can be given (e.g., high-dependency unit/intensive care unit).[7][170][171]

Bear in mind that some patients (e.g., those who are frail) may not be suitable for management in intensive care settings. Consider the patient’s baseline health including their resuscitation status when determining the limits of treatment. Use this to feed into a personalised care plan appropriate to the individual patient.

Consult local protocols for specific escalation routes but in general:

Ensure immediate review by a senior clinician (CT3/ST3 or higher in the UK) of any patient with a NEWS2 score of 5 or more, or who meets one or more of the NICE sepsis high-risk criteria. Also ensure the patient is discussed with a consultant[3][41]
Discuss with the admitting consultant[3] and consider alerting critical care immediately if the patient is acutely unwell and:
- Has a NEWS2 score of 7 or more, persisting high lactate (more than 4 mmol/L [36 mg/dL]) despite fluid resuscitation, or a systolic blood pressure of less than 90 mmHg[3]
  - Discuss with the admitting consultant[3]
- Has hypotension that doesn’t respond to initial fluid resuscitation
- Is likely to require central venous access and the initiation of inotropes or vasopressors[3]
- Has any feature of septic shock
  - See our topic Shock
- Has neutropenia
  - See our topic Febrile neutropenia
- Is immunodeficient
Urgently discuss with a consultant or call them to attend if the patient:[41]
- Is treated with intravenous antibiotics and/or a fluid bolus for sepsis
- Does not respond to initial therapy (antibiotics/fluid resuscitation/oxygen) within the first hour.[3][41] Failure to respond to treatment is defined as:[3]
  - Systolic blood pressure remains less than 90 mmHg
  - Persistent reduced level of consciousness
  - Respiratory rate more than 25 breaths per minute or the new need for mechanical ventilation
  - Lactate has not reduced by more than 20%
Refer to critical care any patient who is likely to require central venous access and initiation of inotropes or vasopressors[3]
- This includes any patient with evidence of circulatory dysfunction or shock, or those who do not respond to initial therapy (as outlined above)
- ECG can be used to determine which vasoactive drug(s) to proceed with in critical care.[43]

Practical tip

Ensure a clear escalation plan has been discussed and agreed with the clinical team; include specific points of contact for nursing staff if you are leaving a patient for later review.

Involve a senior colleague and/or consider transferring to critical care sooner rather than later if the patient is not improving, or deemed high-risk. Examples include if the patient:

Is not responding to fluids
Needs inotropic support
Has a low Glasgow Coma Scale score
Needs ventilatory support.

Plus – monitor comorbidities during hospital stay

Plus – identify the infection source

Treatment recommended for ALL patients in selected patient group

Make intensive efforts to identify the anatomical site of infection as soon as possible.[3][43] Consider the need for urgent source control as soon as the patient is stable.

Start with a thorough and focused clinical history and examination, as well as initial investigations including imaging.[3]
Consider all lines and catheters as potential sources. Take cultures from the line tip. Remove lines where appropriate.[43]
- Assume that any intravenous route is likely to either be the source of the infection, or will seed infections in the bloodstream, making eradication particularly difficult. Therefore, the priority for source control is often to remove any intravenous devices after vascular access has been obtained.[43]
If you suspect an abdominal or pelvic source, involve the relevant surgical team early, particularly if surgery is likely.[3]
- In practice, this may mean early transfer of the patient to a surgical centre if there are no facilities at your hospital.

Common non-specific signs and symptoms include:[21][43]

Those associated with a specific source of infection. Signs and symptoms of possible infection sources external link opens in a new window The most common sources are:[60]
- Respiratory tract (cough/pleuritic chest pain)
- Urinary tract (flank pain/dysuria)
- Abdominal/upper gastrointestinaI tract (abdominal pain)
- Skin/soft tissue (abscess/wound/catheter site)
- Surgical site or line/drain site
Tachypnoea
High (>38°C [>100.4°F]) or low (<36°C [<96.8°F]) temperature, sometimes with rigors
Tachycardia
Acutely altered mental status
Low oxygen saturation
Hypotension
Decreased urine output
- Ask the patient when they last passed urine
Poor capillary refill, mottling of the skin, or ashen appearance
Cyanosis
Malaise/lethargy
Nausea/vomiting/diarrhoea
Purpura fulminans (a very late sign but may be seen on presentation)
Ileus
Jaundice.

Practical tip

Jaundice is a rare sign of sepsis unless it is associated with a specific source of infection (biliary sepsis).

Capillary refill time. Top image: normal skin tone; middle image: pressure applied for 5 seconds; bottom image: time to hyperaemia measured [Figure caption and citation for the preceding image starts]: Capillary refill time. Top image: normal skin tone; middle image: pressure applied for 5 seconds; bottom image: time to hyperaemia measuredFrom the collection of Ron Daniels, MB, ChB, FRCA; used with permission [Citation ends]. Severe purpura fulminans; classically associated with meningococcal sepsis but can occur with pneumococcal sepsis [Figure caption and citation for the preceding image starts]: Severe purpura fulminans; classically associated with meningococcal sepsis but can occur with pneumococcal sepsisFrom the collection of Ron Daniels, MB, ChB, FRCA; used with permission [Citation ends].

If your examination of the patient identifies a clear source of infection, consider the need for urgent source control, as soon as the patient is stable, particularly for:[43]

Gastrointestinal sources (such as visceral abscesses, cholangitis, or peritonitis secondary to perforation)
Severe skin infections (e.g., necrotising fasciitis)
Infection involving an indwelling device, where a procedure or surgery is likely to be required.

Give immediate, targeted antibiotics in people with sepsis thought to arise from a central nervous system source (e.g., suspected meningitis or meningococcal sepsis).[3]

Immediately give a third-generation cephalosporin such as ceftriaxone or cefotaxime.
In community settings, pre-hospital administration of benzylpenicillin is recommended.
Follow local policy and consider discussing with microbiology/infectious disease colleagues to determine the most appropriate choice; use a 'start smart then focus' approach.[41][136]

Practical tip

If intravenous access is not feasible or is likely to lead to a delay in starting antibiotics and fluids, use intra-osseous access as an interim measure.

Plus – monitor comorbidities during hospital stay

Consider – fluid resuscitation

Treatment recommended for SOME patients in selected patient group

Give 500 mL of crystalloid fluid, with a sodium content between 130 mmol/L and 154 mmol/L (130 to 154 mEq/L) (e.g., 0.9% sodium chloride or Hartmann’s solution), over less than 15 minutes to patients who need fluid resuscitation (if there is any sign of circulatory insufficiency).[3][41][66]

Reassess the patient’s haemodynamic status after the first bolus to consider whether a second is required.[3] If there is no response to either the first or second bolus, seek senior support.[3]

Intravenous fluid resuscitation may be lifesaving in patients with hypotension. This is because in sepsis there is vasodilation and capillary leakage, which means that patients can rapidly become intravascularly deplete.[3]

In patients with sepsis-induced hypoperfusion (as indicated by a systolic blood pressure <90 mmHg, a raised lactate level, or signs of organ dysfunction), the Surviving Sepsis Campaign international guideline recommends a total of at least 30 mL/kg of intravenous crystalloid over the first 3 hours.[43]
- If the patient’s initial lactate level is raised, the guideline recommends serial lactate measurements to guide the need for further intravenous fluids (with the goal of normalising lactate levels).[43]

Practical tip

The delivery of appropriate rapid fluid challenges is intended to restore the imbalance between oxygen supply and demand to the tissues. Patients who do not respond to rapid delivery of adequate volumes of intravenous fluids are in septic shock and need immediate referral to critical care. The immediate priority in this group of patients is to restore the circulation and oxygen delivery.

Practical tip

Monitor patients closely for signs of fluid overload such as pulmonary or systemic oedema before and after each additional fluid bolus, as they may require large volumes of fluid to support their circulating volume.[43][152][153]



Evidence: Choice of fluid

Latest evidence suggests a balanced crystalloid may have marginal benefits over saline, but either option is a reasonable choice.[3][154]

Although early fluid resuscitation is a cornerstone of sepsis treatment that is given high priority by both Sepsis Six and the UK National Institute for Health and Care Excellence, choice of fluid has been the source of much discussion. In particular, there has been extensive debate over the choice between a balanced crystalloid (such as Hartmann’s solution, Ringer’s lactate, or PlasmaLyte) and normal saline (an unbalanced crystalloid). There have been very few high-quality studies, but latest evidence from critically ill patients points to marginal benefits from using a balanced crystalloid in preference to saline.[154]
- A 2018 US multicentre cluster-randomised trial among 15,802 critically ill adults receiving care in the intensive care unit found small benefits from balanced crystalloid compared with saline. The 30-day outcomes showed 10.3% mortality in the balanced crystalloid group compared with 11.1% in the saline group (P = 0.06), and a major adverse kidney event rate of 14.3% compared with 15.4% in the two groups, respectively (marginal odds ratio 0.91, 95% CI 0.84 to 0.99).[154]
Colloids (e.g., starches, dextrans, gelatins, albumin, or fresh frozen plasma) are no longer used in emergency medicine in the UK.



Debate: Volume and rate of fluids

Studies have not shown benefits from early goal-directed therapy and your focus should instead be on adjusting treatment according to i) lactate level and ii) clinical assessment of the patient’s haemodynamic response to initial fluids.[43][155][156][157][158][159][160]

Previous versions of the Surviving Sepsis Campaign (SSC) guidelines recommended a protocoled approach to resuscitation, otherwise known as early goal-directed therapy (EGDT).[161][162][163] EGDT involves the use of a series of 'goals' including central venous pressure and central venous oxygen saturation (ScvO₂). This recommendation was largely based on data from one study that showed a significant survival benefit for patients receiving EGDT.[155]
This approach has since been challenged following the failure to show a mortality reduction in three subsequent large multicentre randomised controlled trials: PROCESS, ARISE, and PROMISE.[156][157][158]
- It is worth bearing in mind that these three trials included patients who were less severely ill (lower baseline lactate levels, ScvO₂ at or above the target value on admission, and lower mortality in the control group) than the patients in the original study that outlined EDGT as a recommended approach.
Based on this latest evidence, the SSC no longer specifically recommends EDGT; it does, however, acknowledge the need for guidance on how to approach this group of patients who have significant mortality and morbidity.[43] The SSC therefore recommends that you:[43]
- View these patients as having a medical emergency that necessitates urgent assessment and treatment
- Begin initial fluid resuscitation with 30 mL/kg of crystalloid within the first 3 hours
  - This fixed volume of fluid enables initiation of resuscitation while giving an opportunity to ascertain more specific information about the patient and while awaiting more precise measurements of haemodynamic status
  - Although scant data are available to support this volume of fluid, interventional studies have described this as usual practice in the early stages of resuscitation, and observational evidence supports the practice[159][160]
  - The average volume of fluid pre-randomisation given was approximately 30 mL/kg in the PROCESS and ARISE trials, and approximately 2 L in the PROMISE trial.[156][157][158] The SSC acknowledges that many patients will need more fluid than this, and advocates giving further fluid to this group in line with functional haemodynamic measurements.[43]
The UK National Institute for Health and Care Excellence guideline on sepsis focuses on initial management and treatment, and therefore makes no recommendations regarding intensive monitoring such as that used in EGDT.[3]

Practical tip

To guide the need for further intravenous fluids, it can sometimes be helpful to use bedside ultrasound to monitor changes in inferior vena cava (IVC) diameter during respiration.[164][165]

In the spontaneously breathing patient: consider additional fluid resuscitation if there is a collapsed (or collapsing) IVC.
In the mechanically ventilated patient: an increase in IVC size >18% (or visible to the naked eye) with positive pressure ventilation suggests fluid-responsiveness.

Practical tip

Use the passive leg-raising test to predict fluid-responsiveness if adequate monitoring is available.[66][166]

This is a useful indicator of fluid-responsiveness, which should be assessed using devices that can continuously monitor cardiac output in real time (e.g., Pulse index Continuous Cardiac Output [PiCCO] monitor or oesophageal Doppler), usually in an intensive care unit rather than a general ward setting.
Sit the patient upright at 45° and tilt the entire bed through 45°.
Patients with a positive test have a >10% increase in cardiac output or stroke volume, indicating more fluids may be required.
The passive leg-raise response may be misleading in conscious patients who are uncomfortable or in pain when lying flat.

Plus – monitor comorbidities during hospital stay

Consider – oxygen

Treatment recommended for SOME patients in selected patient group

If indicated, give oxygen to maintain target oxygen saturations >94%.[3][43][46][47] Latest evidence suggests that liberal use of supplemental oxygen (target SpO₂ >96%) in acutely ill adults is associated with higher mortality than more conservative oxygen therapy.[138] Therefore, a reasonable approach in practice is to maintain a target oxygen saturation of 94% to 96% in acutely ill patients who are not at risk of hypercapnia.

Target saturation of 88% to 92% if the patient is at risk of hypercapnic respiratory failure (e.g., patients with COPD).[3][43][46][47]



Evidence: Target oxygen saturation in acutely ill adults

Too much supplemental oxygen increases mortality.

Latest evidence supports a 96% upper limit for target oxygen saturation in non-hypercapnic acutely ill adults.

Guidelines differ in their recommendations on target oxygen saturation in acutely unwell adults who are receiving supplemental oxygen. The 2017 British Thoracic Society guideline recommends a target SpO₂ range of 94% to 98% for patients not at risk of hypercapnia,[47] whereas the 2015 Thoracic Society of Australia and New Zealand guideline recommends 92% to 96%.[167]
A 2018 systematic review including a meta-analysis of data from 25 randomised controlled trials found that in adults with acute illness, liberal oxygen therapy (broadly equivalent to a target saturation >96%) is associated with higher mortality than conservative oxygen therapy (broadly equivalent to a target saturation ≤96%).[138] In-hospital mortality was 11 per 1000 higher for the liberal oxygen therapy group versus the conservative therapy group (95% CI 2-22 per 1000 more). Mortality at 30 days was also higher in the group who had received liberal oxygen (risk ratio 1.14, 95% CI 1.01 to 1.29). The trials included adults with sepsis, critical illness, stroke, trauma, myocardial infarction, and cardiac arrest, and patients who had emergency surgery. Studies that were limited to people with chronic respiratory illness or psychiatric illness, patients on extracorporeal life support, patients receiving hyperbaric oxygen therapy, or those having elective surgery were all excluded from the review.
An upper SpO₂ limit of 96% is therefore reasonable when administering supplemental oxygen to medical patients with acute illness who are not at risk of hypercapnia. However, a higher target may be appropriate for some specific conditions (e.g., pneumothorax, carbon monoxide poisoning, cluster headache, and sickle cell crisis).[168]

There is no specific evidence to show that giving oxygen improves clinical outcomes in sepsis. However, respiratory failure will lead to tissue hypoxia and anaerobic respiration. This is likely to lead to acidosis and consequently a poorer outcome.[169]

Plus – monitor comorbidities during hospital stay

Consider – standard intensive care unit supportive care

Treatment recommended for SOME patients in selected patient group

Bear in mind that some patients (e.g., those who are frail) may not be suitable for management in intensive care settings. Consider the patient’s baseline health including their resuscitation status when determining the limits of treatment. Use this to feed into a personalised care plan appropriate to the individual patient.

Consult local protocols for specific escalation routes but in general:

Ensure immediate review by a senior clinician (CT3/ST3 or higher in the UK) of any patient with a National Early Warning Score 2 (NEWS2) score of 5 or more, or who meets one or more of the UK National Institute for Health and Care Excellence sepsis high-risk criteria. Also ensure the patient is discussed with a consultant[3][41]
Discuss with the admitting consultant[3] and consider alerting critical care immediately if the patient is acutely unwell and:
- Has a NEWS2 score of 7 or more, persisting high lactate (more than 4 mmol/L [36 mg/dL]) despite fluid resuscitation, or a systolic blood pressure of less than 90 mmHg[3]
  - Discuss with the admitting consultant[3]
- Has hypotension that doesn’t respond to initial fluid resuscitation
- Is likely to require central venous access and the initiation of inotropes or vasopressors
- Has any feature of septic shock
  - See our topic Shock
- Has neutropenia
  - See our topic Febrile neutropenia
- Is immunodeficient
Urgently discuss with a consultant or call them to attend if the patient:[41]
- Is treated with intravenous antibiotics and/or a fluid bolus for sepsis
- Does not respond to initial therapy (antibiotics/fluid resuscitation/oxygen) within the first hour.[3][41] Failure to respond to treatment is defined as:[3]
  - Systolic blood pressure remains less than 90 mmHg
  - Persistent reduced level of consciousness
  - Respiratory rate more than 25 breaths per minute or the new need for mechanical ventilation
  - Lactate has not reduced by more than 20%
Refer to critical care any patient who is likely to require central venous access and initiation of inotropes or vasopressors[3]
- This includes any patient with evidence of circulatory dysfunction or shock, or those who do not respond to initial therapy (as outlined above)
- ECG can be used to determine which vasoactive drugs(s) to proceed with in critical care.[43]

Practical tip

Ensure a clear escalation plan has been discussed and agreed with the clinical team; include specific points of contact for nursing staff if you are leaving a patient for later review.

Involve a senior colleague and/or consider transferring to critical care sooner rather than later if the patient is not improving, or deemed high-risk. Examples include if the patient:

Is not responding to fluids
Needs inotropic support
Has a low Glasgow Coma Scale score
Needs ventilatory support.

For any patient with suspected sepsis, consider the need for referral to a high-dependency unit for management by the critical care team.[155][174]

The following interventions should only be initiated by experienced members of the critical care team:[3][43][175]

Glycaemic control
Vasoactive drugs (vasopressors/inotropes)
Corticosteroids.

Additional intensive care measures that will be considered include:[43][176][177]

Stress ulcer prophylaxis (in people at risk of gastrointestinal bleeding)
- With an H2 antagonist or proton-pump inhibitor
Deep venous thrombosis prophylaxis
- With heparin and compression stockings
Enteral or parenteral nutrition
Administration of human albumin solution 4% to 5% in patients with sepsis and shock[3][43] who have not responded to substantial volumes of crystalloids
Transfusion of packed cells
- Consult local protocols for recommended threshold
- The Surviving Sepsis Campaign recommends using a threshold of 70 g/L (7 g/dL).[43]



Evidence: Threshold for transfusion of packed cells

Studies in the general critical care population have shown no improvement with blood transfusions given at a higher haemoglobin threshold compared with a lower haemoglobin threshold,[178] and have shown potential harm associated with liberal transfusion.[179]

One multicentre parallel group randomised trial analysed data from 998 patients with septic shock, split into two intervention groups with similar baseline characteristics. In the intensive care unit, the group assigned to blood transfusion at a lower haemoglobin threshold received a median of 1 unit of blood (interquartile range, 0 to 3) and the group assigned to blood transfusion at a higher haemoglobin threshold received a median of 4 units (interquartile range, 2 to 7).[178]
- At 90 days after randomisation, 216 of 502 patients (43.0%) assigned to the lower-threshold group, compared with 223 of 496 (45.0%) assigned to the higher-threshold group, had died (relative risk 0.94, 95% CI 0.78 to 1.09; P = 0.44).
- The results were similar in analyses adjusted for risk factors at baseline and in analyses of the per-protocol populations.
- The numbers of patients who had ischaemic events, who had severe adverse reactions, and who required life support were similar in the two intervention groups.
A second multicentre randomised controlled trial compared the rates of death from all causes at 30 days and the severity of organ dysfunction in 838 critically ill patients receiving a restrictive strategy of red-cell transfusion compared with those receiving a liberal strategy.[179]
- Overall, 30-day mortality was similar in the two groups (18.7% vs. 23.3%, P = 0.11).
- However, the rates were significantly lower with the restrictive transfusion strategy among patients who were less acutely ill and in patients younger than 55 years of age (5.7% and 13.0%, respectively; P = 0.02), but not in those with clinically significant cardiac disease (20.5% and 22.9%, respectively; P = 0.69).
- The mortality rate during hospitalisation was significantly lower in the restrictive-strategy group (22.3% vs. 28.1%, P = 0.05).

There may be a case to consider giving transfusions at a higher haemoglobin level in people with myocardial ischaemia, severe hypoxaemia, acute haemorrhage, cyanotic heart disease, or lactic acidosis.[43]

In the initial resuscitative phase, transfusion to achieve a higher haematocrit of ≥30% may be appropriate.[155]

In patients requiring prolonged ventilatory support, give lung-protective ventilation using minimal peak inspiratory pressures (<30 cm H₂O) and permissive hypercapnia to specifically limit pulmonary compromise.[180]

Titrate fraction of inspired oxygen (FiO₂) to lowest effective levels to prevent oxygen toxicity and maintain central venous oxygen tension.
Place patients in a semi-recumbent position with the head elevated to 30° to 45°.[43]

Glycaemic control

Although patients with sepsis are often hyperglycaemic, the optimal glucose target is unknown.

The Surviving Sepsis Campaign guideline recommends targeting a blood glucose level <10.0 mmol/L (<180 mg/dL).[43] The guideline also recommends a ‘sliding scale’ variable-rate intravenous insulin infusion.[43]

The UK National Institute for Health and Care Excellence makes no recommendations on glycaemic control in sepsis.[3]



Evidence: Glycaemic control

Recent years have seen a shift in opinion and practice regarding glycaemic control in critically ill people. Since 2001, the use of tight glycaemic control has been advocated in people with sepsis. More recent evidence, however, suggests an increase in adverse events (e.g., severe hypoglycaemia) in patients managed with very tight glycaemic control (targeting a blood glucose below 6.1 mmol/L [110 mg/dL]).[181][182]The conflicting evidence has led to variations in recommendations in different countries and settings. Follow your local protocol.

An international randomised controlled trial (RCT) of 6104 critically ill medical and surgical patients found increased 90-day mortality (odds ratio 1.14, 95% CI 1.02 to 1.28) with tighter glucose control, possibly due to more frequent episodes of hypoglycaemia.[183]
A 2010 systematic review of 6 RCTs and a meta-analysis investigating tight glucose control (4.4 to 6.1 mmol/L [80-110 mg/dL]) versus less strict glucose control in critically ill patients in the intensive care unit setting found no significant improvement in mortality with tight glucose control, but it was associated with significantly more hypoglycaemic episodes compared with less strict glucose control.[184]
An RCT of critically ill patients in a primarily surgical intensive care setting found lower patient mortality with tight glucose control, 4.4 to 6.1 mmol/L (80-110 mg/dL), compared with ‘conventional’ more liberal glucose control.[185]

Plus – monitor comorbidities during hospital stay

Consider – vasopressor (should only be initiated by experienced members of the critical care team)

Treatment recommended for SOME patients in selected patient group

Vasopressors are used in a critical care setting to maintain a mean arterial pressure (MAP) ≥65 mmHg if the patient is unresponsive to fluid resuscitation.[3][43][175]

Failure to respond to initial fluid resuscitation is a sign of septic shock.[1]
Noradrenaline (norepinephrine) is the vasopressor of choice, mainly because it increases MAP.[43]
- Noradrenaline is the vasopressor recommended by the Surviving Sepsis Campaign guideline.[43] The UK National Institute for Health and Care Excellence makes no recommendation on the choice of vasopressor.[3]
- If further vasopressor therapy is required to maintain adequate blood pressure or the noradrenaline dose needs to be reduced, add vasopressin or adrenaline (epinephrine) to noradrenaline.[43]
- Dopamine is an option, but has been associated with higher mortality than noradrenaline.[186][187] [ ] Therefore, it is only recommended in patients with a low risk of tachyarrhythmias and bradycardia.[43] It is rarely used in the UK. Do not use low-dose dopamine for renal protection.[43]

Practical tip

All infusions of vasoactive drugs to correct shock should be given via a secure catheter in a central vein with high flow, such as a central venous catheter. These patients should also have an arterial catheter inserted as soon as possible to ensure more accurate monitoring of arterial blood pressure.[43]



Evidence: Choice of vasopressor

Although a systematic review of 23 randomised trials of patients with shock found no convincing evidence for the superiority of one vasopressor over another,[188] more recent meta-analyses reported a higher mortality associated with dopamine than with noradrenaline.[186]

Primary options

noradrenaline (norepinephrine): 0.4 to 0.8 mg/hour intravenous infusion initially, adjust dose according to response

Secondary options

noradrenaline (norepinephrine): 0.4 to 0.8 mg/hour intravenous infusion initially, adjust dose according to response

-- AND --

vasopressin: 0.01 units/minute intravenous infusion initially, adjust dose according to response, maximum 0.03 units/minute

adrenaline (epinephrine): 2-10 micrograms/minute intravenous infusion initially, adjust dose according to response

dopamine: 2-5 micrograms/kg/minute intravenous infusion initially, adjust dose according to response

Plus – monitor comorbidities during hospital stay

Consider – inotrope (should only be initiated by experienced members of the critical care team)

Treatment recommended for SOME patients in selected patient group

Inotropes can be considered for patients with low cardiac output despite adequate fluid resuscitation and vasopressor therapy.[3][43]

Dobutamine is recommended first line by the Surviving Sepsis Campaign guideline for people with measured or suspected low cardiac output in the presence of adequate left ventricular filling pressure (or clinical assessment of adequate fluid resuscitation) and adequate mean arterial pressure.[43]
The UK National Institute for Health and Care Excellence makes no specific recommendations on inotrope selection in patients with sepsis.[3]

Practical tip

Suspect low cardiac output if the clinical examination reveals prolonged capillary refill times, low urine output, or poor peripheral perfusion. Confirm with cardiac output monitoring or by sampling central venous or pulmonary arterial blood to measure oxygen saturations.

When using inotropes, keep the patient’s heart rate at less than 100 beats per minute to minimise myocardial ischaemia.[175]

Primary options

dobutamine: 2.5 to 10 micrograms/kg/minute intravenous infusion initially, adjust dose according to response, maximum 40 micrograms/kg/minute

Plus – monitor comorbidities during hospital stay

Consider – corticosteroid (should only be initiated by experienced members of the critical care team)

Treatment recommended for SOME patients in selected patient group

The Surviving Sepsis Campaign guideline recommends intravenous hydrocortisone as an option to consider for patients who are unresponsive to both fluid resuscitation and vasopressor therapy.[43]

The UK National Institute for Health and Care Excellence does not give any recommendations on the use of corticosteroids for managing sepsis in adults.[3]



Evidence: Benefits and harms of corticosteroids

Corticosteroids are a late critical-care intervention for patients in whom all other attempts to raise their blood pressure have failed.[3] In this critically ill group, corticosteroids may result in a small reduction in mortality.[189] Possible harms include an increased risk of neuromuscular weakness, hyperglycaemia, and hypernatraemia with corticosteroids compared with no corticosteroids.[189]

An international panel reviewing the inconsistent conclusions of large randomised controlled trials (RCTs) on the topic suggested in 2018 that the evidence for the use of corticosteroids was weak, but that “fully informed patients who value avoiding death over quality of life and function would likely choose corticosteroids”, although a no-corticosteroid approach remained reasonable.[189] This judgement was based on an assessment that corticosteroids may reduce mortality by around 2% (this effect was seen in sepsis with and without shock although the greatest benefit was among patients with septic shock), but can increase the risk of neuromuscular weakness and resulting functional deterioration. Most notably, the panel reviewed the following trials:
- ADRENAL: 3658 patients who had septic shock[191]
  - No statistically significant difference in 90-day mortality between the hydrocortisone and placebo groups
- APROCCHSS: 1241 patients who had septic shock[192]
  - Hydrocortisone plus fludrocortisone reduced 90-day mortality.
  [Figure caption and citation for the preceding image starts]: BMJ Rapid Recommendations: intravenous corticosteroids plus usual care versus usual care onlyLamontagne F, et al. BMJ 2018;362:k3284 [Citation ends].
A more recent systematic review and meta-analysis (comprising 37 RCTs, incorporating both ADRENAL and APROCHSS) included 9564 people with sepsis.[190] The review found corticosteroid use to be associated with significant improvement in healthcare outcomes as shown by:[190]
- Reduced 28-day mortality (risk ratio [RR] 0.90, 95% CI 0.82 to 0.88) compared with placebo or standard supportive care
- Reduced intensive care unit mortality (RR 0.85, 95% CI 0.77 to 0.94; I² = 0%) compared with placebo or standard supportive care
- Reduced in-hospital mortality (RR 0.88, 95% CI 0.79 to 0.99; I² = 38%) compared with placebo or standard supportive care.
However, corticosteroid use was also associated with increased risk of hyperglycaemia (RR 1.19, 95% CI 1.08 to 1.30) and hypernatraemia (RR 1.57, 95% CI 1.24 to 1.99) compared with placebo or standard supportive care.[190]

Primary options

hydrocortisone sodium succinate: 200 mg/day intravenously

Plus – monitor comorbidities during hospital stay

in the community: sepsis highly suspected and bacterial infection confirmed or highly suspected

1st line – refer for emergency medical care in hospital

Use your clinical judgement. Use National Early Warning Score 2 (NEWS2) scoring (encouraged by NHS England) to refer urgently to hospital any acutely unwell patient with suspected or confirmed infection according to the following triggers:[41] NHS England: Sepsis external link opens in a new window

Score 7 or more
- Make an emergency referral to hospital (via blue-light ambulance) for immediate critical care input.
Score 5-6 total, or 3 or more on any single parameter
- Make an immediate referral to an acute care setting and ensure the patient is reviewed by an acute clinician within an hour.

Track and trigger map developed by the West of England Academic Health Science Network National Early Warning Score project team external link opens in a new window

Alternatively, refer for emergency medical care in hospital (usually by blue-light ambulance in the UK) any acutely unwell patient with suspected or confirmed infection who:[3]

Meets one or more of the UK National Institute for Health and Care Excellence (NICE) high-risk criteria (red flags)
- Objective evidence of new altered mental state (e.g., new deterioration in Glasgow Coma Scale score/AVPU ['Alert, responds to Voice, responds to Pain, Unresponsive'] scale)
- Respiratory rate: ≥25 breaths per minute OR new need for oxygen (40% or more fraction of inspired oxygen [FiO₂]) to maintain saturation >92% (or >88% in known chronic obstructive pulmonary disease)
- Heart rate: >130 beats per minute
- Systolic blood pressure ≤90 mmHg or more than 40 mmHg below normal
- Not passed urine in previous 18 hours, or for catheterised patients passed <0.5 mL/kg of urine per hour
- Mottled or ashen appearance
- Cyanosis of skin, lips, or tongue
- Non-blanching rash of skin
Is at risk of neutropenic sepsis and presents with symptoms and signs of infection
- See our topic Febrile neutropenia.

Carefully consider whether emergency medical care is required or whether the patient can be safely managed in the community with safety netting advice.[3] See box below on safety netting advice.

Practical tip

If you need to refer a patient for emergency medical care in hospital, it is important to inform the hospital clinical team that the patient is on the way. This will enable the hospital to initiate appropriate treatment as soon as the patient arrives.

In a patient with signs and symptoms of an infection and evidence of physiological deterioration, presume sepsis until it can safely be excluded. Take a cautious approach when deciding whether it is safe to treat the patient’s infection in the community. Using your clinical judgement in making a decision is paramount. In particular, carefully consider the need for hospital admission if:[50][74][140]

The patient has one or more risk factors for sepsis (as listed above)
The patient appears seriously unwell to you, based on experience and clinical judgement
The patient lives alone with poor access to communication and/or transport
A carer or parent expresses serious concern about the patient (e.g., “they’re just not right”).

See our Diagnosis recommendations section for details of the NICE risk criteria.

Treat the patient’s infection in line with local protocols and accepted practice. Antimicrobial prescribing guidelines from Public Health England and NICE are available for general practitioners in the UK.[203][204]

Practical tip

If you decide that the patient is safe to treat in the community, written and verbal safety netting is vital.[50] Ensure the information is clear and specific rather than generalised advice; for example, do not say “come back if you get worse” – instead, specify key symptoms to watch out for (such as a non-blanching rash, change in behaviour or mental state, mottled skin, or ashen appearance) and explain where and how to access immediate medical care both in and out of hours.[50]

If you give the patient any safety netting advice, ensure you document this clearly in their medical notes, along with the patient’s observations and whether you have offered them any antibiotics. The 2015 national confidential enquiry into sepsis deaths found recorded evidence that safety netting advice had been provided in fewer than one quarter of cases.[60]

The UK Sepsis Trust advises the following acronym:[140]

Slurred speech or confusion
Extreme shivering or muscle pain
Passing no urine (in a day)
Severe breathlessness
‘I feel I might die’
Skin mottled, ashen, blue, or very pale.

Advise the patient to call the emergency services if any of these symptoms develop. If the patient has a change in condition or deterioration that is not covered by the acronym above, advise them to arrange another appointment to see their general practitioner or to call their out of hours service provider.

It is also good practice to consider arranging a next-day review appointment or telephone call; if you will be unable to review the patient yourself, provide a written handover for your colleagues.[139]

Plus – monitor comorbidities during hospital stay

Consider – oxygen

Treatment recommended for SOME patients in selected patient group

If you have decided to refer the patient for emergency medical care and have called for an emergency ambulance, the UK Sepsis Trust/UK National Institute for Health and Care Excellence general practitioner toolkit recommends that, if indicated, you should start oxygen therapy while awaiting the ambulance if resources are available to do so.[139][140]

Give oxygen immediately to maintain target oxygen saturations >94%.[3][43][46][47] Latest evidence suggests that liberal use of supplemental oxygen (target SpO₂ >96%) in acutely ill adults is associated with higher mortality than more conservative oxygen therapy.[138] Therefore, a reasonable approach in practice is to maintain a target oxygen saturation of 94% to 96% in acutely ill patients who are not at risk of hypercapnia.
Target saturation of 88% to 92% in people at risk of hypercapnic respiratory failure (e.g., those with COPD.[3][43][46][47]



Evidence: Target oxygen saturation in acutely ill adults

Too much supplemental oxygen increases mortality.

Latest evidence supports a 96% upper limit for target oxygen saturation in non-hypercapnic acutely ill adults.

Guidelines differ in their recommendations on target oxygen saturation in acutely unwell adults who are receiving supplemental oxygen. The 2017 British Thoracic Society guideline recommends a target SpO₂ range of 94% to 98% for patients not at risk of hypercapnia,[47] whereas the 2015 Thoracic Society of Australia and New Zealand guideline recommends 92% to 96%.[167]
A 2018 systematic review including a meta-analysis of data from 25 randomised controlled trials found that in adults with acute illness, liberal oxygen therapy (broadly equivalent to a target saturation >96%) is associated with higher mortality than conservative oxygen therapy (broadly equivalent to a target saturation ≤96%).[138] In-hospital mortality was 11 per 1000 higher for the liberal oxygen therapy group versus the conservative therapy group (95% CI 2-22 per 1000 more). Mortality at 30 days was also higher in the group who had received liberal oxygen (risk ratio 1.14, 95% CI 1.01 to 1.29). The trials included adults with sepsis, critical illness, stroke, trauma, myocardial infarction, and cardiac arrest, and patients who had emergency surgery. Studies that were limited to people with chronic respiratory illness or psychiatric illness, patients on extracorporeal life support, patients receiving hyperbaric oxygen therapy, or those having elective surgery were all excluded from the review.
An upper SpO₂ limit of 96% is therefore reasonable when administering supplemental oxygen to medical patients with acute illness who are not at risk of hypercapnia. However, a higher target may be appropriate for some specific conditions (e.g., pneumothorax, carbon monoxide poisoning, cluster headache, and sickle cell crisis).[168]

Plus – monitor comorbidities during hospital stay

Consider – broad-spectrum antibiotics

Plus – monitor comorbidities during hospital stay

ACUTE

in hospital: sepsis highly suspected and clear source of bacterial infection identified

1st line – targeted antibiotics according to local protocols

Once a definitive source has been identified, if appropriate to continue treating the patient with antibiotics, choose a treatment regimen in line with local or national policy (which will take into account specialist knowledge of resistance patterns).[43][137] Also consider discussing with microbiology/infectious disease colleagues to determine the most appropriate choice.

Assess the need to de-escalate antimicrobial therapy daily.[43]

Studies have shown that daily prompting about antimicrobial de-escalation is effective and may be associated with improved outcomes.[193][194]

Use the shortest effective course of antibiotics.[195]

Unnecessarily prolonged antibiotic treatment is associated with resistance.



More info: Antimicrobial resistance

Start smart – in the context of sepsis:[136]

Do not start antimicrobial therapy unless there is clear evidence of infection
Take a thorough drug allergy history
Initiate prompt effective antibiotic treatment within 1 hour of diagnosis (or as soon as possible) in patients with sepsis or life-threatening infections. Avoid inappropriate use of broad-spectrum antibiotics
Comply with local antimicrobial prescribing guidance
Document clinical indication (and disease severity if appropriate), drug name, dose, and route on drug chart and in clinical notes
Include review/stop date or duration
Obtain cultures prior to starting therapy where possible (but do not delay therapy).

Then focus – in the context of sepsis:[136]

Review the clinical diagnosis and the continuing need for antibiotics at 48 to 72 hours* and document in a clear plan of action – the ‘antimicrobial prescribing decision’
The ‘antimicrobial prescribing decision’ options are:
1. Stop antibiotics if there is no evidence of infection
2. Switch antibiotics from intravenous to oral
3. Change antibiotics – ideally to a narrower spectrum, or broader if required
4. Continue and document next review date or stop date
It is essential that the review and subsequent decision is clearly documented in the clinical notes and on the drug chart where possible (e.g., ‘stop antibiotic’).

*In clinical practice, daily prompting about de-escalation is encouraged.

Consult local microbiology guidance for other specific recommendations on de-escalation.

Most protocols will recommend switching from intravenous to oral antibiotics as soon as possible.

The SSC guideline suggests considering shorter courses in some patients, particularly those with rapid clinical resolution following effective source control of intra-abdominal or urinary sepsis and those with anatomically uncomplicated pyelonephritis.[43]
Longer courses of treatment may be appropriate in patients who have:[43]
- A slow clinical response
- Undrainable foci of infection
- Bacteraemia with Staphylococcus aureus
- Some fungal and viral infections
- Immunological deficiencies, including neutropenia.

Baseline serum procalcitonin is increasingly being used in critical care settings to guide decisions on how long to continue antibiotic therapy.[43][105][106][107]

Procalcitonin is a peptide precursor of calcitonin, which is responsible for calcium homeostasis.
It is currently excluded from key guidelines, but increasingly used in practice.

Respiratory

Ensure treatment regimens cover common respiratory pathogens and atypical organisms such as Legionella pneumophila.

The respiratory tract is the most common site of infection in people with sepsis.[20][60]
See our topic Overview of pneumonia.

Abdominal

Ensure gram-positive and gram-negative organisms including anaerobes are covered.[172]

Arrange urgent surgical drainage or percutaneous drainage (where appropriate) for peritonitis or intra-peritoneal abscesses.[173]

Urinary tract

Ensure gram-negative coliforms and Pseudomonas are covered. Ensuring patency of the urinary tract is vital.

In people older than 65 years of age, genitourinary tract infections are the most common cause of sepsis.[21][22]

Soft tissue and joint

Includes septic arthritis, wound infections, cellulitis, and acute super-infections arising from chronic ulceration. Most infections are polymicrobial. Ensure gram-positive and gram-negative organisms including anaerobes are covered.

Beware necrotising fasciitis, which requires immediate surgical intervention (as does septic arthritis).

Practical tip

Necrotising fasciitis is notoriously difficult to diagnose. The initial symptoms are non-specific and the clinical course is often slower than might be expected. Typically, the first sign is pain disproportionate to the clinical findings, followed or accompanied by fever.[76]

See our topic Necrotising fasciitis.

Central nervous system

Relatively uncommon but potentially devastating source of sepsis. Beware meningococcal sepsis, which can be extremely rapidly fatal; if survived, can lead to greater morbidity than other forms of sepsis.

Give immediate, targeted antibiotics in people with sepsis thought to arise from a central nervous system source.[3]

Immediately give a third-generation cephalosporin, such as ceftriaxone or cefotaxime, for suspected meningitis or meningococcal sepsis.[3]

Plus – monitor comorbidities during hospital stay

Plus – reassess and monitor

Treatment recommended for ALL patients in selected patient group

Ensure frequent and ongoing monitoring.[3]

Standard monitoring of vital signs, pulse oximetry, level of consciousness, and urinary output is important for any patient with suspected sepsis.
The UK National Institute for Health and Care Excellence (NICE) recommends continuous or half-hourly monitoring (depending on setting) for any patient considered to be at high risk of deterioration (defined in the NICE guideline as meeting one or more of its high-risk criteria for severe illness or death from sepsis).[3]
See the Risk stratification subsection of Diagnosis recommendations for more information.

Use a track-and-trigger scoring system such as National Early Warning Score 2 (NEWS2) to identify any signs of deterioration.[3] Your monitoring should include:

Vital signs: heart rate, blood pressure, oxygen saturations, respiratory rate, and temperature
- Measure blood pressure via an arterial line if the patient does not respond to initial treatment or needs vasoactive drugs. It provides precise, continuous monitoring, and access for arterial blood sampling
Hourly urine output[3][43][46]
Lactate
- The lactate level should decrease if the patient is clinically improving
- Frequency of repeat lactate measurement depends on the cause of sepsis and treatment given.

Measure serum lactate, on a blood gas, to monitor response to treatment.[3][43][46]

Lactate is a marker of stress and may be a marker of a worse prognosis (as a reflection of the degree of stress). Raised serum lactate highlights the possibility of tissue hypoperfusion and may be present in many conditions.[77][78]
Lactate may normalise quickly after fluid resuscitation. Patients whose lactate levels fail to normalise after adequate fluids are the group that fare worst.
Lactate >4 mmol/L (>36 mg/dL) is associated with worse outcomes.
- One study found in-hospital mortality rates as follows:[79]
  - Lactate <2 mmol/L (<18 mg/dL): 15%
  - Lactate 2.1 to 3.9 mmol/L (19 to 35mg/dL): 25%
  - Lactate >4 mmol/L (>36 mg/dL): 38%.

Sepsis guidelines from NICE and NHS England recommend escalating treatment depending on lactate level.[3][41] Alert critical care immediately if the patient is acutely unwell and has persistent lactate >4 mmol/L (>36 mg/dL)[3] despite fluid resuscitation.

Do not be falsely reassured by a normal lactate (<2 mmol/L [<18 mg/dL]).

This does not rule out the patient being acutely unwell or at risk of deterioration or death due to organ dysfunction. You must take into account the full clinical picture of the individual patient in front of you including their NEWS2 score.

Practical tip

Lactate is typically measured using a blood gas analyser, although laboratory analysis can also be performed.



Evidence: Lactate clearance

In one study that looked at patients with septic shock who were treated to normalise central venous and mean arterial pressure, additional management to normalise lactate clearance compared with additional management to normalise ScvO₂did not result in significantly different in-hospital mortality.[198]
- Of 300 patients enrolled, 150 were assigned to each group and patients were well matched by demographics, comorbidities, and physiological features. There were no differences in treatments administered during the initial 72 hours of hospitalisation.
- Thirty-four patients (23%) in the ScvO₂ group died while in the hospital (95% CI 17% to 30%) compared with 25 (17%, 95% CI 11% to 24%) in the lactate clearance group. This observed difference between mortality rates did not reach the predefined -10% threshold (intent-to-treat analysis: 95% CI for the 6% difference, -3% to 15%). There were no differences in treatment-related adverse events between the groups.
Several trials have assessed the diagnostic accuracy of percentage lactate clearance over 0 to 6 hours. It is worth noting that these studies provide very low-quality evidence, owing mainly to a presumed lack of blinding of treating physicians to the patient’s lactate status.
- The studies’ findings agree that lactate clearance early in the hospital course is associated with decreased mortality rate. Patients with higher lactate clearance after 6 hours of emergency department intervention had improved outcomes compared with those with lower lactate clearance.[197][199][200][201][202]

In the UK, use physiological track-and-trigger systems to monitor all adult patients in acute hospital settings.[3]

Practical tip

AVPU should raise concerns if the assessment shows the patient is anything other than 'alert'.

Bear in mind that some patients (e.g., those who are frail) may not be suitable for management in intensive care settings. Consider the patient’s baseline health including their resuscitation status when determining the limits of treatment. Use this to feed into a personalised care plan appropriate to the individual patient.

Consult local protocols for specific escalation routes but in general:

Ensure immediate review by a senior clinician (CT3/ST3 or higher in the UK) of any patient with a NEWS2 score of 5 or more, or who meets one or more of the NICE sepsis high-risk criteria. Also ensure the patient is discussed with a consultant[3][41]
Discuss with the admitting consultant[3] and consider alerting critical care immediately if the patient is acutely unwell and:
- Has a NEWS2 score of 7 or more, persisting high lactate (more than 4 mmol/L [36 mg/dL]) despite fluid resuscitation, or a systolic blood pressure of less than 90 mmHg[3]
  - Discuss with the admitting consultant[3]
- Has hypotension that doesn’t respond to initial fluid resuscitation
- Is likely to require central venous access and the initiation of inotropes or vasopressors[3]
- Has any feature of septic shock
  - See our topic Shock
- Has neutropenia
  - See our topic Febrile neutropenia
- Is immunodeficient
Urgently discuss with a consultant or call them to attend if the patient:[41]
- Is treated with intravenous antibiotics and/or a fluid bolus for sepsis
- Does not respond to initial therapy (antibiotics/fluid resuscitation/oxygen) within the first hour.[3][41] Failure to respond to treatment is defined as:[3]
  - Systolic blood pressure remains less than 90 mmHg
  - Persistent reduced level of consciousness
  - Respiratory rate more than 25 breaths per minute or the new need for mechanical ventilation
  - Lactate has not reduced by more than 20%
Refer to critical care any patient who is likely to require central venous access and initiation of inotropes or vasopressors[3]
- This includes any patient with evidence of circulatory dysfunction or shock, or those who do not respond to initial therapy (as outlined above)
- ECG can be used to determine which vasoactive drug(s) to proceed with in critical care.[43]

Practical tip

Ensure a clear escalation plan has been discussed and agreed with the clinical team; include specific points of contact for nursing staff if you are leaving a patient for later review.

Involve a senior colleague and/or consider transferring to critical care sooner rather than later if the patient is not improving, or deemed high-risk. Examples include if the patient:

Is not responding to fluids
Needs inotropic support
Has a low Glasgow Coma Scale score
Needs ventilatory support.

Plus – monitor comorbidities during hospital stay

Plus – urgent source control

Treatment recommended for ALL patients in selected patient group

Once a site of infection has been identified, early and adequate source control is critical. Consider the need for urgent source control, as soon as the patient is stable, particularly for:[43]

Gastrointestinal sources (such as visceral abscesses, cholangitis, or peritonitis secondary to perforation)
Severe skin infections (e.g., necrotising fasciitis)
Infection involving an indwelling device, where a procedure or surgery is likely to be required.

Give immediate, targeted antibiotics in people with sepsis thought to arise from a central nervous system source (e.g., suspected meningitis or meningococcal sepsis).[3]

Immediately give a third-generation cephalosporin such as ceftriaxone or cefotaxime.
In community settings, pre-hospital administration of benzylpenicillin is recommended.
Follow local policy and consider discussing with microbiology/infectious disease colleagues to determine the most appropriate choice; use a ‘start smart then focus’ approach.[41][136]

Practical tip

If intravenous access is not feasible or is likely to lead to a delay in starting antibiotics and fluids, use intra-osseous access as an interim measure.

Plus – monitor comorbidities during hospital stay

Consider – fluid resuscitation

Treatment recommended for SOME patients in selected patient group

Reassess the patient’s haemodynamic status after the first bolus to consider whether a second is required.[3] If there is no response to either the first or second bolus, seek senior support.[3]

In patients with sepsis-induced hypoperfusion (as indicated by a systolic blood pressure <90 mmHg, a raised lactate level, or signs of organ dysfunction), the Surviving Sepsis Campaign international guideline recommends a total of at least 30 mL/kg of intravenous crystalloid over the first 3 hours.[43]
- If the patient’s initial lactate level is raised, the guideline recommends serial lactate measurements to guide the need for further intravenous fluids (with the goal of normalising lactate levels).[43]

Practical tip



Evidence: Choice of fluid

Latest evidence suggests a balanced crystalloid may have marginal benefits over saline, but either option is a reasonable choice.[3][154]

Although early fluid resuscitation is a cornerstone of sepsis treatment that is given high priority by both Sepsis Six and the UK National Institute for Health and Care Excellence, choice of fluid has been the source of much discussion. In particular, there has been extensive debate over the choice between a balanced crystalloid (such as Hartmann’s solution, Ringer’s lactate, or PlasmaLyte) and normal saline (an unbalanced crystalloid). There have been very few high-quality studies, but latest evidence from critically ill patients points to marginal benefits from using a balanced crystalloid in preference to saline.[154]
- A 2018 US multicentre cluster-randomised trial among 15,802 critically ill adults receiving care in the intensive care unit found small benefits from balanced crystalloid compared with saline. The 30-day outcomes showed 10.3% mortality in the balanced crystalloid group compared with 11.1% in the saline group (P = 0.06), and a major adverse kidney event rate of 14.3% compared with 15.4% in the two groups, respectively (marginal odds ratio 0.91, 95% CI 0.84 to 0.99).[154]
Colloids (e.g., starches, dextrans, gelatins, albumin, or fresh frozen plasma) are no longer used in emergency medicine in the UK.



Debate: Volume and rate of fluids

Previous versions of the Surviving Sepsis Campaign (SSC) guidelines recommended a protocoled approach to resuscitation, otherwise known as early goal-directed therapy (EGDT).[161][162][163] EGDT involves the use of a series of ‘goals’ including central venous pressure and central venous oxygen saturation (ScvO₂). This recommendation was largely based on data from one study that showed a significant survival benefit for patients receiving EGDT.[155]
This approach has since been challenged following the failure to show a mortality reduction in three subsequent large multicentre randomised controlled trials: PROCESS, ARISE, and PROMISE.[156][157][158]
- It is worth bearing in mind that these three trials included patients who were less severely ill (lower baseline lactate levels, ScvO₂ at or above the target value on admission, and lower mortality in the control group) than the patients in the original study that outlined EDGT as a recommended approach.
Based on this latest evidence, the SSC no longer specifically recommends EDGT; it does, however, acknowledge the need for guidance on how to approach this group of patients who have significant mortality and morbidity.[43] The SSC therefore recommends that you:[43]
- View these patients as having a medical emergency that necessitates urgent assessment and treatment
- Begin initial fluid resuscitation with 30 mL/kg of crystalloid within the first 3 hours
  - This fixed volume of fluid enables initiation of resuscitation while giving an opportunity to ascertain more specific information about the patient and while awaiting more precise measurements of haemodynamic status
  - Although scant data are available to support this volume of fluid, interventional studies have described this as usual practice in the early stages of resuscitation, and observational evidence supports the practice[159][160]
  - The average volume of fluid pre-randomisation given was approximately 30 mL/kg in the PROCESS and ARISE trials, and approximately 2 L in the PROMISE trial.[156][157][158] The SSC acknowledges that many patients will need more fluid than this, and it advocates giving further fluid to this group in line with functional haemodynamic measurements.[43]
The UK National Institute for Health and Care Excellence guideline on sepsis focuses on initial management and treatment, and therefore makes no recommendations regarding intensive monitoring such as that used in EGDT.[3]

Practical tip

To guide the need for further intravenous fluids, it can sometimes be helpful to use bedside ultrasound to monitor changes in inferior vena cava (IVC) diameter during respiration.[164][165]

In the spontaneously breathing patient: consider additional fluid resuscitation if there is a collapsed (or collapsing) IVC.
In the mechanically ventilated patient: an increase in IVC size >18% (or visible to the naked eye) with positive pressure ventilation suggests fluid-responsiveness.

Practical tip

Use the passive leg-raising test to predict fluid-responsiveness if adequate monitoring is available.[66][166]

This is a useful indicator of fluid-responsiveness, which should be assessed using devices that can continuously monitor cardiac output in real time (e.g., Pulse index Continuous Cardiac Output (PiCCO) monitor or oesophageal Doppler), usually in an intensive care unit rather than general ward setting.
Sit the patient upright at 45° and tilt the entire bed through 45°.
Patients with a positive test have a >10% increase in cardiac output or stroke volume, indicating more fluids may be required.
The passive leg-raise response may be misleading in conscious patients who are uncomfortable or in pain when lying flat.

Plus – monitor comorbidities during hospital stay

Consider – oxygen

Treatment recommended for SOME patients in selected patient group

If indicated, give oxygen to maintain target oxygen saturations 94% to 96%.[3][43][46][47] Latest evidence suggests that liberal use of supplemental oxygen (target SpO₂ >96%) in acutely ill adults is associated with higher mortality than more conservative oxygen therapy.[138] Therefore, a reasonable approach in practice is to maintain a target oxygen saturation of 94% to 96% in acutely ill patients who are not at risk of hypercapnia.

Target saturation of 88% to 92% if the patient is at risk of hypercapnic respiratory failure (e.g., those with COPD).[3][43][46][47]



Evidence: Target oxygen saturation in acutely ill adults

Too much supplemental oxygen increases mortality.

Latest evidence supports a 96% upper limit for target oxygen saturation in non-hypercapnic acutely ill adults.

Guidelines differ in their recommendations on target oxygen saturation in acutely unwell adults who are receiving supplemental oxygen. The 2017 British Thoracic Society guideline recommends a target SpO₂ range of 94% to 98% for patients not at risk of hypercapnia,[47] whereas the 2015 Thoracic Society of Australia and New Zealand guideline recommends 92% to 96%.[167]
A 2018 systematic review including a meta-analysis of data from 25 randomised controlled trials found that in adults with acute illness, liberal oxygen therapy (broadly equivalent to a target saturation >96%) is associated with higher mortality than conservative oxygen therapy (broadly equivalent to a target saturation ≤96%).[138] In-hospital mortality was 11 per 1000 higher for the liberal oxygen therapy group versus the conservative therapy group (95% CI 2-22 per 1000 more). Mortality at 30 days was also higher in the group who had received liberal oxygen (risk ratio 1.14, 95% CI 1.01 to 1.29). The trials included adults with sepsis, critical illness, stroke, trauma, myocardial infarction, and cardiac arrest, and patients who had emergency surgery. Studies that were limited to people with chronic respiratory illness or psychiatric illness, patients on extracorporeal life support, patients receiving hyperbaric oxygen therapy, or those having elective surgery were all excluded from the review.
An upper SpO₂ limit of 96% is therefore reasonable when administering supplemental oxygen to medical patients with acute illness who are not at risk of hypercapnia. However, a higher target may be appropriate for some specific conditions (e.g., pneumothorax, carbon monoxide poisoning, cluster headache, and sickle cell crisis).[168]

Plus – monitor comorbidities during hospital stay

Consider – standard intensive care unit supportive care

Treatment recommended for SOME patients in selected patient group

Bear in mind that some patients (e.g., those who are frail) may not be suitable for management in intensive care settings. Consider the patient’s baseline health including their resuscitation status when determining the limits of treatment. Use this to feed into a personalised care plan appropriate to the individual patient.

Consult local protocols for specific escalation routes but in general:

Ensure immediate review by a senior clinician (CT3/ST3 or higher in the UK) of any patient with a National Early Warning Score 2 (NEWS2) score of 5 or more, or who meets one or more of the UK National Institute for Health and Care Excellence (NICE) sepsis high-risk criteria. Also ensure the patient is discussed with a consultant[3][41]
Discuss with the admitting consultant[3] and consider alerting critical care immediately if the patient is acutely unwell and:
- Has a NEWS2 score of 7 or more, persisting high lactate (more than 4 mmol/L [36 mg/dL]) despite fluid resuscitation, or a systolic blood pressure of less than 90 mmHg[3]
  - Discuss with the admitting consultant[3]
- Has hypotension that doesn’t respond to initial fluid resuscitation
- Is likely to require central venous access and the initiation of inotropes or vasopressors[3]
- Has any feature of septic shock
  - See our topic Shock
- Has neutropenia
  - See our topic Febrile neutropenia
- Is immunodeficient
Urgently discuss with a consultant or call them to attend if the patient:[41]
- Is treated with intravenous antibiotics and/or a fluid bolus for sepsis
- Does not respond to initial therapy (antibiotics/fluid resuscitation/oxygen) within the first hour.[3][41] Failure to respond to treatment is defined as:[3]
  - Systolic blood pressure remains less than 90 mmHg
  - Persistent reduced level of consciousness
  - Respiratory rate more than 25 breaths per minute or the new need for mechanical ventilation
  - Lactate has not reduced by more than 20%
Refer to critical care any patient who is likely to require central venous access and initiation of inotropes or vasopressors[3]
- This includes any patient with evidence of circulatory dysfunction or shock, or those who do not respond to initial therapy (as outlined above)
- ECG can be used to determine which vasoactive drug(s) to proceed with in critical care.[43]

Practical tip

Ensure a clear escalation plan has been discussed and agreed with the clinical team; include specific points of contact for nursing staff if you are leaving a patient for later review.

Involve a senior colleague and/or consider transferring to critical care sooner rather than later if the patient is not improving, or deemed high-risk. Examples include if the patient:

Is not responding to fluids
Needs inotropic support
Has a low Glasgow Coma Scale score
Needs ventilatory support.

For any patient with suspected sepsis, consider the need for referral to a high-dependency unit for management by the critical care team.[155][174]

The following interventions should only be initiated by experienced members of the critical care team:[3][43][175]

Glycaemic control
Vasoactive drugs (vasopressors/inotropes)
Corticosteroids.

Additional intensive care measures that will be considered include:[43][176][177]

Stress ulcer prophylaxis (in people at risk of gastrointestinal bleeding)
- With an H2 antagonist or proton-pump inhibitor
Deep venous thrombosis prophylaxis
- With heparin and compression stockings
Enteral or parenteral nutrition
Administration of human albumin solution 4% to 5% in patients with sepsis and shock[3][43] who have not responded to substantial volumes of crystalloids
Transfusion of packed cells
- Consult local protocols for recommended threshold
- The Surviving Sepsis Campaign recommends using a threshold of 70 g/L (7 g/dL).[43]



Evidence: Threshold for transfusion of packed cells

Studies in the general critical care population have shown no improvement with blood transfusions given at a higher haemoglobin threshold compared with a lower haemoglobin threshold,[178]and have shown potential harm associated with liberal transfusion.[179]

One multicentre parallel group randomised trial analysed data from 998 patients with septic shock, split into two intervention groups with similar baseline characteristics. In the intensive care unit, the group assigned to blood transfusion at a lower haemoglobin threshold received a median of 1 unit of blood (interquartile range, 0 to 3) and the group assigned to blood transfusion at a higher haemoglobin threshold received a median of 4 units (interquartile range, 2 to 7).[178]
- At 90 days after randomisation, 216 of 502 patients (43.0%) assigned to the lower-threshold group, compared with 223 of 496 (45.0%) assigned to the higher-threshold group, had died (relative risk 0.94, 95% CI 0.78 to 1.09; P = 0.44).
- The results were similar in analyses adjusted for risk factors at baseline and in analyses of the per-protocol populations.
- The numbers of patients who had ischaemic events, who had severe adverse reactions, and who required life support were similar in the two intervention groups.
A second multicentre randomised controlled trial compared the rates of death from all causes at 30 days and the severity of organ dysfunction in 838 critically ill patients receiving a restrictive strategy of red-cell transfusion compared with those receiving a liberal strategy.[179]
- Overall, 30-day mortality was similar in the two groups (18.7% vs. 23.3%, P = 0.11).
- However, the rates were significantly lower with the restrictive transfusion strategy among patients who were less acutely ill and in patients younger than 55 years of age (5.7% and 13.0%, respectively; P = 0.02), but not in those with clinically significant cardiac disease (20.5% and 22.9%, respectively; P = 0.69).
- The mortality rate during hospitalisation was significantly lower in the restrictive-strategy group (22.3% vs. 28.1%, P= 0.05).

In the initial resuscitative phase, transfusion to achieve a higher haematocrit of ≥30% may be appropriate.[155]

Titrate fraction of inspired oxygen (FiO₂) to lowest effective levels to prevent oxygen toxicity and maintain central venous oxygen tension.
Place patients in a semi-recumbent position with the head elevated to 30° to 45°.[43]

Glycaemic control

Although patients with sepsis are often hyperglycaemic, the optimal glucose target is unknown.

The Surviving Sepsis Campaign guideline recommends targeting a blood glucose level <10.0 mmol/L (<180 mg/dL).[43] The guideline also recommends a ‘sliding scale’ variable-rate intravenous insulin infusion.[43]

NICE makes no recommendations on glycaemic control in sepsis.[3]



Evidence: Glycaemic control

An international randomised controlled trial (RCT) of 6104 critically ill medical and surgical patients found increased 90-day mortality (odds ratio 1.14, 95% CI 1.02 to 1.28) with tighter glucose control, possibly due to more frequent episodes of hypoglycaemia.[183]
A 2010 systematic review of 6 RCTs and a meta-analysis investigating tight glucose control (4.4 to 6.1 mmol/L [80-110 mg/dL]) versus less strict glucose control in critically ill patients in the intensive care unit setting found no significant improvement in mortality with tight glucose control, but it was associated with significantly more hypoglycaemic episodes compared with less strict glucose control.[184]
An RCT of critically ill patients in a primarily surgical intensive care setting found lower patient mortality with tight glucose control, 4.4 to 6.1 mmol/L (80-110 mg/dL), compared with ‘conventional’ more liberal glucose control.[185]

Plus – monitor comorbidities during hospital stay

Consider – vasopressor (should only be initiated by experienced members of the critical care team)

Treatment recommended for SOME patients in selected patient group

Vasopressors are used in a critical care setting to maintain a mean arterial pressure (MAP) ≥65 mmHg if the patient is unresponsive to fluid resuscitation.[3][43][175]

Failure to respond to initial fluid resuscitation is a sign of septic shock.[1]
Noradrenaline (norepinephrine) is the vasopressor of choice, mainly because it increases MAP.[43]
- Noradrenaline is the vasopressor recommended by the Surviving Sepsis Campaign guideline.[43] The UK National Institute for Health and Care Excellence makes no recommendation on the choice of vasopressor.[3]
- If further vasopressor therapy is required to maintain adequate blood pressure or the noradrenaline dose needs to be reduced, add vasopressin or adrenaline (epinephrine) to noradrenaline.
- Dopamine is an option, but has been associated with higher mortality than noradrenaline.[186][187] [ ] Therefore, it is only recommended in patients with a low risk of tachyarrhythmias and bradycardia.[43] It is rarely used in the UK. Do not use low-dose dopamine for renal protection.[43]

Practical tip



Evidence: Choice of vasopressor

Primary options

noradrenaline (norepinephrine): 0.4 to 0.8 mg/hour intravenous infusion initially, adjust dose according to response

Secondary options

noradrenaline (norepinephrine): 0.4 to 0.8 mg/hour intravenous infusion initially, adjust dose according to response

-- AND --

vasopressin: 0.01 units/minute intravenous infusion initially, adjust dose according to response, maximum 0.03 units/minute

adrenaline (epinephrine): 2-10 micrograms/minute intravenous infusion initially, adjust dose according to response

dopamine: 2-5 micrograms/kg/minute intravenous infusion initially, adjust dose according to response

Plus – monitor comorbidities during hospital stay

Consider – inotrope (should only be initiated by experienced members of the critical care team)

Treatment recommended for SOME patients in selected patient group

Inotropes can be considered for patients with low cardiac output despite adequate fluid resuscitation and vasopressor therapy.[3][43]

Dobutamine is recommended first line by the Surviving Sepsis Campaign guideline for people with measured or suspected low cardiac output in the presence of adequate left ventricular filling pressure (or clinical assessment of adequate fluid resuscitation) and adequate mean arterial pressure.[43]
The UK National Institute for Health and Care Excellence makes no specific recommendations on inotrope selection in patients with sepsis.[3]

Practical tip

When using inotropes, keep the patient’s heart rate at less than 100 beats per minute to minimise myocardial ischaemia.[175]

Primary options

dobutamine: 2.5 to 10 micrograms/kg/minute intravenous infusion initially, adjust dose according to response, maximum 40 micrograms/kg/minute

Plus – monitor comorbidities during hospital stay

Consider – corticosteroid (should only be initiated by experienced members of the critical care team)

Treatment recommended for SOME patients in selected patient group

The UK National Institute for Health and Care Excellence does not give any recommendations on the use of corticosteroids for managing sepsis in adults.[3]



Evidence: Benefits and harms of corticosteroids

Corticosteroids are a late critical-care intervention for patients in whom all other attempts to raise their blood pressure have failed.[3] In this critically ill group, corticosteroids may result in a small reduction in mortality.[189][190] Possible harms include an increased risk of neuromuscular weakness, hyperglycaemia, and hypernatraemia with corticosteroids compared with no corticosteroids.[189]

An international panel reviewing the inconsistent conclusions of large randomised controlled trials (RCTs) on the topic suggested in 2018 that the evidence for the use of corticosteroids was weak, but that “fully informed patients who value avoiding death over quality of life and function would likely choose corticosteroids”, although a no-corticosteroid approach remained reasonable.[189] This judgement was based on an assessment that corticosteroids may reduce mortality by around 2% (this effect was seen in sepsis with and without shock although the greatest benefit was among patients with septic shock), but can increase the risk of neuromuscular weakness and resulting functional deterioration. Most notably, the panel reviewed the following trials:
- ADRENAL: 3658 patients who had septic shock[191]
  - No statistically significant difference in 90-day mortality between the hydrocortisone and placebo groups
- APROCCHSS: 1241 patients who had septic shock[192]
  - Hydrocortisone plus fludrocortisone reduced 90-day mortality.
  [Figure caption and citation for the preceding image starts]: BMJ Rapid Recommendations: intravenous corticosteroids plus usual care versus usual care onlyLamontagne F, et al. BMJ 2018;362:k3284 [Citation ends].
A more recent systematic review and meta-analysis (comprising 37 RCTs, incorporating both ADRENAL and APROCHSS) included 9564 people with sepsis.[190] The review found corticosteroid use to be associated with significant improvement in healthcare outcomes as shown by:[190]
- Reduced 28-day mortality (risk ratio [RR] 0.90, 95% CI 0.82 to 0.88) compared with placebo or standard supportive care
- Reduced intensive care unit mortality (RR 0.85, 95% CI 0.77 to 0.94; I² = 0%) compared with placebo or standard supportive care
- Reduced in-hospital mortality (RR 0.88, 95% CI 0.79 to 0.99; I² = 38%) compared with placebo or standard supportive care.
However, corticosteroid use was also associated with increased risk of hyperglycaemia (RR 1.19, 95% CI 1.08 to 1.30) and hypernatraemia (RR 1.57, 95% CI 1.24 to 1.99) compared with placebo or standard supportive care.[190]

Primary options

hydrocortisone sodium succinate: 200 mg/day intravenously

Plus – monitor comorbidities during hospital stay

Use of this content is subject to our disclaimer

Sepsis in adults

Treatment algorithm

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Glycaemic control

Practical tip

Primary options

Secondary options

These drug options and doses relate to a patient with no comorbidities.

Primary options

Secondary options

Drug choice, dose and interactions may be affected by the patient's comorbidities. Check your local drug formulary.

Primary options

Secondary options

Practical tip

Primary options

These drug options and doses relate to a patient with no comorbidities.

Primary options

Drug choice, dose and interactions may be affected by the patient's comorbidities. Check your local drug formulary.

Primary options

Primary options

These drug options and doses relate to a patient with no comorbidities.

Primary options

Drug choice, dose and interactions may be affected by the patient's comorbidities. Check your local drug formulary.

Primary options

Practical tip

Practical tip

Respiratory

Abdominal

Urinary tract

Soft tissue and joint

Practical tip

Central nervous system

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Practical tip

Glycaemic control

Practical tip

Primary options

Secondary options

These drug options and doses relate to a patient with no comorbidities.

Primary options

Secondary options

Drug choice, dose and interactions may be affected by the patient's comorbidities. Check your local drug formulary.

Primary options

Secondary options

Practical tip

Primary options

These drug options and doses relate to a patient with no comorbidities.

Primary options

Drug choice, dose and interactions may be affected by the patient's comorbidities. Check your local drug formulary.

Primary options

Primary options

These drug options and doses relate to a patient with no comorbidities.

Primary options

Drug choice, dose and interactions may be affected by the patient's comorbidities. Check your local drug formulary.

Primary options