- Front matter
- Definition of septic shock
- Source of septic shock & evaluation for source
- Interventions of no real value
- Resuscitative endpoints ??
- Questions & discussion
- PDF of this chapter (or create customized PDF)
Septic shock is the defining illness of medical critical care. It is important because it is common, potentially lethal, and highly treatable. The importance of septic shock has attracted attention, guidelines, politics, and controversy. Which makes writing this chapter challenging, because – let's admit it – everyone has a different approach to treating septic shock.
This chapter describes my approach to septic shock. It is evidence-based, but there's inadequate evidence available to prove that it is superior to innumerable other ways of treating septic shock. Overall, it is a fairly simple approach that doesn't involve a lot of fancy gadgetry. As such, this approach should be accessible to a wide variety of practitioners.
Like the remainder of the IBCC, this chapter will be aggressively written and then aggressively revised. Some bold claims will be made. If readers find evidence to overturn them, I will make revisions accordingly. The ability to write controversial material and then revise it based on post-publication peer review is a unique strength of this medium. If you disagree with material below, please comment and include citation(s) of supporting evidence.
brief history of septic shock treatment
Rivers Trial, 2001
The history of modern sepsis care can largely be traced back to the Rivers Trial of Early Goal Directed Therapy in septic shock (11794169). This study used an aggressive resuscitative package to decrease mortality from sepsis. The study has numerous major flaws, which have grown more apparent over time:
- Major conflict of interest (the primary investigator held a patent for a catheter to measure mixed central venous oxygen saturation).
- Single-center design with conspicuous lack of blinding (patients in the experimental arm were often treated personally by Dr. Rivers).
- The mortality rate in the control group (47%) was quite high. This suggests either a statistical anomaly, research misconduct, or lack of generalizability to other contexts.
- The fragility index was four (very respectable, but also not irrefutable).
- 25 patients were initially included in the study, but then removed for reasons which remain murky and contentious. Notably, the number of patients removed from the study is greater than the fragility index of four. According to an investigational report in the Wall Street Journal, an intention-to-treat analysis including these 25 patients would have given study negative results (McKenna 2008).
- Attempts to replicate the Rivers trial have failed. Over time, most components of the Rivers Protocol for Early Goal Directed Therapy have been explicitly disproven or have fallen out of favor (e.g. central venous pressure monitoring, nitroglycerine for septic shock, transfusion to hemoglobin >10 mg/dL, mixed venous oxygen saturation monitoring). Based on this fact alone, it must be concluded that the Rivers Trial was invalid (according to science, findings which cannot be replicated are not valid).
The Rivers Trial is a cautionary tale in evidence-based medicine. Much like the NINDS trial of tPA in ischemic stroke, this was an inspirational study which sparked widespread enthusiasm, leading to broad acceptance. If the Rivers Trial were published today, it would be met with considerable criticism and calls for immediate replication. Unfortunately, the exciting results of the trial and high profile of the NEJM allowed the Rivers Trial to misdirect thinking in septic shock for a decade.
The Rivers trial did promote an aggressive stance towards treating septic shock, which is beneficial. However, most of the interventions from the trial are incorrect. In order to move forwards with scientific and evidence-based therapy for septic shock, we need to have a clear-eyed view of this trial.
Surviving Sepsis Campaign
This campaign was initiated as a joint marketing effort by Eli Lilly and Edward Life Sciences to promote activated protein C (XIGRIS) and central venous catheters which measure the mixed venous oxygen saturation (yes, the same one Rivers held a patent for). The original backbone of the guidelines was the Rivers Trial.
The Surviving Sepsis Campaign has a track record of being sluggish to change based on the emergence of new data. For example, the Surviving Sepsis Campaign continued to recommend the use of central venous pressure and mixed venous oxygen saturation even after the PROCESS and ARISE trials demonstrated that these were non-beneficial. The campaign also has a history of making strong and arbitrary recommendations pulled out of thin air (e.g. 3-hour and 6-hour bundles of care involving fixed volumes of fluid resuscitation).
In 2018, the Surviving Sepsis Campaign issued an update recommending initiation of antibiotics and 30 cc/kg fluid bolus within sixty minutes of emergency department triage. This is a bizarre departure from prior recommendations, which came without any solid evidentiary basis. This update provoked widespread dissent, which led the Surviving Sepsis Campaign to temporarily withdraw it. Subsequently, however, these recommendations were re-issued.
Currently, the Surviving Sepsis Campaign lumbers on, serving as an impediment to the development of guidelines which are truly evidence-based. The Surviving Sepsis Campaign guidelines have formed the basis of governmental sepsis regulations in the United States (SEP-1), which thwart individualized care and are frankly dangerous.
- Mythbusting the Surviving Sepsis 2016 guidelines
- Petition to retire the surviving sepsis guidelines
- The Survivign Sepsis 1-hour guidelines are… back?
- Marik et al. Pro-Con debate: Should the Surviving sepsis campaign guidelines be retired? CHEST 2019; 155:12-20.
- Kalantari A and Rezaie SR. Challenging the one-hour sepsis bundle. Western Journal of Emergency Medicine, in press.
- Spiegel R et al. The 2018 Surviving Sepsis Campaign's Treatment Bundle: When guidelines outpace the evidence supporting their use. Annals of Emergency Medicine, 2018.
definition of septic shock
The consensus definition of septic shock was updated from Sepsis-II to Sepsis-III recently. These definitions may be summarized as follows:
- Sepsis II definition of septic shock: Infection causing persistent hypotension, despite fluid resuscitation.
- Sepsis III definition of septic shock: Infection causing vasopressor requirement to maintain a MAP > 65 mm (despite fluid resuscitation) plus a serum lactate >2 mM.
Neither of these definitions is perfect, but they focus our attention on two key bits that together help delineate septic shock:
- Bit #1: Overt hypotension
- Bit #2: Hyperlactatemia – this is generally a reflection of aerobic lactate production due to endogenous epinephrine. It may be grossly conceptualized here as a measurement of the patient's endogenous epinephrine release.
Sepsis III requires both bits, which is a mistake because they measure different things:
Lactate is not specific to septic shock. Lactate may be elevated by a panoply of conditions (including any shock state, physiologic stress, beta-agonists, seizure, or hepatic dysfunction). Elevated lactate does often identify patients who are at increased risk of mortality, who require more intensive investigation and treatment.
Ultimately, formal definitions are more relevant to clinical trials than bedside management of individual patients. These blunt tools are inadequate to direct patient management. Instead, a diagnosis of septic shock should be made carefully on an individual basis, using clinical judgement and consideration of the following factors:
- (1) Type of underlying infection
- Some infections (e.g. necrotizing fasciitis, ascending cholangitis) are more likely to cause septic shock. There should be a lower threshold to diagnose septic shock and initiate aggressive management in these patients.
- (2) Degree of hemodynamic instability
- Shock index (heart rate / systolic blood pressure)
- Blood pressure, compared to patient's baseline pressure
- Vasopressor requirement
- Evidence of end-organ hypoperfusion (e.g. urine output, skin perfusion)
- (3) Degree of hyperlactatemia
- Presence of other factors which may increase lactate levels (e.g. albuterol, hepatic dysfunction).
- Lactate >4 mM suggests a significant mortality (28248722).
- (4) Other end-organ failures (e.g. delirium, shock liver)
Septic shock encompasses a broad range of infections in a diverse range of patients. The table below shows common signs and symptoms of sepsis. This can be a difficult diagnosis, because different patients will present with different constellations of these findings.
In many cases, septic shock may be suspected before the underlying infection is definitively diagnosed. In such cases, it is generally best to empirically treat as if the patient has septic shock, while obtaining additional information (e.g., culture data, definitive CT imaging).
- For patients who are clinically in shock with no obvious cause (despite evaluation of history, physical exam, and echocardiography), there should be a high index of suspicion for septic shock.
With increased focus on sepsis and septic shock, other disorders are increasingly likely to be misdiagnosed as septic shock (27692840). Common mimics and their evaluation are shown here:
common sources & evaluation of source
- ? Localizing symptoms
- ? Travel to areas with endemic infections (e.g. tick-borne infections, malaria)
- ? Hardware (e.g. prosthetic joints, pacemakers, implanted ports, central lines)
- Physical exam: see figure above
- Basic labs: Chem-10 (including Ca/Mg/Phos), CBC with differential, Coags, Liver function tests
- Cultures: Blood cultures x2, culture of any indwelling line present >48 hours
- Urinalysis and culture
- Specific tests directed at other potential infections (e.g. PCR for anaplasma and ehrlichia; more here)
- Chest X-ray
- Low threshold to obtain CT abdomen/pelvis if source of infection remains unclear. Especially in elderly patients, severe abdominal pathology may occur with minimal findings on history and physical exam. Consider a CT angiogram if mesenteric ischemia is possible.
- Additional tests as clinically warranted, for example:
- CT head and lumbar puncture if altered mental status or nuchal rigidity
- Ultrasound evaluation of any potential sites of infection
- Paracentesis if ascites (even if previously present)
increase MAP>65mm promptly with peripheral vasopressor infusion
- Correlational evidence suggests that a longer duration of hypotension increases the risk of renal failure.
- The traditional strategy of flogging the patient with fluid for hours before starting pressors is ill-conceived.
- Peripheral vasopressor may be started without delay to support the blood pressure.
- If the patient improves following fluid resuscitation, then pressor may be weaned off.
- If vasopressor requirements escalate, then transition to a central line may be considered.
choice of the first pressor: there is no “first-line” agent
- Traditional dogma favors a specific sequence of vasopressors for all patients (typically beginning with norepinephrine, then with sequential addition of vasopressin, and finally epinephrine). Evidence doesn't really support this:
- Initial vasopressor support with vasopressin was potentially beneficial in the VANISH trial (27483065). This shows that a pure vasoconstrictor can be used as a front-line pressor for septic shock.
- Epinephrine was shown to yield similar results to norepinephrine in the CAT trial 😺 (18654759) .
- The only prospective study comparing norepinephrine to phenylephrine found nearly identical hemodynamic responses (19017409).
- For most patients, norepinephrine is a good choice. However, the selection of pressors may be individualized based on patient physiology, as shown below.
- The only pressor which shouldn't be used is dopamine (based on evidence of harm in prospective RCTs)(20200382, 26323041).
- The main concern with peripheral vasopressor infusion is extravasation into the skin leading to necrosis. Fortunately, this is rare. Overall, the risk of skin necrosis should never be prioritized over the risk of systemic hypoperfusion and death.
- Skin necrosis is both less common and less important than mortality and systemic hypoperfusion.
- Norepinephrine is safe for peripheral infusion, but the risk of extravasation does increase if infusions are used for prolonged periods of time (25669592). Peripheral norepinephrine has been validated in a medical ICU with a rigorous protocol for placing IV lines and monitoring their function closely (26014852).
- Phenylephrine or epinephrine are the safest agents to use peripherally. Both agents have been administered via a subcutaneous route intentionally in the past. There appear to be no reports in the literature of these agents' causing skin necrosis. It's sensible to avoid infusion into the wrist or hand, but overall these agents are unlikely to cause necrosis if extravasation occurs.
- Note that phenylephrine infusion is physiologically very similar to norepinephrine.
- Vasopressin administration peripherally should probably be avoided. If vasopressin extravasates, there is no way to counteract its effect (unlike catecholamine extravasation, which can be treated with local infiltration with phentolamine).
- Vasopressin is a non-catecholamine pure vasoconstrictor that may improve renal function (compared to norepinephrine).
- Drawbacks of vasopressin:
- Hard to titrate (half-life of ~30 minutes, so it will take a long time to reach steady state). Vasopressin can be used as a titratable pressor at doses ranging from 0-0.06 U/min, but don't expect dose adjustments to have an immediate effect (27483065).
- The combination of vasopressin plus norepinephrine can cause digital ischemia and necrosis.
- How to use vasopressin?
- Best for patients with warm extremities (vasodilatory shock) and acute kidney injury.
- Monitor extremity perfusion and stop vasopressin immediately if digital ischemia occurs.
- Consider combining vasopressin with epinephrine (rather than norepinephrine). Theoretically, this may allow taking advantage of the renal-perfusion benefits of vasopressin while avoiding excessive vasoconstriction.
epinephrine might be the preferred inotrope for septic shock
- Traditionally, there has been a preference for catecholamine inotrophes that selectively affect the beta-1 receptor. This is based on the following two concepts, which are both wrong:
- Epinephrine may actually have advantages over dobutamine:
- (1) Epinephrine may have greater efficacy as an inotrope (figure below).
- (2) Dobutamine causes vasodilation, which may decrease the blood pressure and exacerbate vasodilatory shock. This can cause problems if the dobutamine isn't titrated very thoughtfully. In contrast, epinephrine contains alpha-adrenergic stimulation which prevents it from having a net vasodilatory effect.
- Titration of multiple vasopressors involves avoiding excessive vasoconstriction or excessive inotropy (figure below). The optimal balance will vary between patients.
- When in doubt, the best approach is often to empirically up- and down-titrate pressors in order to sort out what the various agents are doing.
- If a patient strongly responds to up/down titration of an agent, that drug is more likely to be causing benefit.
- If a patient has no/minimal response to adjusting the dose of an agent, that drug probably isn't helping much (and may be causing harm).
- Norepinephrine down-titration: In addition to adding pressors, it's also useful to down-titrate pressors which aren't helping. Sometimes norepinephrine may over-constrict patients, leading to excessive afterload and a drop in ejection fraction (the heart is unable to tolerate the excessive afterload). If adjusting the norepinephrine dose doesn't affect blood pressure much, that suggests that the norepinephrine dose is excessive. The goal should always be to use the minimal dose of vasopressor(s) necessary to achieve hemodynamic targets.
- Contraindications to epinephrine challenge:
- Significant tachycardia (e.g. heart rate >120 b/m)
- Echocardiography shows the left ventricle is already hyperkinetic.
- Indications for epinephrine challenge
- (1) Hypoperfusion (e.g. poor urine output, cold extremities, mottling, poor capillary refill)
- -plus, ideally-
- (2) Some indicator that epinephrine might help, for example
- i) Bedside echo shows ejection fraction is reduced or (inappropriately) normal
- ii) Heart rate is inappropriately slow or normal (e.g. <80 b/m)
- iii) Lactate is inappropriately low relative to the severity of illness (inappropriate normolactatemia, suggesting an inadequate endogenous sympathetic response)
- How to perform: Start an epinephrine infusion at ~4-5 mcg/min.
- Judging response to the epinephrine challenge:
- (1) Effect on skin perfusion (e.g. capillary refill time, mottling, and temperature).
- (2) Effect on blood pressure. On average, epinephrine has a relatively equivalent effect on blood pressure compared to norepinephrine (e.g. 1 mcg/min epinephrine ~ 1 mcg/min norepinephrine). If the epinephrine has a considerably greater effect on blood pressure than norepinephrine, this suggests that it's helping (e.g. adding 4 mcg/min epinephrine allows the norepinephrine dose to be decreased by 10 mcg/min).
- (3) Effect on heart rate – an excessive tachycardic response to epinephrine may be harmful.
- If the patient responds well to epinephrine, then this should be continued. In many cases, other vasopressors may be down-titrated (try to achieve MAP and perfusion targets with the minimal cumulative dose of vasopressor).
- If the patient responds poorly to epinephrine, then stop it.
- Early norepinephrine to stabilize the MAP
- Epinephrine challenge
- Peripheral pressors
major considerations in antibiotic selection
- Any prior culture results (especially drug-resistant organisms)
- Antibiotic allergies
- Possible source(s)
- Recent antibiotic exposures (consider using an agent the patient hasn't been recently exposed to)
- A broad-spectrum beta-lactam is generally the most important antibiotic used.
- Due to increasing resistance and toxicity, fluoroquinolones are a poor choice for treatment of critically ill patients.
- Don't be scared off by beta-lactam allergies. As explored in this chapter, there is less cross-allergic response between different antibiotics than generally believed.
- Even in a patient with a history of numerous, severe allergies, meropenem can generally be used.
- Good choices
- Avoid third-generation cephalosporins (e.g. ceftazidime or ceftriaxone) for the following reasons:
- Sub-optimal gram-positive coverage (even if vancomycin is given, vancomycin levels will often be inadequate).
- Failure to cover gram-negatives with inducible beta-lactamase (e.g. Serratia or Enterobacter spp.) – even if they may appear to cover these organisms on an antibiogram (more on this here).
- Not every septic patient needs vancomycin!
- MRSA isn't a pathogen involved in urosepsis or community-acquired intra-abdominal infection. Patients with these sources of infection don't need MRSA coverage.
- MRSA coverage should be considered for patients with pneumonia, soft-tissue infection, line infection, or endocarditis.
- Either vancomycin or linezolid may be used.
- If vancomycin is chosen, it should be dosed correctly (ideally with pharmacokinetic monitoring).
- Linezolid has no risk of nephrotoxicity and is easier to dose adequately.
coverage of other pathogens
- Atypical pulmonary pathogens should be covered in patients with pneumonia (azithromycin or doxycycline).
- Possible tick-borne illness may be covered with doxycycline.
- Clostridioides difficile coverage in patients with colitis or diarrhea (especially if recent antibiotic exposure).
- Clindamycin may be considered for toxin suppression in patients with necrotizing fasciitis, toxic shock syndrome, or severe group A streptococcal infections (e.g. streptococcal cellulitis).
- Approach to allergy to various beta-lactam antibiotics (IBCC chapter)
- Antibiotics (IBCC chapter)
Failure to achieve source control might be the most common cause of death from septic shock within a modern healthcare system. All other interventions described in this chapter will often fail if there is inadequate procedural source control.
common examples of source control
- Infected hardware (e.g. catheter) may need to be removed.
- Nephrolithiasis causing obstruction and infection may require decompression.
- Ascending cholangitis requires ERCP or percutaneous intervention for decompression.
- Perforated or obstructed bowel requires surgical repair.
- Necrotizing fasciitis may require debridement.
investigation regarding possible need for source control
- Ensure that the source has been fully investigated (e.g. a patient with urosepsis should be imaged to exclude obstruction).
- For patients in whom no infection is identified, aggressive imaging may be helpful to search for the source (e.g. CT abdomen/pelvis).
Rivers of ink have been spilled discussing the optimal strategy for fluid administration in septic shock. However, there's little evidence that fluid administration is truly beneficial. Excessive fixation on fluid status may serve to divert attention from other more important aspects of care.
foundational concepts of fluid management in septic shock
- The primary physiologic problems in septic shock are vasodilation and maldistribution of blood to organs (sometimes with cardiac dysfunction as well). None of these problems can be solved with fluid administration.
- The vast majority of crystalloid administered will leak out of vasculature into the interstitial tissue (for example, 95% in some studies of fluid boluses!)(22165353).
- Fluid bolus therapy isn't evidence-based and should arguably be avoided when possible (partially due to #2 above).
- There is no high-quality evidence that large-volume fluid resuscitation is beneficial in septic shock. Available RCT evidence shows that fluid administration may be harmful (27686349, 28973227).
- A low central venous pressure and collapsed inferior vena cava may result from vasodilation and distribution of blood out of the central veins (rather than true hypovolemia; explained further here). Therefore, a collapsed IVC shouldn't be interpreted as an indication to give fluid.
- Fluid responsiveness has never been shown to improve clinical outcomes in septic shock (likely due to #1-3 above). Likewise, sophisticated hemodynamic monitoring (e.g. Swan-Ganz catheter, vigelo FloTrac) has never been shown to improve clinical outcomes (29149934).
- Every physician is convinced that they understand how to apply fluids to beneficial effect, yet there is zero agreement on this topic. This demonstrates the existence of widespread over-estimation of our ability to use fluids in a beneficial manner.
video explaining why fluid boluses are rarely helpful (more on this here)
initial fluid resuscitation (generally in the emergency department)
- The optimal strategy for initial volume resuscitation is unknown. This is currently the subject of ongoing clinical trials (e.g. CLOVERS trial).
- The above schema seems reasonable, but will obviously need to be tailored to the individual patient.
- For patients with pneumonia and mild hypotension, their primary physiologic problem is often hypoxemia. If you're concerned primarily about pulmonary decompensation then fluid won't help – such patients may benefit from vasopressor support rather than fluids.
management of fluid status following initial resuscitation
Yep, looks something like this:
Is the pt in the ICU?
Stop giving fluid!
— Rory Spiegel (@EMNerd_) June 7, 2019
- Unless the patient has substantial ongoing fluid losses (e.g. severe diarrhea), it may be wise to stop giving additional crystalloid following the initial resuscitation.
- Most patients will receive ~1.5 liters per day of fluid along with various infusions and antibiotics. The addition of enteral nutrition will often increase this to >2-3 liters per day! This fluid alone is already excessive (without the use of any additional crystalloid).
- Follow electrolytes daily. If hypernatremia develops then the patient has a free water deficiency which must be treated with enteral water or intravenous D5W.
- A common mistake is to continue large-volume fluid resuscitation for too long. For example, a patient may receive 2-3 liters of fluid in the emergency department (appropriately), and then the ICU team will re-resuscitate with an additional 2-3 liters of fluid (inappropriately). Then a different ICU team will rotate on shift and resuscitate the patient a third time!
- Another related mistake is not keeping track of fluid inputs during transitions in care.
- Available evidence supports a fluid-restrictive strategy following initial resuscitation (CLASSIC & FACT trials). In the CLASSIC trial, additional fluid caused no improvement in hemodynamics, vasopressor dose, or urine output. In fact, a liberal fluid strategy seemed to increase the risk of kidney injury (27686349).
- Serial assessment of fluid responsiveness may lead to a vicious cycle of perpetual fluid administration (figure below).
- Fluid administration does often cause a transient improvement in hemodynamics, which reinforces this misguided behavior.
- Fluid micro-management (e.g. serial echocardiography with fluid boluses) is extremely time-consuming and probably not beneficial (see: futility of fluid boluses above #2).
traditional dogma: over-resuscitation followed by de-resuscitation
- A traditional concept is that septic patients should initially be volume overloaded, and subsequently diuresed during their recovery (“you need to swell to get well, you need to pee to be free”).
- There is no evidence to support this concept. Volume overload isn't a viable therapeutic strategy. A better strategy seems to be targeting euvolemia and then keeping patients there.
- If the patient does become volume overloaded, diuresis should be performed as soon as it is safe to remove fluid. Diuresis can often be commenced ~24-48 hours after admission, once the patient is stabilizing hemodynamically and vasopressor doses are decreasing. Signs of intravascular volume overload (e.g. distension of the inferior vena cava, portal vein pulsatility) support the safety of diuresis.
- Fluid choice can be guided by the presence of pH abnormalities (pH-guided fluid resuscitation).
- For most patients, a balanced crystalloid is a good choice (e.g. Lactated Ringers or Plasmalyte).
- For patients with spontaneous bacterial peritonitis and/or hepato-renal syndrome, albumin is generally the fluid of choice.
- Landmark trials
- Fluid Responsiveness (EMCrit podcast 162)
- Empty IVC doesn't prove volume depletion
- Myth-busting the fluid bolus
- pH-guided resuscitation (IBCC chapter)
- Get SMART: 9 reasons to quit using normal saline
risk of steroid
- Stress-dose steroid (50 mg hydrocortisone IV q6hr) is equivalent to 50 mg prednisone daily. This dose of steroid is routinely used for outpatients with a myriad of diagnoses, without much fuss (e.g. asthma or COPD exacerbation).
- Stress-dose steroid doesn't increase the risk of super-infection. This concept was promoted by inappropriate interpretation of secondary outcomes in the CORTIUS trial (18184957). This fear has been debunked in meta-analyses, as well as in the much larger ADRENAL trial (19489712, 29347874).
- Large RCTs haven't detected a risk of steroid-induced myopathy. However, such a risk may exist in patients who are undergoing therapeutic paralysis.
- Steroid does increase the rate of hyperglycemia (shown in numerous RCTs above).
benefit of steroid
- Both the meta-analysis and ADRENAL trial show that steroid reduces the time on vasopressors, the duration of intubation, and the length of ICU stay (29347874, 29761216). These are important outcomes which may hasten recovery, avoid iatrogenic harms, and reduce costs.
- A potential mortality benefit of steroid is debated. It's possible that one may exist, if steroid is initiated early and in the sickest patients (29490185). Logistically, it is nearly impossible to prove whether such a mortality benefit exists, so this may be the wrong question to fixate upon.
rationale for early steroid administration
- (1) For years, it was believed that a subset of patients with pressor-refractory shock would experience a mortality benefit from steroid, based on subgroup analysis within the Annane 2002 trial. This suggested that only the sickest subset of patients would benefit from steroid. However, the more recent ADRENAL trial showed that patients on any dose of vasopressor could benefit (as discussed above). This suggests that steroid therapy may be used more broadly.
- (2) The entire concept of sepsis resuscitation is to proactively support the patient and prevent further deterioration – not to wait for the patient to deteriorate further before trying to salvage them in a reactive fashion. This is why every single therapy in sepsis care is instituted rapidly – ideally including steroid.
- (3) Starting steroid immediately eliminates the decision about when to start steroids, shifting greater focus on other issues.
- (4) Immediately initiating maximal medical therapy can help rapidly clarify whether the patient can respond to medical therapy. This may expedite decisions about whether to move to a more aggressive surgical procedure for source control.
- (5) Antibiotics often cause bacterial cell lysis, which releases bacterial products into the bloodstream, causing clinical deterioration (Jarisch-Herxheimer reaction). Front-loaded steroid therapy could theoretically blunt this phenomenon.
- Above is one potential approach to steroid in septic shock. This strategy prioritizes liberation from pressors and ventilation over the risk of hyperglycemia.
- Please note that patients with known adrenal insufficiency or chronic steroid use should definitely be treated with stress-dose steroid.
steroid dose in septic shock
- The best studied dose of steroid in septic shock is 200 mg hydrocortisone total daily.
- Clinical studies often use a continuous infusion of hydrocortisone. However, in clinical practice the use of divided IV doses (50 mg IV q6hr) has the following advantages:
- (1) Immediately establishes effective drug level (rather than an infusion that takes time to reach steady state).
- (2) Doesn't tie up an intravenous line, making it easier to administer.
- (3) One small RCT demonstrated that intermittent administration of hydrocortisone resulted in improved shock resolution compared to a continuous infusion (30628950).
- If hydrocortisone isn't immediately available, any equivalent dose of steroid may be used. A nice option is methylprednisolone 40 mg IV daily, because this is widely available in emergency departments.
fludrocortisone is unnecessary
- The APROCHSS trial used a combination of hydrocortisone and fludrocortisone. This has led some to question the value of adding fludrocortisone on top of hydrocortisone.
- Using fludrocortisone currently doesn't seem justified for several reasons:
- (1) Hydrocortisone itself has mineralocorticoid effects, making the addition of fludrocortisone unnecessary.
- (2) Fludrocortisone administration to critically ill patients often fails to achieve a measurable serum drug level – so it's probably not doing much (27416887).
- (3) The ADRENAL trial demonstrated numerous benefits of steroid – without fludrocortisone administration.
- (4) Septic patients are typically treated with an excessive quantity of exogenous sodium, making fludrocortisone-induced sodium retention unnecessary.
- The Bottom Line reviews:
- Steroid in septic shock: Four misconceptions and one truth
- ADRENAL trial & implications for metabolic resuscitation (PulmCrit)
- APROCCHSS vs. ADRENAL: Really discordant? (PulmCrit)
DVT prophylaxis with heparin
heparin may protect the endothelial glycocalyx
- The endothelial glycocalyx is composed of heparinoids, which have a similar molecular structure compared to exogenous heparin. This raises the possibility that exogenous heparin could potentially inhibit the activity of endogenous heparinase enzymes that degrade the endothelial glycocalyx (thereby protecting the glycocalyx). Some laboratory studies suggest that low molecular-weight heparin may protect the endothelial glycocalyx (28347755, 30046671, 22507823, 22310127). Benefits from heparin may extend beyond this, to include immunomodulation as well (28832958).
- Patients with septic shock are at high risk of deep vein thrombosis (DVT), so they merit treatment for DVT regardless of the above data. A potential benefit of heparin in septic shock argues for initiating this therapy sooner rather than later.
interventions of no real value
serial fluid bolus after initial resuscitation
- This is explored briefly above in the section on fluid management.
- Further discussion regarding why fluid boluses are unlikely to help patients in septic shock is here. This intervention has neither diagnostic nor therapeutic merit.
central venous pressure (CVP) transduction
- Central venous pressure is an extremely complex variable, which reflects a nexus of volume status, cardiac function, and vascular tone.
- The concept that central venous pressure reflects volume status has been thoroughly debunked (18628220).
- Measuring the central venous pressure should be avoided, as this value will almost invariably be misinterpreted.
mixed venous oxygen saturation (%svcO2)
- Similar to central venous pressure, the mixed venous oxygen saturation is a highly complex variable which reflects numerous contributing factors.
- Based on the number of factors affecting svcO2%, there is a dramatic error range in the value.
- Measurement of mixed venous oxygen saturation should be avoided because it is extremely complex, subject to considerable random error, and easily misinterpreted. Furthermore, numerous RCTs revealed that measuring svcO2% didn't improve outcomes compared to conventional therapy (the PROCESS, PROMISE, and ARISE trials).
resuscitative endpoints: general philosophy
Resuscitative endpoints are targets which we would ideally like our patients to reach. The evidentiary basis for most resuscitative endpoints can be summarized roughly as follows:
- There is usually excellent retrospective evidence that a resuscitative endpoint correlates with improved outcomes.
- There is usually minimal prospective RCT evidence that tailoring resuscitation to intentionally reach a specific endpoint improves outcomes.
Prospective RCTs tend to compare different resuscitative endpoints, which is extremely murky (because none of these endpoints is supported by any hard data). These studies often build upon one another, with an exquisitely shaky foundation. For example:
- The Rivers trial used resuscitative endpoints including central venous pressure (CVP) and mixed venous oxygen saturation. As discussed above, this is a profoundly flawed trial which at this point has been largely disproven (discussed above).
- The Jones trial showed equivalence between mixed venous oxygen saturation and lactate. This is probably demonstrating the equivalence of two awful resuscitative endpoints.
- The ANDROMEDA-SHOCK trial suggested that capillary refill could be superior to using lactate as a resuscitation target. Given the lack of evidence supporting lactate as an endpoint, this trial is nearly impossible to interpret. It's possible that this trial reveals more about the harm of chasing lactate than the benefit of chasing capillary refill.
how to use resuscitative endpoints wisely?
- As discussed above, the science behind resuscitative endpoints is very weak. This creates a risk of harm if we chase these endpoints too aggressively.
- (1) Using resuscitative endpoints as a trigger for fluid administration is generally not a good idea.
- It's well established that the vast majority of fluid administered will rapidly extravasate out of the vasculature.
- Unless there is a source of ongoing fluid loss (e.g. diarrhea or high-output fistula), ongoing fluid resuscitation is unlikely to be beneficial.
- More discussion on fluid resuscitation above.
- (2) Resuscitative endpoints may be most useful for titrating vasopressors and inotropes.
- Exactly how patients will respond to vasopressors or inotropes may be impossible to predict from an initial echocardiogram (which represents a snapshot of cardiac function in time).
- The optimal hemodynamic targets (e.g. MAP goal) may vary between patients.
- The best way to determine the optimal MAP goal and vasopressor dose for any specific patient may be empiric titration with close observation of the effects.
- (3) Failure to meet resuscitative endpoints should prompt overall re-evaluation of the patient, for example:
- Is there a failure to detect the correct source of sepsis?
- Is the antibiotic selection correct?
resus targets: MAP
why MAP is generally the most useful blood pressure parameter:
- It is the mean pressure driving perfusion.
- It is what noninvasive oscillometric blood pressure cuffs actually measure.
- It is the most reproducible parameter when measured in different locations and via different techniques (e.g. noninvasive vs. invasive monitoring).
- Septic patients with vasodilation often have low diastolic blood pressure – so an adequate systolic blood pressure may be falsely reassuring.
conventional MAP target (>65 mm)
- MAP > 65mm is the usual target.
- This is generally a reasonable place to start – particularly in an undifferentiated patient whose baseline hemodynamics are unclear.
higher MAP targets (e.g. >75-80 mm)
- SEPSISPAM trial showed that a MAP goal of >80-85 mm improved renal function in the subset of patients with chronic hypertension, but higher MAP targets were associated with increased risk of atrial fibrillation.
- Higher MAP targets may be trialed in patients with chronic hypertension and sluggish urine output. If there is an improvement in urine output at higher MAP, then this strategy may be continued; otherwise, the MAP target may be decreased (vasopressor challenge).
- Patients with cirrhosis and renal failure who have a component of hepatorenal syndrome might benefit from higher MAP targets.
lower MAP targets (e.g. >60 mm)
- Situations where lower MAP targets may be sensible:
- Patients with chronic hypotension (e.g. younger women, patients with cirrhosis without hepatorenal syndrome).
- Patients with end-stage renal failure on hemodialysis (main concern with low MAP is renal injury, which isn't an issue here).
- Lower MAP values may be tolerated in the face of adequate perfusion (e.g. urine output and peripheral perfusion).
resus targets: heart rate
optimal heart rate?
- Septic patients should have a bit of tachycardia. This is a compensatory response that improves cardiac output.
- Remember: Cardiac Output = (Heart Rate)(Stroke Volume)
- The optimal heart rate for a patient with septic shock is unknown, and may vary between patients (e.g. patients with diastolic dysfunction could do better with a slightly slower heart rate). In general, a heart rate of ~90-110 might be reasonable for most patients.
management of bradycardia or “inappropriate normocardia” (e.g. <80 b/m) with systemic hypoperfusion
- If systemic perfusion is inadequate, a normal heart rate is probably not ideal.
- Epinephrine may be trialed in this situation, to determine if it could improve the heart rate and cardiac output (see above: epinephrine challenge).
- For patients with permanent pacemakers, pacemaker rate may be increased.
management of excessive tachycardia (e.g. heart rate >140 b/m)
- Excessive tachycardia probably isn't great either:
- May impair ventricular filling (especially with diastolic dysfunction).
- Over time, may increase the risk of stress cardiomyopathy.
- Potential interventions which may be considered:
- Transition to vasopressors with less beta-adrenergic stimulation (e.g. vasopressin or phenylephrine).
- Digoxin initiation in patients with chronic atrial fibrillation.
- Removal of other stimuli which may be driving tachycardia (e.g. under-treated agitation, pain, or withdrawal).
- Esmolol infusion was found to be beneficial in one study, but generalization from that study is impossible (because patients were on high-dose pressors and often levosemindan)(24108526). It's possible that the benefit from esmolol reflected excessive beta-adrenergic administration. Esmolol infusion in septic shock generally isn't generally recommended.
resus targets: urine output
typical urine output
- The usual goal is >0.3-0.5 cc/kg/hr ideal body weight.
- There is no single, specific cutoff rate for the urine output. The risk of renal failure increases based on how low the urine output is, and how long oliguria persists. For example:
- Urine output 0.4 cc/kg for an hour with subsequent improvement is probably fine.
- Urine output <0.3 cc/kg for several hours suggests an increased risk of renal failure.
good urine output is generally reassuring
- Some factors may falsely elevate urine output: diuretics, severe hyperglycemia (osmotic diuresis), or hypothermia.
- In the absence of these factors, adequate urine output is strong evidence of adequate perfusion.
low urine output is nonspecific
- Rarely oliguria or anuria may be due to urinary obstruction (e.g. dysfunctional Foley catheter).
- Oliguria may represent acute-onset hypoperfusion (“pre-renal AKI”), which will respond to improved perfusion. However, oliguria often represents acute tubular necrosis (“intrinsic AKI”), which won't respond to hemodynamic manipulation. In this scenario, trying to increase the urine output will be futile and potentially harmful (especially if fluid is used).
potential approach to oliguria
- After the initial resuscitation, additional fluid boluses should be avoided unless there is compelling evidence of hypovolemia (e.g. fluid loss from diarrhea).
- Depending on the current hemodynamics, vasopressor or inotrope challenges may be useful. If there is evidence of systemic congestion or a high suspicion for intrinsic renal failure, a furosemide stress test may be performed.
- This is discussed further in the chapter on acute kidney injury.
resus targets: fluid balance
- Keep track of the net fluid balance of the patient, which includes the volume of crystalloid given, as well as the volume of antibiotics and other infusions.
- Very high fluid balance correlates with increased mortality (28130687). Unless there is strong evidence of pre-existing hypovolemia, avoid running the fluid balance above ~5 liters positive.
resus targets: lactate
- Lactate is not an indicator of perfusion, systemic oxygenation, organ failure, or anaerobic metabolism. In most cases it serves as an index of endogenous epinephrine production.
- Cycling the lactate and trying to “normalize” it isn't evidence-based (for more on this, see Myth #3 here).
- Following the lactate is reasonable until it falls, but this data should be used in an intelligent and thoughtful fashion.
following lactate in patients not on epinephrine
- If the lactate is continuing to increase, this may be a sign of “missed injury”
- Undrained source of infection.
- Inappropriate antibiotic selection.
- Incorrect diagnosis entirely.
- A rising lactate isn't an indication to blindly give fluid or inotropes, but rather should be a sign that something is wrong and the entire patient needs to be reconsidered carefully.
*Breezing into the adult ICU for my first day of service*
Lab tech: Oh, um hello
Me: Whatcha doin
LT: A reflex lactate
LT: I'll just go ahead and cancel it
LT: I'll cancel all of the reflex lactates
Me: <whispers> no more lactates forever
— Omnintensivist (@GoodishIntent) April 4, 2019
following lactate in patient on epinephrine
- Epinephrine should cause the lactate to increase. In fact, rising lactate following initiation of epinephrine is generally a positive prognostic sign! (20016405).
- However, if the lactate increases to very high levels (e.g. >10 mM), then the epinephrine infusion should be down-titrated (and replaced by dobutamine if an inotropic agent remains necessary).
- Garcia-Alvarez M, Marik P, Bellomo R. Sepsis-associated hyperlactatemia. Critical Care 2014; 18(5) 503 (open access).
- Understanding lactate and using it to our advantage (PulmCrit)
resus targets: skin perfusion
(1) loss of pulse oximetry waveform
- If the extremities are very poorly perfused, then the pulse oximeter becomes unable to recognize any pulsatile blood flow in the digits.
- This isn't sensitive for poor perfusion (because pulse oximeters are designed to amplify small pulsations).
- When encountered, this should be recognized as an objective sign of terrible perfusion.
- Mottling refers to extreme vasoconstriction of cutaneous blood vessels, which creates a patchwork appearance (skin areas with very poor perfusion will appear violaceous).
- Mottling has been linked to poor prognosis (including mortality) in several studies. Patients with mottling are extremely sick and should be taken very seriously.
(3) capillary refill time
Capillary Refill Time. What a useful hemodynamic monitoring tool when done in a standardised way. Especially in resource limited setting. Even in dark-skinned individual. pic.twitter.com/QeqqpwcvFH
— Dr Supradip Ghosh (@dr_supradip) May 25, 2019
- Preferred technique:
- Compress fingertip using a slide with enough pressure to cause blanching for ten seconds. (Any clear surface can also be used, such as a plastic urine specimen container.)
- Release and measure the time until return of normal color using a chronograph (watch or smartphone).
- Normal capillary refill time is <3.5 seconds, whereas >5 seconds is delayed capillary refill (27908340).
- In the ANDROMEDA-SHOCK trial, patients were resuscitated with a goal of decreasing the capillary refill time below <3 seconds. Further validation is needed before capillary refill could be used as a primary driver of resuscitation, but it's certainly something worth paying attention to.
(4) skin temperature
- Warmth of extremities may be used as an indirect measurement of cardiac output.
- Septic shock is often initially associated with a hyperdynamic state, which is associated with warm extremities and elevated cardiac output (“warm shock”). Some patients may subsequently transition to a state of reduced cardiac output (e.g. due to septic cardiomyopathy). One sign of this transition may be the development of cold extremities (“cold shock”).
multimodal assessment of skin perfusion
- The above signs of skin perfusion generally trend together (especially temperature and capillary refill). It may be most accurate to assess them and consider them all as a group:
- If all signs are consistent with one another, then this is more likely to be accurate.
- If different signs disagree with one another, then this data may be unreliable.
skin perfusion may be used to guide the selection and titration of vasopressors
- (1) Signs of skin hypoperfusion may reflect excessive vasoconstriction (above). In some patients, these will improve with a shift towards use of an inotrope.
- (2) Vasopressin
- One of the most feared complications of using vasopressin is digital ischemia, which may lead to necrosis and digit loss.
- Cool digits may be a sign of excess vasoconstriction and the need to down-titrate vasopressin.
- Digital ischemia (e.g. blue fingers) should be an indication to immediately stop vasopressin.
The above interventions will suffice to stabilize the vast majority of patients. However, rare patients may remain in refractory shock. Below is a list of potential interventions for such patients.
- Metabolic resuscitation if not already initiated (see above).
- Placement of a central arterial line (femoral or axillary)
- Radial arterial lines may under-estimate blood pressure.
- Transition to more hemodynamically stable analgo-sedative regimen
- Propofol or dexmedetomidine are excellent sedatives, but they do tend to reduce the blood pressure.
- In the face of refractory shock, more hemodynamically stable agents may be preferable (e.g. ketamine infusion).
- Trial of various different pressor agents (e.g. epinephrine challenge, addition of vasopressin).
- For patients with right ventricular dysfunction and hypoxemia on mechanical ventilation, inhaled pulmonary vasodilators could be trialed (e.g. inhaled epoprostanol).
- Intravenous calcium (often causes transient improvements in blood pressure which are short-lived).
- Reducing MAP target
- If very high doses of norepinephrine are required to achieve a MAP >65 mm (e.g. >1 ug/kg/min), this may be causing more harm than good. Reducing the MAP target to >55 or >60 could potentially represent a more beneficial balance of vasopressor benefit vs harm.
- Nitric oxide inhibitors
- Avoid these if possible (in one RCT, inhibition of nitric oxide increased mortality)(14707556). Should be used only as a last-ditch effort for profound vasodilatory shock.
- Methylene blue:
- Dose: bolus 1-2 mg/kg q4-6 hours; infusion of 0.25-1 mg/kg/hr.
- Can increase pulmonary vascular resistance
- Dose: 5 grams IV x1.
- Angiotensin II may be considered if your hospital has this.
Part I: Initial resuscitation (Primary Survey)
- Investigations [more on source evaluation]
- Electrolytes, CBC with differential, Coags, Liver function tests
- UA/UCx, peripheral blood culture x2, culture of any line in place >48 hours
- Chest X-ray
- Procalcitonin, Lactate
- Exam with ultrasonography
- CT abdomen/pelvis if source unclear
- Additional tests as warranted (e.g. CT head/LP if concern for meningitis)
- Antibiotics [more on antibiotics]
- Review prior cultures & antibiotic exposure data if available
- Key is a good beta-lactam backbone (piperacillin-tazobactam, meropenem, or cefepime).
- MRSA coverage only if soft tissue infection, infected line, nosocomial infection, or possibly for PNA (not for community-acquired abdominal or urinary source).
- Additional antibiotics depending on source (e.g. azithromycin if pneumonia).
- Source control [more on source control]
- Consider hardware removal (e.g. port, tunneled line, central line).
- Otherwise depends on source (e.g., decompress hydronephrosis, ERCP for cholangitis, abscess drainage).
- Adjunctive therapies
Part II: Follow-up after initial resuscitation (Secondary Survey)
- Correct diagnosis? Right antibiotics & source control measures?
- Review medications
- Have antibiotics been administered? Are they scheduled & dosed optimally?
- Resus targets: Overall philosophy [more]
- It's generally unclear how hard to chase resuscitative targets.
- Additional fluid should usually not be given based on these targets. Resuscitation targets might be best utilized to fine-tune the dose and choice of vasopressors & inotropes (figure below).
- Resus targets: MAP [more]
- Usually target >65 mm.
- Vasopressor challenge: For patients with chronic HTN and poor urine output, increase MAP target to >80mm and determine whether this improves urine output.
- Patients with chronic hypotension and excellent urine output: may target MAP >60 mm
- Resus targets: Heart rate [more]
- Hypoperfusion & heart rate <80 b/m: Consider epinephrine challenge.
- Marked tachycardia (heart rate >140 b/m): Consider vasopressors with less beta-agonist activity (e.g. phenylephrine, vasopressin)
- Resus targets: Fluid balance [more]
- Keep track of net fluid balance (inputs – outputs).
- Avoid running net fluid balance greater than +5 liters (unless history of severe volume depletion prior to admission).
- Resus targets: Skin perfusion [more]
- Evaluate for mottling, cool extremities, and sluggish capillary refill.
- Poor skin perfusion may be an indication that the patient is receiving excess vasoconstrictors (and perhaps could benefit from an epinephrine challenge).
- Resus targets: Lactate [more]
- The use of lactate as a resuscitative target is extremely dubious, but this will conventionally be measured.
- Rising lactate should prompt global re-evaluation of the patient (e.g. echocardiography & adequacy of source control and antibiotics).
- Epinephrine infusions will increase lactate, making lactate measurements meaningless in such patients.
- Resus targets: Urine output [more]
- Falling urine output is concerning regarding hypoperfusion. However, this is nonspecific (may reflect hypoperfusion or intrinsic renal failure).
- Oliguria requires thoughtful evaluation, ideally including echocardiogram and bladder ultrasound.
Most Important Tables & Figures
beware of sepsis mimics:
meticulous search for the source:
Follow us on iTunes
The Podcast Episode
Want to Download the Episode?
Right Click Here and Choose Save-As
questions & discussion
To keep this page small and fast, questions & discussion about this post can be found on another page here.
- Inadequate focus on antibiotic selection and source control. Failure of either of these is most likely to cause morbidity or mortality.
- Blind administration of 30 cc/kg fluid is often not the best treatment. For morbidly obese patients, 30 cc/kg should be be dosed based on ideal body weight, not actual weight.
- Delaying the initiation of vasopressors while waiting to see if fluid-loading works. If the patient is really in septic shock, delaying hemodynamic stabilization isn't advisable.
- Incorrect diagnosis of sepsis due to pneumonia (in response to minor CXR abnormalities from atelectasis) or urosepsis (due to UA abnormalities reflecting asymptomatic bactiuria). CXR and UA are often abnormal, so be careful of premature diagnostic closure based solely on these tests.
- Inadequate diagnostic investigation (e.g., for patients with an unknown or unclear source of sepsis, get a CT abdomen/pelvis).
- Excessive bombardment of all septic patients with vancomycin. In particular, patients with urosepsis or community-acquired abdominal sepsis won't benefit from vancomycin – all vancomycin will achieve is nephrotoxicity.
- Antibiotics should always be ordered first dose now, STAT. Beware of electronic medical record systems which will auto-schedule antibiotics to start hours in the future!