Early Goal-Directed Therapy:  A house collapsing in slow motion

The original foundation of the Surviving Sepsis Campaign was the Rivers trial on early goal-directed therapy.  This is basically the NINDS trial of the critical care world:  a study with ~300 patients showing implausibly positive results, published in NEJM, and rapidly brainwashing an entire discipline.  The NINDS and Rivers trials became dogma not because they were rigorous, but rather because people wanted to believe them.  After struggling for decades to treat stroke and sepsis, these finally represented some hope.

The unfortunate reality is that the Rivers trial was a flimsy study.  It was a single-center study with dodgy methodology, including mysterious disappearance of 25 patients who were randomized but never analyzed.  Dr. Rivers had major conflicts of interest, including patenting a catheter to monitor svcO2.

The Rivers trial and the Surviving Sepsis Campaign popularized sepsis protocols, which saved lives.  Massive accomplishment.  However, that doesn't validate the individual components of early goal-directed therapy.  Any protocol involving early identification and intensive care of septic patients probably would have worked.

Over time, nearly every component of early goal-directed therapy was disproven.  Transfusing patients to a hemoglobin of 10 mg/dL is now recognized to be foolish.  Central venous pressure predicts fluid responsiveness about as well as flipping a coin (Marik 2008).  Monitoring the svcO2 saturation has been proven to be unnecessary.  Massive fluid resuscitation encouraged by the protocol is increasingly recognized to be harmful (in the initial Rivers study, protocoled patients received an average of 13.4 liters!).  The only core component that has survived the test of time is maintaining a mean arterial pressure above 65mm.

Although the components of the Rivers protocol have been disproven, it remains blasphemous to admit that the Rivers trial was wrong.  The Rivers trial is the underpinning of a decade of research in sepsis.  If the Rivers trial isn't valid, then many studies based on it aren't valid either.  Admitting Rivers is wrong is like admitting that the foundation of our house has rotted and is about to collapse.

The Surviving Sepsis Campaign was founded on the basis of early goal-directed therapy.  As this was disproven, the campaign stubbornly clung to it (for example, continuing to recommend CVP monitoring in 2014).  The current guidelines have finally removed early goal-directed therapy, since to do otherwise would have rendered them irredeemably obsolete.

However, many vestiges of early goal-directed therapy remain in the current guidelines (e.g. serial lactate measurement).  We must take a hard, objective look at such therapies to determine if they still stand up, now that early goal-directed therapy has collapsed.

Myth #6: Combination therapy is beneficial for gram-negative septic shock.

Combination therapy is defined as using two drugs against gram-negative pathogens, with a goal of synergy (box above).  This has been proven to be ineffective in numerous RCTs and meta-analyses (example below).  It remains controversial whether to initially provide double-coverage for gram-negative pathogens.  However, even advocates of double-coverage generally admit that combination therapy doesn't work (the main argument for double-coverage is to increase the likelihood that at least one drug covers the pathogen).

The Surviving Sepsis Campaign has been on the fence about this issue since its inception.  The guidelines are slowly inching away from combination therapy.  For example, the recommended duration of combination therapy has shrunk from 3-5 days (2012 guidelines) to “the first few days” (2016 guidelines).  The indications for combination therapy have been pared back to only the sickest patients:

The guidelines attempt to draw a sharp distinction between septic shock and sepsis.  This isn't possible, because definitions have shifted over time and many studies of antibiotics contain a mixture of both conditions.  Regardless, there is no high-quality evidence to support combination therapy for gram-negatives in either sepsis or septic shock (1).

Myth #5: Norepinephrine is the best vasopressor for all septic patients.

Norepinephrine is a great drug for septic shock.  Everyone likes norepinephrine.  However, this consensus is threatening to create a dogma that norepinephrine is the best drug for every patient.

Norepinephrine is clearly superior to dopamine.  Beyond this, the evidence is unclear:

• One prospective RCT and a subsequent meta-analysis show equivalent outcomes comparing epinephrine vs. norepinephrine (Myburgh 2008, Avni 2015).
• Recent publication of the VANISH and VANCS trials support the use of vasopressin as a front-line vasopressor in patients with sepsis.

Overall, norepinephrine, epinephrine, and vasopressin are all supported by evidence as potential first-line vasopressors.  There is no proof that any is superior to any other.  It's possible that the best drug might depend on an individual's physiology:

Myth #4:  If one vasopressor doesn't work, additional drugs should be added in a sequential fashion.

The guidelines recommend the strategy above for titrating vasopressors.  This algorithm ignores individual differences in hemodynamics and vasopressor responsiveness.  For example, rapidly up- and down-titrating vasopressors may reveal how a patient responds to different drugs.  This is safe and easy: make a change, observe, if it doesn't help then return to your prior regimen.  Vasopressor challenge may be useful to select agents that work, while avoiding agents that aren't helping.

One problem with the strategy recommended in the guidelines is that patients often wind up on high-dose norepinephrine plus vasopressin.  This combination provides a lot of afterload, which may drop cardiac output in patients with cardiomyopathy (yes, that's right: vasopressor-induced shock).  Such patients often respond nicely to up-titration of epinephrine with simultaneous down-titration of norepinephrine.

Myth #3: Lactate is a measure of tissue perfusion. Normalization of lactate should be used as a resuscitation target.

It remains widely believed that lactate is a measurement of perfusion, systemic oxygenation, and anaerobic metabolism.  This is a myth.  Among septic patients, the primary cause of lactate elevation is beta-2 agonist stimulation (from endogenous and exogenous epinephrine).  Beta-2 agonist stimulation causes the liver to secrete lactate via an aerobic mechanism.  Lactate may be used by the heart and brain as a metabolic fuel, suggesting that this is an adaptive mechanism designed to handle physiologic stress.

Once we understand lactate as an index of endogenous epinephrine, using it as a resuscitation target stops making sense:

• First, let's imagine a young woman who is doing great, after receiving some fluids and norepinephrine.  Her blood pressure, urine output, and skin perfusion are excellent.  Her lactate is 6 mM, revealing the presence of endogenous epinephrine.  Giving her additional norepinephrine and dobutamine can overdrive suppress her sympathetic nervous system, shutting down endogenous epinephrine production.  This will reduce lactate, but it probably won't help her.  It's basically using exogenous norepinephrine and dobutamine to replace her endogenous epinephrine (which was working perfectly well).
• Second, let's imagine an elderly man who is doing poorly despite receiving fluids and norepinephrine.  His heart rate is in the 70s, his urine output is poor, and his extremities are mottled.  His lactate is 0.5 mM.  This patient is suffering from an inadequate sympathetic response, with a deficiency of endogenous epinephrine.  After an epinephrine infusion is started, he improves dramatically with excellent urine output and skin perfusion.  However, the exogenous epinephrine causes his lactate to increase to 6 mM.  This increase in lactate is actually a positive prognostic sign, indicating that he is likely to improve (Wutrich 2010).  Rising lactate doesn't mean that he's getting sicker – it means that the epinephrine is working (2).

There is no persuasive evidence that trying to “normalize” lactate is beneficial.  The new guidelines recommend using lactate to guide resuscitation on the basis of five studies:

• Jones 2010:  This showed that chasing lactate was equivalent to chasing svcO2.  However, recent studies suggest that chasing svcO2 isn't helpful… so chasing lactate and chasing svcO2 may be equally ineffective.
• Jansen 2010:  This prospective RCT found no mortality benefit among patients whose lactate levels were measured. However, a post-hoc adjusted analysis did find benefit.
• The remaining three studies aren't available in English (Lyu 2015, Tian 2012, Yu 2013).

These five studies have been combined into a meta-analysis, showing that targeting lactate clearance was beneficial:

However, these studies are horrifically heterogeneous.  For example, Tian 2012 used lactate clearance >10% in the control group, whereas Yu 2013 used this same goal in the experimental group (table below, red text).  It's simply not valid to combine such heterogeneous studies into a meta-analysis.

Ultimately, the available evidence is limited (3).  However, it is doubtful that trying to “normalize” the lactate is beneficial.  Septic patients probably aren't supposed to have a “normal” lactate (an example of inappropriate euboxia).

Myth #2: All septic patients must receive 30 ml/kg fluid initially.

I recently cared for a patient nearly killed by this concept.  A morbidly obese young woman presented to an outside hospital with influenza pneumonia.  She met qSOFA criteria due to tachypnea and a blood pressure of 95/60 (her baseline).  To comply with CMS regulations, the hospital recently implemented a sepsis order set including 30 cc/kg fluid.  The trigger was pulled.  Following a single keystroke, she received nearly four liters of saline (dosed per actual weight) plus several IV bags of antibiotics.  That pushed her into full-blown ARDS, precipitating a one-week stint on the ventilator while we diuresed her.

Let's be honest:  nobody really knows the right amount of fluid to give to a patient with sepsis.  30 cc/kg is a reasonable dose for most patients.  However, the concept that every patient must receive 30 cc/kg fluid is nuts.  For example, septic shock does sometimes coexist with hypervolemia in patients with CHF or pulmonary hypertension (an example of Hickam's Dictum, above).

The above algorithm may be frankly dangerous.  For example, consider the recommendation that a patient with ARDS and either hypotension or hyperlactatemia may be intubated to facilitate fluid administration (4).  This is dubious for many reasons:

• We must always consider the entire patient to answer the key question: what physiologic process represents this patient's greatest life-threat?  For many patients with ARDS and mild hypotension/hyperlactatemia, their primary life-threat is ARDS.  In that case, the risk of large-volume resuscitation outweighs the possible benefit.
• Intubation, sedation, and positive-pressure ventilation typically cause patients to become more hemodynamically unstable.  In many cases, the amount of hemodynamic instability caused by intubation and sedation will outweigh any hemodynamic benefit from 30 cc/kg fluid.
• There is no evidence to support the practice of pre-emptive intubation to facilitate fluid loading in the context of ARDS.  If anything, evidence might suggest the opposite:  diuresis to facilitate extubation (FACTT trial).

Myth #1: It's helpful to mandate that specific sepsis therapies be given within a rigid time frame.

It is widely believed that early antibiotics are essential, with each hour being critical.  This is based on retrospective studies which correlate the delay to antibiotics with outcome.  However, delay to antibiotics also correlates with numerous confounders (atypical clinical presentation, mental status alteration, day of the week, time of day, delay to other treatments, etc.).  These correlational studies should be used only for hypothesis generation.

To make matters even muddier, studies disagree.  A meta-analysis detected no benefit from early antibiotics (Sterling 2015).  Ultimately, there is no scientific answer to this question (5).

However, let's suppose that earlier antibiotics do help (probably true).  The next question is whether it is beneficial for guidelines to mandate that antibiotics be given within a certain time-frame.  This may accelerate care, but it may also force practitioners to cut corners.  On the balance, is this helpful?  What is better:  fast and sloppy care, or slow and accurate care?  This, too, is unknown.

A cautionary tale is provided by the case of antibiotics in pneumonia.  Retrospective studies in 1997 and 2004 suggested a correlation between reduced mortality and antibiotics administration within four hours.  Subsequently, numerous regulatory bodies adopted this as a quality benchmark.  Over the following years, it was shown that this benchmark increased misdiagnosis and inappropriate antibiotic use.  Finally, the four-hour benchmark was abandoned.

• We have a history of over-estimating our knowledge of sepsis and the benefit of our interventions.
• Acknowledging uncertainty is uncomfortable, but it may ultimately make us more thoughtful and flexible clinicians.
• We need to be honest with ourselves about what is actually known about sepsis therapy (which is surprisingly little).   For example:
  • Intentional double-coverage of gram-negative bacilli to achieve synergy has no proven benefit.
  • Norepinephrine is a good vasopressor, but it is unclear whether it is the best drug for every patient.  The ideal drug selection and dose is unknown.
  • Lactic acid is not a reliable measurement of tissue perfusion.  It is doubtful that using lactate as a resuscitation endpoint is beneficial.
  • Nobody knows how much fluid should be given to septic patients or how quickly it should be given.
  • The benefit of early therapy is unproven.  Insufficient evidence exists to make any concrete recommendation regarding the timing of interventions.

More information

• Surviving Sepsis 2016
  • Full guidelines (PDF)
  • Users’ Guide to the Guidelines (PDF)
  • Podcasts from EMCrit and FOAMcast
  • Blog from PulmCCM (Jon-Emile Kenny)
• Myth #6, Combination therapy for gram-negatives
  • The siren's call: Double coverage for VAP (PulmCrit)
  • Double-coverage of gram negatives with a fluoroquinolone? (PulmCrit)
• Myth #5, NE best vasopressor for everyone
  • Renal microvascular hemodynamics & vasopressin (PulmCrit)
  • Renoresuscitation, vasopressin, vepinephrine & VANISH (PulmCrit)
  • Phenylephrine infusions: not pure evil? (PulmCrit)
• Myth #4, Blind sequential vasopressor addition
  • Epinephrine challenge in sepsis (PulmCrit)
• Myth #3, Lactate abuse
  • Understanding lactate and using it to our advantage (PulmCrit)
  • Understanding lactate: SMACC talk (Marik)
  • SMACC-Back on Marik & Lactate (EMCrit)
  • Euboxia & Ab(Normality) (LITFL, Nickson)
• Myth #2, 30 ml/kg fluid for everyone
  • Marik Fluid Lecture (Marik on EMCrit)
  • Renoresuscitation: Sepsis resus to avoid long-term complications (PulmCrit)
  • Mythbusting: Empty IVC plus hyperkinetic heart doesn't equal volume depletion (PulmCrit)
• Myth #1, Timing dogma
  • The fallacy of time-to-intervention studies (PulmCrit)
  • Insanity of CMS Core Measures & Micromanagement (EMCrit)

Notes:

1. Combination therapy is evidence-based for management of severe pneumonia (azithromycin + beta-lactam) and toxic shock syndrome (clindamycin + beta-lactam).  However, the use of combination therapy for gram negatives isn't.
2. I've seen this many times, and it's generally a beautiful phenomenon.   The lactate spikes in response to epinephrine, and often patients improve a lot and come off vasopressors entirely within the next 24 hours.
3. The most meaningful question is whether trending lactate would be beneficial after we stop measuring CVP.  None of these trials answer that question (all were performed in the bygone era of early goal-directed therapy).
4. The disease previously known as ALI (Acute Lung Injury) has now been re-defined as ARDS.
5. It would be nearly impossible to design a RCT to test this, using time as an independent variable.

• Author
• Recent Posts

Social Me

Josh Farkas

Josh is the creator of PulmCrit.org. He is an associate professor of Pulmonary and Critical Care Medicine at the University of Vermont.

Social Me

Latest posts by Josh Farkas (see all)

• PulmCrit: How to quickly create a useful professional account in BlueSky - November 28, 2024
• PulmCrit Wee: Why MedTwitter should move to Bluesky - November 15, 2024
• PulmCrit Wee – A better classification of heart failure (HFxEF-RVxEF) - August 26, 2024