Escalation-deescalation
There are roughly two strategies for adjusting the intensity of treatment:
- Titrated strategy: Treatment intensity is adjusted to match the severity of the disease.
- Escalation-deescalation strategy: Treatment intensity is increased rapidly to exceed disease severity and gain control of the disease. After the patient improves, treatment intensity is reduced.
The best strategy depends on the situation. A titrated strategy is good for an outpatient with cellulitis: start with cefazolin and broaden antibiotics later if necessary. Septic shock from cellulitis is best treated with escalation-deescalation: start with broad-spectrum antibiotics, then deescalate after the patient responds.
The importance of early escalation in septic shock is generally recognized (e.g. early antibiotics, early resuscitation). However, steroid has traditionally been withheld until the patient was failing to respond to other treatments. This runs contrary to basic principles of sepsis management. The reasons for withholding steroid are two misconceptions:
- Steroid therapy has serious side-effects.
- Only the sickest patients benefit from steroid.
Why delaying steroid therapy doesn't make sense.
Misconception #1: Steroid is very risky and increases infection rates.
A few decades ago, using steroid for sepsis did cause lots of problems, because huge doses were used. We now use much lower doses (equivalent to ~50 mg/day prednisone). Short courses of low-dose steroid are well-tolerated. Such doses are often prescribed to outpatients. If primary care docs are comfortable prescribing this to outpatients with zero monitoring, then we shouldn't be scared of it either.
A common concern regarding steroid is an increased risk of infection. Low-dose steroid is unlikely to exacerbate infection in the absence of tuberculosis or fungal infection. Where the incidence of tuberculosis or fungal infection is low, this is safe. Meta-analysis of prior studies of steroid in sepsis revealed no increase in infection rates (Sligl 2009). The ADRENAL trial confirms this.
Steroid does have side-effects. However, these seem to be fairly minor and tolerable. Hyperglycemia occurs, but it's debatable whether this truly causes harm. Delirium can occur. In the ADRENAL trial, the rate of severe treatment-related adverse events was 4/1835 with steroid vs. 2/1829 with placebo (nonsignificant).
My read on #ADRENAL is that they are indeed harmless
— Scott K Aberegg MD MPH (@medevidenceblog) January 20, 2018
The physiology of sepsis could protect patients from steroid side-effects. Sepsis induces a state of glucocorticoid resistance (Dendoncker 2017). Thus, giving stress-dose steroid to a septic patient should cause fewer side-effects than giving the same dose to a healthy person.
Misconception #2: Only the sickest patients benefit from steroid.
Steroid accelerates shock reversal. This effect is observed across all levels of shock severity. For example the CORTICUS trial, which has been criticized for including patients who weren't that sick, found accelerated shock reversal in all patient subgroups (even the least ill subgroups). Within both ADRENAL and CORTICUS, shock reversal was accelerated among patients on any dose of norepinephrine.
It is conceivable that sicker patients may benefit more. Probably so. However, what the evidence shows is that steroid accelerates shock resolution among septic patient on any vasopressor dose.
less time in shock, on a ventilator and in ICU? Will be hard not to use steroids given the size and robustness of #Adrenal #CCR18
— Rob Mac Sweeney (@CritCareReviews) January 19, 2018
Steroid did accelerate recovery in ADRENAL
Some study results may initially appear to contradict each other. The speed of recovery was reported using two types of variables in the manuscript:
- Median time-to-event (e.g. days until extubation)
- Average time free of support (e.g., vent-free days, ICU-free days)
The median time-to-event is significant in every case, but the averages are not:
The reason is simple. An average value is very sensitive to outliers (extreme data points). This causes the standard deviation of the average to be quite high (red boxes above). Large standard deviations make it impossible for differences in the average value to reach statistical significance. That doesn't mean that difference doesn't exist, it just means that this particular statistic isn't powered well enough to detect them (4).
Thus, the average number of support-free days is a noisy variable which can easily obscure clinically meaningful differences. A more sensitive approach to analyzing this data is time-to-event analysis, which was shown in the appendix. Surely enough, highly significant differences are detected:
The overall picture here is simple and clear: steroids accelerate recovery. Shock resolution, extubation, and ICU discharge were all hastened. These differences were highly significant when measured either using the median time-to-event or using hazard ratios.
Using steroid to accelerate recovery is universally accepted.
There is some debate about whether steroid can still be useful, even if it doesn't improve mortality. That's silly. Steroid is widely used to accelerate recovery in many diseases. For example, the use of steroid for COPD exacerbation is universally accepted. RCT evidence shows that steroid doesn't affect mortality in COPD exacerbation, it merely accelerates recovery by one day (Niewoehner 1999). Furthermore, the steroid dose used in sepsis is far lower than steroid doses commonly used for COPD (60-125 mg methylprednisolone q6hr, equivalent to 300-625 mg prednisone/day!). It's illogical to accept the use of steroid for COPD, but not for septic shock.
#Adrenal shows no mortality effect but more rapid shock resolution, reduced ICU LOS, and less transfusion. Unanswered Qs but surely, on the balance of probabilities, the answer is ‘yes' for now?
— Paul Young (@DogICUma) January 19, 2018
Implications for hydrocortisone-ascorbate-thiamine (HAT) therapy
Metabolic sepsis resuscitation with hydrocortisone, ascorbic acid, and thiamine (HAT) remains highly controversial. Until now, the strongest argument against HAT has been concern about side-effects from hydrocortisone (thiamine and ascorbate are water-soluble vitamins with excellent track records for safety)(2). The ADRENAL trial addresses these concerns nicely.
The rationale for using HAT is roughly two-fold:
- HAT might reduce mortality.
- HAT might accelerate recovery.
Whether or not HAT improves mortality is hard to say currently. Numerous RCTs are ongoing, which hopefully will give us a better answer within the next few years.
Regardless, evidence is accumulating that HAT does accelerate recovery from septic shock. Numerous RCTs consistently show that steroid accelerates shock reversal (e.g. CORTICUS and ADRENAL). Likewise, a small RCT found that intravenous ascorbic acid accelerated shock reversal (Zabet 2016). Given that both steroid and ascorbate accelerate recovery when used alone, the combination ought to work quite well (and might even work synergistically)(3). The notorious before-after study by Marik did indeed show that.
Rapid recovery is good for patients and good for the healthcare system. For the patient, this means less time on ventilation, less time in the ICU, and less time with a central line (or avoiding a line entirely)(1). Reducing ICU length of stay is also beneficial for the healthcare system. Hospitals have a limited number of ICU beds. Giving HAT to a patient with mild septic shock won't save that individual's life, but it could save the life of another patient awaiting that ICU bed.
Marik described the use of HAT therapy as immediate, up-front treatment for any patient with septic shock. Nonetheless, most providers using HAT reserve it for a tiny subset of patients with refractory shock who are failing all conventional measures. The ADRENAL trial supports steroid administration for patients with all severities of shock. Once you're willing to accept the concept of steroid administration, the addition of thiamine and ascorbic acid isn't a very large leap of faith. Thus, the ADRENAL trial moves us closer to using HAT broadly for patients with septic shock.
Evolution of sepsis therapy into a front-loaded strategy
Historically, sepsis therapy would start with antibiotics and fluids (figure below). After giving a lot of fluid (customarily four liters), the need for vasopressor would be grudgingly acknowledged. A couple hours delay would ensue to allow for placement of a central line and X-ray confirmation. Later on, if the pressors weren't working, steroid might be added on. This process of treatment escalation might take half a day.
A better approach to septic shock is shown above. Antibiotic, fluid, and peripheral vasopressor are all started immediately. Additional vasopressors are added within minutes if needed. Metabolic resuscitation with hydrocortisone, ascorbate, and thiamine are started immediately, for anyone on vasopressors. As patients improve, vasopressors and metabolic therapy are weaned off. Rapid escalation stabilizes the patient faster, which overall reduces the ICU length of stay. A less fluid-centric strategy avoids volume overload and “ARDS.”
Critical care medicine exists in a perpetual haze of uncertainty. Currently, RCTs are testing HAT therapy. Arguments about steroid will surely rage on forever. However, the above strategy seems congruent with the best available evidence. In an uncertain world, it is acceptable to strike hard, strike fast, and have no remorse.
- The ADRENAL trial showed that stress-dose steroids are fairly safe, have no mortality benefit, and accelerate clinical improvement.
- ADRENAL shows that steroid offers meaningful patient-centered benefit to patients on any dose of vasopressor (not solely patients on high-dose vasopressor).
- Treatment for septic shock involves immediate initiation of numerous therapies to stabilize the patient as rapidly as possible. It makes sense to initiate steroid promptly, rather than withholding it until the patient is failing to respond to high doses of vasopressor.
- Metabolic sepsis resuscitation with hydrocortisone, ascorbic acid, and thiamine remains controversial. By demonstrating the safety and efficacy of hydrocortisone, the ADRENAL study provides indirect support for this combination.
Related
- Fore-runners of this post:
- Steroids in sepsis 2015 (PulmCrit) – this post rounded up available evidence and correctly predicted the ADRENAL results. It remains valid.
- Accelerated Goal Directed Therapy for sepsis 2015 (PulmCrit)
- Metabolic Sepsis Resuscitation (PulmCrit) – evidence behind all the vitamin C stuff.
- ADRENAL trial
- Manuscript at NEJM
- Live presentation of results at CCR meeting
- The Bottom Line review – contains all the details of the study
- The Adrenal Trial – Salim Rezaie (Rebel EM
- CC Nerd (Rory Spiegel) – a skeptical view of the ADRENAL trial.
- Why negative trials are generally positive (BMJ, Peter Brindley)
- EM Lit of Note (Ryan Radecki) discusses power & NNT
- End of the roid? Dan Horner at St Emlyns Blog
Notes
- For patients with very mild sepsis who need only a very low dose of vasopressor, one strategy is to use peripheral vasopressors (e.g. peripheral phenylephrine) combined with HAT. Such patients can often be weaned off vasopressors promptly, thereby never requiring a central line. A typical scenario where this may be useful is an elderly patient who is DNR and refuses placement of a central line.
- It's generally accepted that 200 mg IV thiamine BID for four days is completely safe. IV ascorbic acid can cause oxalate nephropathy or hemolysis among patients with G6PD deficiency, but these events are reported at higher doses and/or longer durations than used in HAT therapy (6 grams daily for up to four days; Colliou 2017).
- The effect of steroid and ascorbic acid on endothelial function has been shown to be synergistic (Marik 2017). From a theoretical standpoint, the benefit from HAT would be expected to be greater than the sum from each ingredient given alone.
- Looking at event-free days remains important to make sure that time-to-event statistics are generated by accelerating recovery (rather than accelerating death). The fact that trends in event-free days match up with median time-to-event data is reassuring in this regard (as is the fact that there was trend towards reduced mortality among patients treated with steroid). Therefore, event-free days are a very appropriate statistic to include in the manuscript. It's simply a matter that this statistic shouldn't be over-interpreted: absence of a significant difference in the average number of event-free days doesn't exclude a clinically meaningful benefit.
- Pulmcrit wee: The cutoff razor - April 15, 2024
- PulmCrit Blogitorial – Use of ECGs for management of (sub)massive PE - March 24, 2024
- PulmCrit Wee: Propofol induced eyelid opening apraxia – the struggle is real - March 20, 2024
Given the title, shouldn’t the photo have been from The Karate Kid?
excellent point!! although I like the visual of fighting in the rain because that captures how I feel sometimes in treating sepsis – I don’t know the exact source, it’s unclear exactly what is going on, I’m just struggling to gain control.
Are there rct‘s already started for HAT?
Yes. Will take a few years but we should have some better information relatively soon. There are a few studies. Hopefully none of them will be stopped early (either for efficacy or futility).
It is funny…. I always thought secondary outcomes (like time on ventilation, …) are only useful to assess the validity of the primary outcome, if it is positive.
But the ADRENAL study showed that mortality isn’t changing (then we can discuss if it was a good primary outcome), so how can someone say : “oh I thing the study was flawed because of this and this and this, but look at Rhodes gorgeous secondary outcomes !”
see discussion above with Swami about primary/secondary endpoint issue. This is tricky and I’m not sure there is a simple answer to it. We don’t get many big RCTs in critical care, so it makes some sense to pay attention to the secondary endpoints.
Josh/Scott, What are your thoughts on deferring steroids in patients with IDDM? The steroid might in theory benefit clearing the infectious process, but may also cause severe hyperglycemia which would in turn inhibit the bodies ability to manage and recover from said infection. Thoughts?
I don’t know, you have to use your judgement based on severity of septic shock vs. how brittle the patient’s diabetes is.
The ADRENAL trial didn’t exclude patients with diabetes. In fact, they didn’t have a lot of exclusion criteria, which is nice. This should theoretically make the results fairly generalizable. Of course, every patient is unique so the results from the entire population may not apply to one specific patient.
Josh – interesting post and I appreciate your comments on ADRENAL and Vit C-Thiamine treatment. My major issue with the argument is that the time to shock reversal was one of many secondary outcomes and, as such, is simply hypothesis generating. A large RCT designed to look specifically at the question of shock reversal as the primary outcome may show no difference . . . or it could support the secondary outcome. I haven’t seen studies using shock reversal as the primary outcome but was wondering if you knew of any in existence already. My reluctance with the whole thing… Read more »
Very interesting issues here. Secondary outcomes and sub-group analyses are often badly abused in efforts to show some sort of benefit. So caution is always in order. However, the secondary/primary endpoint division is a bit artificial. For example, imagine that the ADRENAL trial had been performed with shock reversal as the primary endpoint. Same patients, same study, same data. This would make the study “positive.” But it doesn’t change anything about the actual science or raw data. Steroid isn’t patented, so these trials are generally clinician-designed. Clinicians usually choose mortality as a primary endpoint, leading to “negative” trials. This is… Read more »
Very good thoughts on primary v secondary endpoints (and I like the example of the ATHOS-III, ha). I think we all tend to discount secondary outcomes to a greater proportion than we should. I think this is true for a lot of smaller studies that have contributed to the “metabolic resuscitation” strategy over the last several years, too. For example, some of my favorite work on metabolic (or mitochondrial, more specifically) resuscitation strategies in sepsis have been done by Donnino et al. at BID, such as this RCT of thiamine supplementation in sepsis: https://www.ncbi.nlm.nih.gov/pubmed/26771781 — this was negative for the… Read more »
Josh,
Do you administer IV vitamin C to septic patients in your ICU?
yes. It seems to work – have seen some surprising turn-arounds and overall patients seem to get better really fast.
There will never be a trial with ‘shock reversal’ as the primary endpoint in these major trials (though I don’t think unreasonable to have something other than mortality, esp in ECMO, hypothermia, etc trials – but in this case, why shock reversal, why not ventilator or ICU days or how about primary outcome being 6-12mo QoL/function/etc – which are FAR MORE important IMO for any treatment than if someone lives or dies???). I get the ‘concern’ related to secondary endpoints (but as Josh notes- it doesn’t change a clinically significant findings that reached statistical significance. Do you really think we… Read more »
Hello, I’m passionated with nutrition and I have great hope that we take a metabolic approach on the treatment of (critical care) disease. About the innocuity of treatments, I use to think and teach that every medication is potentially harmfull (i.e. think about oxygen). I’ve been witnessed of a patient with a deterioting neurological state during one night on call. The point of care glucose monitoring was saying that the patient glycemia was ok although it was not ! Ascorbate interferes with these devices ! We have to be aware of this rare but serious issue. https://www.ncbi.nlm.nih.gov/pubmed/25162951
very interesting, Josh. then the question might also be: for very ill patients not yet in “shock”, should one consider HAT? and, what is begun in the ER would best be continued in the ICU. so it might require a paradigm, cultural adjustment at a particular institution. not just one ER doc being a “cowboy”.
thank you, Josh
tom
Hi Josh. Sorry for joining late. Regarding HAT , the problem I have in mind is What is the dose of vitamin c and thiamine and hydrocortisone that makes it work. Suppose you have adult patients weighing 30 kg, another with 60 kg and yet another with a 120 kg. It’s not that rare. So how can I believe that 6g. Vitamin c is enough or not toxic to each patient? When standardised to body weight, it will be 0.2gm/kg,0.1 gm/kgAnd 0.05 gm/kg. Similar with thiamine and hydrocortisone. I never didn’t get why the dose has to be not patient… Read more »
It is beginning to appear that all those prominent figures running around with cries of ‘snake oil’ and ‘charlatan’ levelled squarely at Marik may have yet to swallow those words. I wonder if they will humbly apologise if he is proved to be right. Maybe it really is as simple as we don’t want to believe it to be.
A great post. Thank you.