Sepsis is scurvy?? Vitamin C, thiamine, and steroids

I have avoided commenting on vitamin C for sepsis, because commentary seemed rather unnecessary. The hype was obviously ridiculous. The chances of it helping were clearly low. The evidence of benefit was negligible. Others had widely discussed the papers, and most people seemed to be ignoring this “miracle cure.” However, a new study was just published with what might be the greatest trial name of all time: HYVCTTSSS. With a name like that, I just couldn’t resist. A blog post about HYVCTTSSS is bound to be a viral success. It just roles off your tongue. People are going to be talking about HYVCTTSSS for years. So, because of HYVCTTSSS, I will finally get around to discussing the evidence for vitamin C in sepsis.

The paper that started everything

This is the paper that brought everyone’s attention to vitamin C as a potential treatment for sepsis. Based on some in vitro lab work, the authors decided to give vitamin C and steroids to 3 patients they thought were definitely going to die, and all 3 made a ‘miraculous’ recovery. From that point on, they decided to give all patients in their ICU vitamin C and hydrocortisone (and they also threw in thiamine). They present us with before and after data.

Marik PE, Khangoora V, Rivera R, Hooper MH, Catravas J. Hydrocortisone, Vitamin C, and Thiamine for the Treatment of Severe Sepsis and Septic Shock: A Retrospective Before-After Study. Chest. 2017;151(6):1229‐1238. doi:10.1016/j.chest.2016.11.036 PMID: 27940189

Methods

This is a retrospective chart review, with a before and after design. They started routinely giving vitamin C, hydrocortisone, and thiamine to all adult ICU patients with severe sepsis or septic shock and a procalcitonin over 2 in January 2016. Their protocol provided Vitamin C 1.5 grams IV every 6 hours for 4 days, hydrocortisone 50 mg IV every 6 hours for 7 days, and thiamine 200 mg IV every 12 hours for 4 days. The protocol was stopped if the patient left the ICU. Some patients in the control group also received hydrocortisone at the treating physician’s discretion. They look at the outcomes in the 6 months before and after that protocol change. They don’t comment at all on their chart review methods. Their primary outcome was in-hospital survival. 

The Results

They included 94 patients overall (47 in both groups). The control group was slightly older (mean 62 vs 58 years), had more diabetes and hypertension, and was more likely to be mechanically ventilated (55 vs 47%).

In hospital mortality was drastically lower in the after group (4% vs 20%). They say that none of the patients in the after group died of sepsis, but instead of complications of their underlying disease. There were also some differences in secondary outcomes, such as need for renal replacement therapy and duration of vasopressors. 

My thoughts

The results of this trial are unbelievable. Seriously, the massive difference in mortality is literally not believable. Obviously, we would all love a cure for sepsis. It makes sense that this study was followed up with RCTs, but these results are very unlikely to be true.

This is not randomized data. They try to use propensity matching to compare the two groups, but there are probably hidden confounders that were unaccounted for. It is a single center trial with a very small number of patients. The Hawthorne effect is a possible source of bias. This study was not blinded, and the authors were obviously extremely excited about this protocol (in that they had already adopted it), which dramatically increases the chance of bias. There is a reason that we don’t base major medical decisions on before and after studies. 

The complete lack of chart review methodology is also a huge problem and tremendous source of bias. Agreed upon methods to improve the reliability of chart reviews include training abstractors before starting the study, explicitly defining inclusion and exclusion criteria, precisely defining variables of interest, using a standardized abstraction form, holding periodic meetings to resolve disputes and review coding rules, monitoring the performance of the abstractors, blinding the abstractors to the study hypothesis, and testing inter-rater agreement between multiple abstractors. (Gilbert 1996) This trial included none of them. 

Mortality is a hard, objective outcome. How can bias influence something like mortality? Although bias and subjective outcomes are a worse combination, the use of mortality doesn’t eliminate the influence of bias. We may just select out different populations to compare. For example, if a patient dies of shock despite getting this metabolic cocktail, the physician might decide that the diagnosis couldn’t possibly be sepsis, because he knows this cocktail works, and therefore write cardiogenic shock or some other diagnosis on the death certificate. Ultimately, you haven’t changed anything, but on paper it looks like nobody dies of sepsis anymore. Bias can also change patient outcomes. The number one cause of death in most ICUs is withdrawal of care. If we really believe that patients getting vitamin C will live, we might provide more aggressive care, while more easily giving up on patients who didn’t receive vitamin C, creating a self fulfilling prophecy. 

Therefore, although I think it was reasonable to follow this study with RCTs, this study is completely unconvincing for any real benefit. There are some physiological reasons to think this cocktail might work. I have never been much for physiologic reasoning, because the complex machinery of human homeostasis means that there are always unintended consequences of our actions, and physiologic reasoning so frequently fails. However, if you are interested in reading a little more about the scientific rationale, you can read the discussion section of Marik’s paper, or check out this PulmCrit blog. Although it might be reasonable to start some treatments before strong RCT data, we consistently overestimate benefit and under-estimate harm in medicine. This trial should not have been strong enough data to change anyone’s mind, but it definitely did. Some people even used the awful parachute analogy to claim RCTs weren’t needed. Luckily, those claims were ignored, and we have some follow-up RCTs to discuss.

Other FOAMed

REBEL EM: The Marik Protocol: Have We Found a “Cure” for Severe Sepsis and Septic Shock?

SGEM: Don’t Believe The Hype – Vitamin C Cocktail For Sepsis

EMCrit: Dr. Marik Responds to a Few Tough Questions on the Metabolic Resus of Sepsis Strategy

PulmCrit: Metabolic sepsis resuscitation: the evidence behind Vitamin C

Paper #2: CITRIS-ALI

Fowler AA 3rd, Truwit JD, Hite RD, et al. Effect of Vitamin C Infusion on Organ Failure and Biomarkers of Inflammation and Vascular Injury in Patients With Sepsis and Severe Acute Respiratory Failure: The CITRIS-ALI Randomized Clinical Trial. JAMA. 2019;322(13):1261‐1270. doi:10.1001/jama.2019.11825 PMID: 31573637 NCT02106975

The Methods

This is a multicentre, randomized, double-blind, placebo-controlled trial that compared vitamin C (50 mg/kg every 6 hours for 96 hours) to placebo in adult ICU patients admitted with sepsis who developed ARDS. ARDS onset had to be within 48 hours of enrollment (which led to the exclusion of a lot of patients). They had three primary outcomes (not ideal), and all were disease oriented: change in SOFA score at 96 hours and change in CRP and thrombomodulin levels at 168 hours. Only two of those three primary outcomes were registered as primary outcomes. 

The Results

Out of 1262 eligible patients, they only managed to enroll 167. There was no difference in any of the 3 primary outcomes. There was no difference in 43 of the 46 secondary outcomes. However, just to complicate things, one of the 3 secondary outcomes that was ‘statistically significant’ was 28 day mortality (46% with placebo and 30% with vitamin C, p=0.03, ARR 16.6% 95% CI 2-31%). They did not identify any “unexpected study-related adverse events.”

My thoughts

This is the first large RCT looking at vitamin C in sepsis, and it is clearly a negative study, but it doesn’t slam the door shut on vitamin C in sepsis. It looks at a different population than the Marik study (you needed ARDS) and also used a different protocol (thiamine and hydrocortisone weren’t required). Waiting for ARDS may mean they waited too long to start therapy. They also eliminated a large number of patients, making this a very select population.

Of course, despite being a negative trial, all anyone wants to talk about is the secondary outcome of 28 day mortality. It is always unfortunate when the truly important outcomes are secondary outcomes. Even if they had shown a change in CRP level, I wouldn’t have cared, but demonstrating a decrease in mortality would be groundbreaking. Unfortunately, there were 46 secondary outcomes, so some of them had to be positive by chance alone. The authors even say, “these findings were based on analyses that did not account for multiple comparisons and therefore must be considered exploratory.” 

Perhaps the biggest problem with this trial is survivorship bias. They only measured the disease oriented outcomes in people who were alive. If vitamin C was keeping sicker people alive who would have otherwise died, that could have made the average SOFA score in the group look much worse. Thus, we can’t completely rule out a benefit here. This is why trials should look at important, patient oriented outcomes like mortality, rather than silly lab based outcomes, as their primary outcome.

Other FOAMed

REBEL EM: CITRIS-ALI: Vitamin C in Patients with Sepsis and Severe Acute Respiratory Failure

The Bottom Line: The CITRIS-ALI Randomized Clinical Trial

EM Nerd: The Case of the Sour Remedy

PulmCrit: CITRIS-ALI: Can a secondary endpoint stage a coup d’état?

Paper #3: VITAMINS

Fujii T, Luethi N, Young PJ, et al. Effect of Vitamin C, Hydrocortisone, and Thiamine vs Hydrocortisone Alone on Time Alive and Free of Vasopressor Support Among Patients With Septic Shock: The VITAMINS Randomized Clinical Trial. JAMA. 2020;323(5):423‐431. doi:10.1001/jama.2019.22176 PMID: 31950979

The Methods

This is a multicenter, open-label, parallel-group randomized trial conducted in 10 intensive care units in Australia, New Zealand, and Brazil. They enrolled patients admitted to an ICU with the primary diagnosis of septic shock and compared the same protocol in the original Marik trial (vitamin C 1.5 grams q6h, thiamine 200 mg q12h, and hydrocortisone 50 mg q6h) to an unblinded control group who all also got hydrocortisone, were allowed to get thiamine at the doctor’s discretion, but were not allowed to receive vitamin C. The primary outcome was the time alive and free from vasopressors at 7 days. 

The Results

They include 216 patients out of 786 screened. There was no difference in the primary outcome, time alive and free of vasopressors at 7 days (median of 122 vs 124 hours). There was no difference in all cause mortality at 28 days (22.6% with the intervention and 20.4% with control) or at 90 days (28.6% with intervention and 24.5% with control).

My thoughts

This is the first RCT of the Marik protocol. It is an important trial, although it is far from perfect. It is a relatively small open label trial. The primary outcome is not a patient oriented outcome and the trial wasn’t powered for the outcome we really care about – mortality. (Contrary to the CITRIS-ALI trial, mortality was actually a little bit higher with vitamin C in this trial.) Furthermore, the fact that everyone in the control group received steroids and was allowed to receive thiamine may have biased the results towards no effect. There is also some debate about the timing of therapy. The first dose of vitamin C wasn’t given until about 12 hours, and it is possible that is too late. 

Overall, this trial moves our needle further away from the possibility that the Marik protocol is going to save any lives, but doesn’t definitively prove a lack of benefit. (You can’t actually definitively prove a lack of benefit. The interesting debate is how much time and money should be spent exploring a hypothesis that always had a very low chance of being true.)

Other FOAMed

FOAM Cast: VITAMINS Trials – Vitamin C + Hydrocortisone + Thiamine in Septic Shock

REBEL EM: REBEL Cast Ep74 – Is it all About the VITAMINS in Sepsis?

PulmCrit: Metabolic Resuscitation: Was the answer inside us all along?

St. Emlyn’s: The Vitamins trial. Hydrocortisone, Vit C and Thiamine (Marik protocol – or not?) in sepsis.

Emergency Medicine Literature of Note: Happy VITAMINS Day!

The paper we have all been waiting for: HYVCTTSSS

Chang P, Liao Y, Guan J, et al. Combined treatment with hydrocortisone, vitamin C, and thiamine for sepsis and septic shock (HYVCTTSSS): A randomized controlled clinical trial. Chest. 2020;S0012-3692(20)30552-3. doi:10.1016/j.chest.2020.02.065 PMID: 32243943

The Methods

This is a single-center, single-blind, randomized, controlled trial. They enrolled adult ICU patients with sepsis (using the sepsis-3 definition) or septic shock, and a procalcitonin > 2ng/mL. The intervention group got the Marik protocol (vitamin C 1.5 grams q6h, thiamine 200 mg q12h, and hydrocortisone 50 mg q6h). The control group was given equal volume of saline, but the clinicians weren’t blinded. The primary outcome was 28 day all cause mortality. 

The Results

They enrolled 80 patients, but they were supposed to enroll 140. The trial was stopped early because of “ineffectiveness” and because there was a high incidence of hypernatremia (13 patients in the treatment group and 3 in the control group). There were no statistical differences, but the confidence intervals are huge because of the small trial size. 28 day mortality was 28% with treatment and 35% in the control group (RR 0.79, 95% CI 0.41-1.52; p=0.47). The secondary outcomes are all negative, but most point estimates are on the side of treatment. 

My thoughts

This trial doesn’t add much to what we already knew. It is a bit of a mess. The protocol called for giving placebo, even though clinicians weren’t blinded, but then 28 of the 40 control group patients weren’t given placebo and just got standard care. Also, it is woefully underpowered, because they stopped the trial early, but as far as I can tell, they violated their own stopping rules to do so. 

Again, there is no good evidence that this vitamin C cocktail works, but there isn’t definitive evidence that it doesn’t.

Other FOAMed

PulmCrit: HYVCTTSSS trial: The latest chapter in the Vitamin C saga

REBEL EM: The HYVCTTSSS Trial: The “Metabolic Cocktail” in Another RCT

Overall discussion

Although there are some physiologic reasons to think that vitamin C might help in sepsis, it is pretty unlikely that it is going to be a magic bullet that reduces mortality. Before this saga began, our pretest probability should have been very low. The Marik experience was worth publishing as a hypothesis generator, but the retrospective before and after design is clearly too flawed to influence our thinking. We now have three ‘negative’ trials, but none of them definitively closes the door on therapy. Mortality was decreases as a secondary outcome in the CITRIS-ALI trial, but it was one of 46 secondary outcomes, and most showed no benefit. The results in VITAMINS were the most clearly negative, but treatment was started late, and the control group all received steroids. (There is still a fair amount of debate about the role of steroids in septic shock.) HYVCTTSSS was too small to make any definitive claims, but the point estimate for mortality was again on the side of treatment helping.

Since the original Marik trial, there have been multiple other before and after studies that also showed benefit. (Kin 2018; Sadaka 2019; Long 2020) There is also one negative before and after trial. (Litwak 2019) Of course, these trials suffer from the same methodologic flaws and potential bias as the first trial, but when looking at observational data, you always want to see a consensus. 

One of the problems with treatment cocktails is that it is hard to tell what, if any, part of the combination is providing benefit. By testing agents independently you potentially risk missing out on synergistic actions between the agents. However, testing them together means that the benefit of one agent could be counteracted  by harms from another. The combination makes research more difficult to interpret. One thing is certain: there is a higher chance of harm when using multiple agents.

Overall, the evidence does not provide us with a definitive answer. However, to date there are no RCTs demonstrating benefit from this approach, and research always underestimates harms. Although it may be reasonable to continue to study this approach, there is clearly no evidence to support its use clinically. 

Don’t worry, there are many trials still to come (such as the VICTAS trial). Unless there is something truly groundbreaking, I doubt I will want to dedicate an entirely new blog post to this topic. However, I will try my best to keep this page updated with new trials as they come out, so feel free to check back in in the future.

Bottom line

Based on the available evidence, there is no reason to believe that a metabolic cocktail of vitamin C, thiamine, and steroids improve outcomes in sepsis, but a benefit cannot be completely excluded. There is really no role for this therapy in current clinical practice, but future research may be warranted. 

Other FOAMed

The SGEM: Vitamin C Not Ready For Graduation To Routine Use

UPDATE

Since originally publishing this post, the ORANGES trial was published. (Iglesias 2020) It doesn’t change my practice, so I am not going to dedicate a full blog post to it. If you want to read a critical appraisal, I think Salim Rezaie gets it completely right in his post.

UPDATE #2

Another RCT has been published, and once again it is negative. This was the VICTAS trial, which randomized 500 patients ICU patients to the metabolic cocktail (vitamin C 1.5g, thiamine 100mg, and hydrocortisone 50mg) or placebo. (Sevransky 2021) There were no differences. Once again, I will rely on Salim Rezaie over at REBEL EM to provide you with the details.

References

Chang P, Liao Y, Guan J, et al. Combined treatment with hydrocortisone, vitamin C, and thiamine for sepsis and septic shock (HYVCTTSSS): A randomized controlled clinical trial [published online ahead of print, 2020 Mar 31]. Chest. 2020;S0012-3692(20)30552-3. doi:10.1016/j.chest.2020.02.065 PMID: 32243943

Fowler AA 3rd, Truwit JD, Hite RD, et al. Effect of Vitamin C Infusion on Organ Failure and Biomarkers of Inflammation and Vascular Injury in Patients With Sepsis and Severe Acute Respiratory Failure: The CITRIS-ALI Randomized Clinical Trial. JAMA. 2019;322(13):1261‐1270. doi:10.1001/jama.2019.11825 PMID: 31573637

Fujii T, Luethi N, Young PJ, et al. Effect of Vitamin C, Hydrocortisone, and Thiamine vs Hydrocortisone Alone on Time Alive and Free of Vasopressor Support Among Patients With Septic Shock: The VITAMINS Randomized Clinical Trial [published online ahead of print, 2020 Jan 17]. JAMA. 2020;323(5):423‐431. doi:10.1001/jama.2019.22176 PMID: 31950979

Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: Where are the methods? Ann Emerg Med 1996;27(3):305–8.

Kim WY, Jo EJ, Eom JS, et al. Combined vitamin C, hydrocortisone, and thiamine therapy for patients with severe pneumonia who were admitted to the intensive care unit: Propensity score-based analysis of a before-after cohort study. J Crit Care. 2018;47:211‐218. doi:10.1016/j.jcrc.2018.07.004 PMID: 30029205

Long M, Frommelt M, Ries M, et al. Early Hydrocortisone, Ascorbate and Thiamine Therapy for Severe Septic Shock: A Retrospective Cohort Analysis. Critical Care and Shock. 2020:Pending

Litwak JJ, Cho N, Nguyen HB, Moussavi K, Bushell T. Vitamin C, Hydrocortisone, and Thiamine for the Treatment of Severe Sepsis and Septic Shock: A Retrospective Analysis of Real-World Application. J Clin Med. 2019;8(4):478. Published 2019 Apr 9. doi:10.3390/jcm8040478 PMID: 30970560

Marik PE, Khangoora V, Rivera R, Hooper MH, Catravas J. Hydrocortisone, Vitamin C, and Thiamine for the Treatment of Severe Sepsis and Septic Shock: A Retrospective Before-After Study. Chest. 2017;151(6):1229‐1238. doi:10.1016/j.chest.2016.11.036 PMID: 27940189

Sadaka F, Grady J, Organti N, et al. Ascorbic Acid, Thiamine, and Steroids in Septic Shock: Propensity Matched Analysis [published online ahead of print, 2019 Jul 17]. J Intensive Care Med. 2019;885066619864541. doi:10.1177/0885066619864541 PMID: 31315499

Sevransky JE, Rothman RE, Hager DN, Bernard GR, Brown SM, Buchman TG, Busse LW, Coopersmith CM, DeWilde C, Ely EW, Eyzaguirre LM, Fowler AA, Gaieski DF, Gong MN, Hall A, Hinson JS, Hooper MH, Kelen GD, Khan A, Levine MA, Lewis RJ, Lindsell CJ, Marlin JS, McGlothlin A, Moore BL, Nugent KL, Nwosu S, Polito CC, Rice TW, Ricketts EP, Rudolph CC, Sanfilippo F, Viele K, Martin GS, Wright DW; VICTAS Investigators. Effect of Vitamin C, Thiamine, and Hydrocortisone on Ventilator- and Vasopressor-Free Days in Patients With Sepsis: The VICTAS Randomized Clinical Trial. JAMA. 2021 Feb 23;325(8):742-750. doi: 10.1001/jama.2020.24505. PMID: 33620405