#450 - [Journal Club] - 🫀 From The Heart - Does Early Hydrocortisone Actually Move the Needle in Fluid-Refractory Shock?

In this double-blind randomized controlled trial from northern India, Nim and Adrianne review whether early hydrocortisone reduces 14-day all-cause mortality in preterm infants with fluid-refractory shock. The primary outcome showed no statistically significant difference between groups, though an 11 percent absolute reduction in mortality in the hydrocortisone group raised clinical interest. A major limitation was the high rate of open-label steroid crossover, with over 70 percent of both groups ultimately receiving hydrocortisone. The study highlights the difficulty of achieving equipoise when clinicians already believe strongly in a therapy, and raises important questions about study design in neonatal shock research.

----

Early hydrocortisone verses placebo in neonatal shock- a double blind Randomized controlled trial. Dudeja S, Saini SS, Sundaram V, Dutta S, Sachdeva N, Kumar P.J Perinatol. 2025 Mar;45(3):342-349. doi: 10.1038/s41372-025-02222-3. Epub 2025 Feb 13.PMID: 39948354 Clinical Trial.

Transcript

Adrianne Rahde Bischoff (00:00.000)
All right, Nim. Are we ready for the next one?

Nim Goldshtrom (00:03.342)
I think so. So challenging paradigms in sepsis with TNE, now moving over to questioning the use of steroids in preterm infants. We are going to move on to the next paper, talking about hydrocortisone versus placebo, a topic with a lot of literature and established paradigms even at our hospital, in the setting of fluid-refractory shock.

This group is from northern India. The first author, and I hope I am pronouncing this correctly, is Sankalp Dudeja, and the last author is Praveen Kumar. This was published in 2025 in the Journal of Perinatology: Early Hydrocortisone versus Placebo in Neonatal Shock, a double-blind randomized controlled trial. Kudos to them for embarking on such an ambitious research question.

The background is that during septic shock, inappropriate hypothalamic-pituitary axis activation in severe stress states can lead to hemodynamic instability, and critically ill neonates with low cortisol levels receiving exogenous steroids have shown improved outcomes in a variety of studies. Hydrocortisone is part of an algorithm for pressor-resistant hypotension, where it can improve both vascular tone and myocardial contractility. In their introduction, which I highly recommend everyone read, they reference two trials that showed promising results and formed the basis of their protocol, which we will discuss at the end.

They set out to evaluate whether early hydrocortisone versus placebo in fluid-refractory shock could reduce 14-day mortality. Their cohort consisted of neonates with fluid-refractory shock, and I will note the first limitation here, which is that they did not clearly define the gestational age criteria, though the cohort ended up consisting of preterm infants. The intervention was early hydrocortisone at one milligram per kilo every six hours for 48 hours, followed by one milligram per kilo every 12 hours for the next three days, compared to a placebo, with randomization and double blinding for delivery. The primary outcome was 14-day all-cause mortality. They also collected echo parameters for outflow and PDA, NEC, and a variety of other clinical factors.

Exclusion criteria included obvious volume loss such as dehydration or blood loss after birth, HIE, cardiac disease, NEC, any exposure to postnatal steroids within the prior seven days. They had robust vital sign monitoring directly from the monitors, blood gases, serum cortisol levels, and a variety of echocardiographic function assessments. Enrolled infants were randomized using computer-generated sequences, with stratification by gestational age into less than 30 weeks and greater than 30 weeks. All investigators and outcome assessors were blinded to the intervention.

Vasoactive drug therapy was managed based on unit protocol, with dopamine as first line for hypotensive neonates and dobutamine for normotensive neonates with peripheral circulatory failure. Hypotension was defined as below the fifth percentile for gestational age plus additional clinical criteria, which you can find in the paper. They had planned a priori interim analysis after two-thirds of expected primary outcomes were reached.

Ultimately they enrolled 84 subjects, 43 in the placebo group and 41 in the hydrocortisone group. The mean gestational age was around 30 and a half weeks with a standard deviation of about three to three and a half weeks, so a fairly wide range, anywhere from roughly 26 to 27 weeks up to nearly 34 weeks. There was stratification across birth weight groups and equal distribution for SGA. There were few differences in clinical variables between the two groups, though the placebo group did trend toward having more culture-positive sepsis.

Their tables show vital sign variables with no statistical significance between groups, including LV outflow, SVR index, cortisol levels, and all-cause mortality at 14 days, which was 72 percent in the hydrocortisone group versus 83 percent in the placebo group. There was little to no gastrointestinal hemorrhage or perforation from steroid use. The difference in mortality of around 11 percent was not statistically significant but may be clinically meaningful. There was also no difference in catecholamine-resistant shock.

In adjusted models, mean vasoactive inotropic score at 24 hours was different between survivors and non-survivors, and those who received early hydrocortisone were more likely to be in the survivor group. However, the most important point comes toward the end of the results, which is that they did not account for crossover in the study design. Ultimately, 71 percent of the steroid group and 81 percent of the placebo group ended up receiving open-label hydrocortisone due to catecholamine-resistant shock. SNAP-2 score was also a significant predictor of mortality.

The authors concluded that while not statistically significant, the study showed a clinically important improvement in 14-day mortality of approximately 11 percent that did not hold up in regression models.

Nim Goldshtrom (06:42.258)
This is an interesting paper in a lot of ways, and while I want to be thoughtful to the authors, it raised a lot of things for me in terms of research and study design in general. First, I want to commend this group. In a country that I have to imagine has resources different from what we are used to in westernized modern medicine, they set out and performed a double-blind randomized study in a high-risk group of preterm infants. These are probably not the babies we are most often thinking about, the less than 28-week, less than 1000-gram infants in early transitional shock or late onset septic shock. This is a more varied group of moderate and late preterm infants. And yet they were able to institute double-blind randomization. The unfortunate part is the inability to control for open-label steroid use, and there are ways to plan for that prospectively.

Additionally, if you go back and read the two promising trials they cited in their introduction, one was essentially a very small case series of 11 or 22 infants comparing hydrocortisone to no hydrocortisone with a limited description showing early outcome benefit in smaller preterm infants. The second was a study on steroids for shock in term babies with HIE on therapeutic hypothermia. These are reasonable primary studies to build from, but all of them were limited by very small numbers.

What I take away for trainees and anyone reading this paper is that all the planning in the world may not be enough. Planning a study properly probably requires two to three times the work of actually executing it, because these problems arise. You invest a lot of time. You do not want the absence of an open-label crossover plan, or the failure to pre-specify an intention-to-treat analysis, to be the limiting factor that takes away your message and prevents you from finding statistical significance that may have been there with a stronger design.

I commend the authors for pursuing and publishing this work, though I imagine it was difficult to defend the lack of open-label crossover consideration. The limitations make me wonder whether this really changes anything. We already have a lot of good data about low cortisol levels, stress states, shock, and hydrocortisone use in preterm infants. Fluid-refractory hydrocortisone use is a well-established tool, and we have plenty of supporting studies. It would have been great for this study to be more methodologically rigorous. I don't think it adds much beyond confirming what we already believe, that if you use it, you are probably going to get some benefit, and that at this five-day dosing strategy, you are probably not exposing infants to profoundly significant adverse effects. But Adrianne, how does this change your practice? What did you think?

Adrianne Rahde Bischoff (09:34.100)
I think it basically shows that we have no equipoise. We feel so strongly about this that it is nearly impossible to even study it. The fact that they gave steroids even to babies randomized to placebo tells you that neonatologists want to use it and believe in it. Belief is not the right way to do science, obviously, but it is a challenging thing to study.

Our practice here is very steroid-heavy and very early. I previously practiced in a place where we were truly using steroids for refractory shock, where you were on a first drug, considering a second, and then thinking about steroids at that point. I feel like more and more we are moving toward starting steroids while still on the first agent, in hopes of preventing the need for a second medication.

Nim Goldshtrom (10:13.100)
And this is all shock steroids, right? Not BPD, not respiratory failure management, not the preterm BPD picture. This is all shock. It is interesting.

Adrianne Rahde Bischoff (10:23.100)
We have a protocol here, including a fluid bolus to assess for preload-recruitable stroke volume, one or two agents, and then steroids for pressure-resistant shock. I do not think our use is as ubiquitous. We have a similar approach with the post-operative cardiac patient, where after multiple boluses and escalating inotropes you hit certain thresholds. Those patients are clearly in a stressed state, and steroids help not just for potentially low cortisol, since some of these are full-term surgical babies who may not have that, but because of the capillary leak and inflammatory response from bypass. But we try to be more judicious, because in that world high-dose steroids are clearly associated with poor wound healing and long-term post-surgical infections. There is a lot of dogma. We have tried to pull it back with evidence, but I understand why it is hard.

The practice I struggle with is the somewhat indiscriminate use of steroids for hypotension as first line, which I have seen in the transitional period. The reasoning being that a preterm infant probably has adrenal insufficiency, so let's start the steroid. The problem is that steroids may fix the blood pressure without fixing the underlying problem, which most often is not pure adrenal insufficiency. Most often there is cardiac dysfunction, pulmonary hypertension, low cardiac output, or something else that then becomes masked by that approach and can actually cause more harm than good.

Nim Goldshtrom (12:43.310)
Correct. Or preserved ejection fraction with low SVR, where what you actually need is an alpha agent. And the vasopressin data for hypotension with preserved ejection and systolic function comes to mind. It is complex physiology, and TNE opens up a world of assessment that we simply did not have before, and I hope it continues to pull us away from harder drugs with significant consequences, linear developmental consequences, hyperglycemia, and other things we want to avoid. These are not to say that some of these kids do not have low cortisol and a genuine need for steroids, but we have a lot of targeted drugs that can address vascular, hemodynamic, and cardiac problems and help these babies ride out their illness without necessarily reaching for steroids in situations that are not truly steroid-refractory.

I wonder if instead of steroids versus no steroids, what we should actually be studying is dose and duration. What is the effective dose to get what you want without running into issues like hyperglycemia and excessive cumulative dosing?

Adrianne Rahde Bischoff (13:47.100)
In combination with structured, standardized TNE to address the hemodynamics first. Fix the ejection fraction, fix the RV or LV problem, fix the vascular problem, and then in truly refractory states, ask whether there is evidence of low cortisol that would suggest a steroid is going to be additive rather than just a trade-off between side effects and a genuine stress-response medication.