#021 - Can Continuous Vital Signs Predict Which Preterm Infants Will Need Pressors?

Hello friends 👋

Can the numbers flashing across the bedside monitor predict which preterm infant will need a pressor? In this episode, Rupa sits down with Dr. Nicholas Heitkamp, a third-year neonatology fellow at the University of Virginia, whose NeoHeart London presentation won the oral presentation award. Nick explains how he used UVA’s continuous vital sign repository to find that heart rate skewness, mean SpO2, and SpO2 variability, combined with gestational age and birth weight, help identify infants under 30 weeks who go on to receive vasopressors. He shares how a poster at PAS led to external validation with WashU, how he found his mentor, what improvement science taught him, and why he’s heading to Iowa for neonatal hemodynamics.

----

Short Bio: Nicholas Heitkamp, MD, MSc, is a third-year Neonatal-Perinatal Medicine fellow at the University of Virginia and a member of UVA's Physician-Scientist Training Program. He earned his MD at Indiana University and completed pediatric residency and an improvement science fellowship at Eastern Virginia Medical School. His research, mentored by Dr. Brynne Sullivan, uses continuous bedside monitor data to identify physiologic signatures that precede shock in extremely preterm infants. This work won the oral presentation award at NeoHeart Europe 2026 in London. He founded and chairs UVA's NICU point-of-care ultrasound program and serves on the editorial board of NeoReviews. In July 2027 he will begin a neonatal hemodynamics fellowship at the University of Iowa, with the goal of building a neonatal hemodynamics clinical and research programs.

----

The transcript of today's episode can be found below 👇

Rupa (00:00.650)

Hi, everyone, this is Rupa’s Fellows Friday. I know it’s been a bit, but I have a really fantastic guest in our studio today. I’m delighted to welcome Dr. Nicholas Heitkamp from the University of Virginia. Nick is a third-year neonatology fellow and a member of the Physician Scientist Training Program. He went to medical school at Indiana University, did his pediatrics residency and an Improvement Science Fellowship at Eastern Virginia Medical School, which is interesting, and I’d like to pick your brain about that fellowship. He’s currently a third-year fellow mentored by Dr. Brynne Sullivan, and he’s given some fantastic presentations. I had the honor of listening to his oral presentation at the NeoHeart conference in London, which also won him the oral presentation award there. He’s done so many other things that we’ll highlight as we go through our conversation today. Welcome, Nick. I’m so excited to have you in the studio.

Nick Heitkamp, MD (01:13.326)

Thanks for the invite. It’s been great meeting you in London and connecting ever since. I can tell we’re going to be friends for a long time.

Rupa (01:21.846)

A hundred percent. We have so many things we talk about, even just for fun. Like I said, I was really impressed in the audience listening to your talk. It was fantastic that you walked us through your fellowship project, your scholarly activity. I’d love for you to describe your project in your own words, so our listeners can get some insight into the presentation you gave in London.

Nick Heitkamp, MD (01:51.503)

Absolutely, and thanks again. What I talked about in London is something I started at the University of Virginia using data science and big data. It’s all centered on something we see and interact with every day in the NICU: the bedside monitors. We constantly see numbers for various features popping up on the screen and changing, but all we see is that number in that moment. What I’m interested in researching is what happens if we take those numbers as a continuous stream of data, graph them over time, and look at their distributions. Can we learn anything from the trends or shapes they form, either on their own or with other trends as a cross-correlation? Can we use them to understand physiology or predict a certain outcome in babies?

Specifically, I looked at preterm infants in their transitional period, the first 72 hours, and whether I could use heart rate and SpO2 (oxygen saturation) data to separate babies who end up with vasopressor-treated hypotension from those who don’t. We’re lucky at UVA to have the world’s largest repository of continuous neonatal vital sign data, terabytes and terabytes on computers in a computer lab. I went back retrospectively five years, looked at all of our babies under 30 weeks, and pulled their continuous vital signs.

We found three main features, heart rate skewness, mean SpO2, and SpO2 standard deviation, that separated the babies who went on to receive a pressor from those who didn’t. When we paired those three features with the two that anyone researching preterm infants uses, gestational age and birth weight, the multivariable model had a really good area under the curve for predicting which babies would be treated, beyond gestational age and birth weight alone. Understandably, it was a small subset of babies, because it’s a single center. But it was a really interesting way to start generating hypotheses, and we have some future steps planned. That was the gist of what we did and talked about at NeoHeart.

Rupa (04:54.590)

That’s fantastic. One thing I was curious about in the audience was how you defined vasopressor-treated shock. I’m sure you’ve felt this as a fellow: we have different definitions of hypotension for our babies. It gets more complicated with preterm babies specifically, where we’ve historically used techniques with very limited evidence to understand what hypotension means in that age group. By the way, for our listeners, Nick’s presentation got a lot of questions. Everyone was curious about his data, and it was very well received. But my question was specifically how you defined vasopressor-treated shock. I have some related questions, but I’d like your perspective first.

Nick Heitkamp, MD (06:04.331)

That’s such a phenomenal question, and something my mentor, Brynne Sullivan, and I went back and forth on. She actually just gave a lecture, right before I jumped on with you, about being a mentee in fellowship and creating a project you can do and finish during fellowship. As a fellow-initiated project, I wanted the scope to be pragmatic. So we defined shock as the point when the clinician decided to start a pressor. We also looked at the subset of infants who got a fluid bolus before the pressor, knowing that, unless there was an abruption or some other delivery event, these infants likely aren’t so fluid-depleted that they need a bolus. But we primarily defined it as infants who were started on a pressor, as the clinician decided.

Just as you alluded to, I was reading a paper describing maybe twenty different definitions of shock used in NICUs in the U.S., which blows my mind and makes it really hard to study.

Rupa (07:26.856)

Not surprised at all.

Nick Heitkamp, MD (07:31.896)

If we could agree on these numbers, it would make things easier. But there’s an aspect of culture, comfort, and how you trained, so I think it will be really hard. I am interested in retrospectively looking at the babies’ blood pressures at the time the pressor was started, to get an idea of what numbers we’re dealing with when we do start them. But in this case, I used the time to pressor onset.

Rupa (08:04.256)

It’s interesting you mention the blood pressure when the vasopressor was started. Sometimes these babies are also started on hydrocortisone for presumed adrenal insufficiency, and sometimes they’re on fairly substantial doses of sedation, which might also bring that number down. There are so many things happening at once in these babies that you don’t know where to start and stop with all these confounding variables. It’s very interesting that you took on this project, because someone needs to do this, so I’m so glad you did.

Nick Heitkamp, MD (08:41.013)

I totally agree. From the perspective of a fellow training as a physician scientist and thinking about research: as neonatologists, we do so many things to these babies at once. How can we go through with a comb and separate which intervention led to which outcome? Sometimes it might be impossible. That’s what I’m hoping we can leverage big data for, and as we learn more and have more data available, I think it’s leading to some really interesting conclusions.

Rupa (09:19.708)

Fantastic. What was the one takeaway from your project, and how is it translating into the future of the project? What does that next phase look like?

Nick Heitkamp, MD (09:35.256)

I’ll start with the future question and then what I learned most. I presented this as a poster at PAS (Pediatric Academic Societies), and my mentor brought over one of her colleagues, Zach Vesoulis at WashU, who does similar work with continuous vital signs and big data. He said, “We could get you a similar, matched retrospective cohort here at WashU, and you could use it as external validation.” One lesson from that as a trainee: sometimes you feel like, “I’m just submitting another abstract to another conference, and I have to go stand by my poster.” But you never know what even 15 minutes at your poster can lead to, because conferences concentrate so many like-minded people into two or three days. We already had a data use agreement, so it was easy to get the data, and I’m analyzing it now. So far, it’s really interesting to see the differences, but actually a lot more similarities, between the sites.

As for what I learned most, it may not be specific to the science, although I did learn how to create a data pipeline and how to work with our data scientists. We’re really lucky to have a lot of data science support, and I learned how to bring questions to them when I’m stuck. So there were a lot of interpersonal lessons about working in a research team.

But mostly, I started fellowship with a totally different interest. I knew I wanted to do research, and like a lot of fellows right now, since it’s a hot topic in neonatology, I knew I wanted experience and depth in point-of-care ultrasound. I was lucky to have some experience at the end of residency, with training from Will Corder at CHOP (Children’s Hospital of Philadelphia). I came to UVA gung-ho about ultrasound, formed a group, created a curriculum, and we’ve been working our way up, first with vascular access and then lung ultrasound.

While I was working on point-of-care ultrasound, I developed a fascination with preterm infants in the transitional phase. That’s when I learned about UVA’s data science resources. This isn’t specific to UVA, it’s a trend at large institutions, but we have what’s called the Center for Advanced Medical Analytics, and one of our neonatology faculty, Brynne, now my mentor, is its director. It’s serendipitous how you rub shoulders with people along the way, talk with them, and find they’re interested in mentoring you. So while I was working on point-of-care ultrasound, which is still a big interest of mine, especially lung, I learned about continuous vital sign analytics and what we have available locally. When I realized we had a repository of hundreds of thousands, even millions, of pieces of vital sign data just waiting for a question to be asked, I knew I was going to tap into it and try to ask what I thought might be a simple question and get some sort of answer.

To summarize, I’m always looking for the next thing where I can plant a seed, even if it doesn’t grow into a full plant, and try something new. Getting into data science and big data wasn’t something I intended, but I used local resources and a really great mentor.

Rupa (13:49.311)

That’s fantastic. You touched on some important points for fellows. I’d say most fellows in their first year have absolutely no clue what they want, and that’s completely okay. For first-years listening, it’s completely fine to come in with an open mind and explore during your first few months, before you take your peds boards, which, it turns out, were this week. It’s lovely to have that phase of “I don’t know what I want, but I’m on service and I’m finding patient populations, or mentors, who might spark that interest.” Still, it helps to have a general idea of the patient population that interests you, whether NEC (necrotizing enterocolitis) research, cardiac physiology, or respiratory physiology. Having a general bucket is always helpful, and then you’ll magically meet the person you relate to in a lot of ways, which sounds like Brynne was for you.

For our listeners, share how you decided, so to speak, that she was going to be your mentor. And what’s the one piece of advice you’d give fellows looking for a mentor? This is around the time first-year fellows might be looking, so what’s your one piece of advice for finding the right one?

Nick Heitkamp, MD (15:28.961)

That’s a really great question. I approached mentorship the way I approach research curiosity. I went to faculty I got along really well with, whether on clinical service or in research. I had an ultrasound mentor I was doing great ultrasound work with, and I did improvement science, so I had an improvement science mentor. I started collecting mentors by happenstance when I realized we had mutual interests.

When I first went to Brynne, I never thought I’d get into anything data-related. I still feel like I need to take some college classes in data science. It’s taken me back to my horrible calculus days, which I thought were behind me, and life has an interesting way of bringing you back. I started by setting regular meetings. Our faculty, and I’m sure it’s like this at a lot of places, are great about saying, “Our calendars are always open. Just put something on my calendar and we’ll meet.” So I did, and it turned into a meeting almost every Monday. Then one day, I remember walking out and saying, “So I think you’re my primary mentor now, right?” And she said, “Yep, that sounds great.”

So these things happen by chance. For fellows starting out who may or may not be interested in research, it helps to be personally motivated and to set your own deadlines. It’s easy to have the intention and desire to do something, and maybe have a few conversations. What has really pushed things forward for me is putting the next meeting on the calendar, even if it’s a month away, because then I know I have to have something ready to talk about. Otherwise it’s going to be awkward, and they’ll wonder why I met with them. So I’d encourage people to put something on the calendar, even weeks out, or set themselves an internal deadline. I noticed my relationships with mentors worked a lot better when I was more proactive about pushing things along.

Rupa (18:03.880)

Agendas and deadlines are great advice. It’s been a while since I finished fellowship, but one thing I think about is that if your mentor is giving you their time and attention, you have to make the best use of it. They’re all busy people. If you get 30 minutes of your mentor’s time, have an agenda, with set discussion points. Include things about your life too: “Just so you’re aware, this is happening in my life.” That’s always helpful for building a personal relationship with your mentor. And the agenda should always end with when your next meeting will be.

Nick Heitkamp, MD (18:58.861)

That’s right.

Rupa (19:02.378)

Fantastic. I hear you’re going to be the incoming neonatal hemodynamics fellow at the University of Iowa, which is amazing. I think I know the answer, but I’d love to hear from you what got you interested in applying, and what your goals are for the fellowship that might apply to your future.

Nick Heitkamp, MD (19:30.593)

First, I want to say how excited I am about the opportunity. I realized I wanted to do it as an offshoot of my ultrasound work. As is well described in several papers now, especially from the NeoPOCUS collaborative, point-of-care ultrasound has its own scope, design, and questions in one sphere, and hemodynamics has other questions and uses. They overlap in a Venn diagram, with cardiac POCUS, which could be defined within either circle in different ways.

As I thought about our point-of-care ultrasound program, I strongly believed everybody, including nursing, should have a foundation in vascular access, if we think of it as a pyramid. Then I found my love and research interest in lung ultrasound. I knew there was always a top piece of the pyramid, and it was going to be cardiac. Based on local and national conversations, it makes a lot of people nervous, for various reasons, to think about using ultrasound at the bedside to answer a question about the heart, especially a sick heart, particularly when credentialing comes up. Indications for cardiac POCUS exist and are very helpful, especially in a code situation, like identifying tamponade, fluid around the heart that you need to tap. That’s a huge one.

But I realized there are so many other questions that looking at the heart and blood flow with ultrasound, and assessing the heart’s dynamic output, could answer and help us change management at the bedside. When I first heard about it, I read all the papers I could and realized, wow, the care you can provide to an infant using serial cardiac ultrasounds performed by a neonatologist to answer a physiology question, which is what differentiates it from the anatomic questions or congenital disease our pediatric cardiology colleagues address. I realized its potential. The more I read, the more I thought it could and should be used in every NICU. The problem right now is training: more centers would love to implement it than there are training centers. So I’m so grateful and excited that Iowa was willing to accept me, and I hope I can give back to the field as much as the field is going to give me with this training.

Rupa (22:32.852)

That’s fantastic. You rightly put it that it enhances your own capabilities as a bedside clinician. We’re moving into a world where phenotyping, specifically cardiac phenotyping, has almost become a vital sign in itself. It’s such an important skill to bring to your clinical decision-making at the bedside. Kudos to you for taking that step. I hope it plays a significant role in all your future endeavors.

Speaking of fellowships, I didn’t give it enough credit earlier, but I’d love to hear about the improvement science fellowship you did during residency. Can you share more?

Nick Heitkamp, MD (23:27.691)

Absolutely. That was another serendipitous whim. We all get lots of institutional emails that we don’t read. In residency, I happened to open one and scroll to the bottom, and there was an advertisement for trainees who wanted to learn about quality improvement. It was a year-long improvement science fellowship led by one of our pediatric emergency medicine doctors, Sandip Godambe, who is now at CHOC (Children’s Hospital of Orange County). It was a concurrent fellowship, not ACGME (Accreditation Council for Graduate Medical Education) accredited at the time: you did residency and the fellowship at the same time. I did an extension, so it ended up being about a year and a half.

It was like being in college again. There were lectures and three textbooks. He taught us the science and theory behind improvement science and how to leverage it. It really is a science: the statistical process control charts you create, and the software used to create them, can show you when variation in your own system is due to an intervention you made, beyond the common variation within your system. That’s when I learned that there’s basic science, clinical research, and improvement science, each in its own category because they serve different ends. Basic science is about the principles of cells and their interactions in the body. Clinical research tests something at the bedside. Improvement science isn’t testing a new intervention on the patient, but on the system overall, to maximize outcomes in your local system. It blew my mind, I went down the rabbit hole, and I really latched onto it.

I started there and continued in fellowship, using improvement science to improve outcomes in golden hour admissions, in our institution for the smallest babies under 28 weeks. You can pick any aspect of the outcome and try to improve it. We focused on overall time, making it a true golden hour, and on what you need in a very complex situation. We all work in very complex systems, so it’s about how to move that needle. It was a really interesting, different way of thinking. I guess you just throw your hat in the ring, and sometimes things work out. I got accepted and thought, “What the heck, let’s give it a shot.”

Rupa (26:33.312)

That’s amazing. I also want to quickly highlight some of Nick’s achievements for our listeners. As I mentioned, Nick won the oral presentation award at the NeoHeart Europe conference in London. He has also presented his work at several conferences, including the AAP (American Academy of Pediatrics) and PAS, and he won the Top Fellow Submission Travel Award at Hot Topics in Neonatology. He’s also on the NeoReviews editorial board, which I’m part of too, and that’s where we became friends. You’ve done some fantastic work in your career so far, Nick. My last question: what’s your big piece of advice for incoming fellows who might be interested in your area of research?

Nick Heitkamp, MD (27:29.707)

As I think about what I’ve learned and what people could keep in mind when starting fellowship, it’s something you said at the beginning. Even if you’re not committed to a particular subject, and maybe you know it’s not research, that’s fine. I absolutely love my co-fellow. We’re like two peas in a pod, and she knew from the gate that research was off the list, and there’s absolutely nothing wrong with that. So look at what remains, and think, often with the advice of a really good mentor, about one small niche you can make your own. That’s your starting interest. Pick something, so that when you talk to people at a conference, or when you apply for jobs, which third-years are thinking about right now, you can say, “This is what I plan to do in the future.”

Having a plan helps, whether or not it ends up happening the way you think. Some of the best careers were never what people envisioned at the beginning. But having some sort of plan you can aim for is really helpful for setting goals along the way and having something to show for it at the end. It’s hard, because some people have no idea, and that’s just as fine. But pick something small you can tackle easily, and see whether any offshoots come from it that excite you, or whether you just cross it off the list and say, “Now I know I don’t want to do that.” That’s equally helpful.

Rupa (29:29.200)

Really wonderful advice. I think that’s so valuable. Nick, this was fantastic. Thank you for coming to hang out with me today and discuss all the fantastic things you’ve done in your career. Good luck with a beautiful year of learning neonatal hemodynamics, and good luck with everything.