Episode Transcript
[00:00:00] Welcome to this week in Heart rate Variability. I'm Matt, and this is the show where we take the newest research on heart rate variability. The small beat to beat changes in the timing of your heartbeat that give us a window onto the autonomic nervous system.
[00:00:11] And we try to make honest sense of it together, not just what a study found, but what it can and cannot tell us and what it might mean for the people we work with and for ourselves. Before we go any further, a brief but important word. Everything we discuss on this program is for educational purposes only. It is not medical advice, and it is not a substitute for guidance from a qualified professional who knows you and your history. Heart rate variability is a powerful lens, but it is a lens, not a diagnosis. If anything you hear today raises a question about your own health or the health of someone you care for, please bring that question to a licensed provider who can see the whole picture. If you have been with us over the last few weeks, you know we have changed how this show is built. We used to gather studies simply because they landed in the same week, and we would move through them one after another, however loosely related they were. Now we organize each episode around a single theme, a single question, so. So that the research can talk to itself and we can think more deeply rather than more broadly. Last month we spent a week on performance, and last week we spent a week on mental health. And the response to that format has been wonderful. So here's a quick reminder for anyone joining us. For the first time, each week has a center of gravity, and today it is medical health. We are looking at what heart rate variability can tell us when the body is under genuine medical strain, when there is diabetes, when there is cardiovascular risk that plays out over decades, when the brain itself has been injured, when cancer has landed someone in critical care, and when and when a chronic neurological disease wears a person down with fatigue. That is a wide territory and I want to lay out the map before we start walking so you always know where we are. We will open with a study that asks a very practical and very humbling question. Whether the rhythm of the heart can be used to sense blood sugar and whether a fashionable statistical fix actually helps. From there, we will move to a long patient study that followed people for more than two decades to see how resting heart rate and heart rate variability relate to the risk of dying from cardiovascular and coronary heart disease and how that relationship changes depending on where someone sits on the spectrum from normal blood sugar to diabetes. Then we will look at something quieter and cleverer. What happens to heart rate variability during a routine brain recording and whether a perfectly normal brain test can still hide a nervous system that is not quite balanced. That is where we will pause for a word from the people who make this show possible. After the break, we will turn to the most fragile patients of all. We will look at brain injured people in a neuro intensive care unit and at how the coupling between breath and heartbeat, a coupling that is normally so reliable most of us never think about it, can flatten out and even reverse when the brain stem is under siege. Then we will widen our view to critically ill cancer patients and a thoughtful review arguing that we should stop looking at the autonomic nervous system through a single instrument and start listening with several at once. And we will close the research with a study on multiple sclerosis, fatigue and whether heart rate variability can serve as a biomarker for the kind of exhaustion that does not lift with a good night's sleep. Then, as always, we will sit back and try to draw the threads together, because the whole point of a themed week is that the last study should change how you heard the first.
[00:02:52] Before we jump in, one more note about how we will move through this. Each of these six studies is doing something different from the others. One is asking a methodological question about how to build honest models. One is asking a long patient question about who dies of what across decades. One is asking whether a routine test carries hidden information.
[00:03:09] One is asking what happens at a bedside when the brain is grievously injured. One is asking us to broaden how we assess the autonomic nervous system altogether. And one is asking whether a specific chronic disease can be understood in part through the lens of autonomic flexibility. What links them and why they belong in a single medical week is that each is trying to say something honest about the relationship between the heart's rhythm and the body state under real medical stress. Some of what you will hear will be encouraging. Some of it will be sobering. All of it, I hope, will reward the patients you bring to it. So let us begin. Our first study is by Ahmed Basidizam and Sarangtami Botambi Singh, who work in the Department of Computer Science and Engineering at the School of Engineering, Tezpor University, India. And I want to start here because this study models something I wish more research did intellectual honesty about its own methods. The question they set out to answer has real commercial and clinical momentum behind it. Right now. Can we use heart rate variability extracted from the electrocardiogram, a recording of the heart's electrical activity to estimate a person's blood sugar level. If that were reliably possible, you could imagine a future where a wearable device on your wrist gives you at least a rough sense of your glycemic state or without ever pricking your finger. That is the dream, and it is a big one, because for people living with diabetes, the finger prick has been a fact of daily life for decades. Anything that could ease the burden of daily testing or catch blood sugar drift earlier would be genuinely valuable. This study is a careful, sober look at how far the reality currently reaches, and I want to walk through it slowly because the way this team asks the question tells us as much as the answer itself. I should also say up front that this is the kind of study we do not spend enough time on in this field. It is easy for research to reach for the biggest claim it can plausibly make. It is much harder to spend the same effort on trying to disprove yourself. What I love about this paper is that the disproving is the point. The authors are not selling you a wearable they are asking whether the very idea of reading blood sugar from the B2B timing of the heart survives an honest test. They worked with data from 43 men who had type 2 diabetes. From each person's electrocardiogram they derived a set of heart rate variability features, mathematical descriptions of how the intervals between heartbeats vary over time, and then asked whether machine learning models could use those features to predict two things. The first was hemoglobin A1C, usually written as HbA1C, which measures average blood sugar over the preceding two to three months. The second was fasting blood glucose Fbg, a snapshot of blood sugar after a period without eating. So one long term marker and one short term marker, both estimated purely from the timing of heartbeats. Now here's where the honesty comes in and I want to slow down because this is the heart of why the study matters. In machine learning, it is disturbingly easy to fool yourself if you're not careful about how you separate the data you train your model on from the data you test it on. Information can leak across that boundary and your model can look brilliant in the lab and then fail completely in the real world. The authors used a rigorous approach called leave one subject out cross validation. The idea is simple to state and demanding to honor. You train the model on everyone except one person, then test it on that single held out person and you repeat that for every individual in the study.
[00:06:03] This is the honest way to ask whether a model will work on someone it has never seen before, which is of course the only question that matters clinically. They also kept all their data pre processing strictly within each fold so that no information from the test set could leak into the training process. With that discipline in place, the best performing model was something called an extra trees regressor. It is a machine learning approach that builds multiple decision trees on the data and combines their predictions. It is generally robust to noisy inputs, but making it a reasonable choice for this kind of problem. And the results were modest. For hemoglobin A1C, the model explained about 22% of the variation with a correlation coefficient of roughly 0.47 and the association was statistically significant.
[00:06:42] For fasting blood glucose it was even weaker, explaining less than 9% of the variation with a correlation of around 0.34. Now if you have not spent much time with numbers like these, let me put them in context. Explaining 22% of the variation means that nearly 4/5 of the variation in hemoglobin A1c among these people was was not captured by the model. So there is a real signal here it is nothing. Heart rate variability is associated with glycemic status and that association is detectable even in a small sample when you are being honest about your validation. But it is a long way from the precision you would want before trusting a number to guide a treatment decision. The gap between there is a signal and we can act on this alone is enormous and studies like this one help us understand where that gap lies.
[00:07:24] And then comes the finding I most wanted to share with you because it runs directly counter to a popular assumption. We know that heart rate variability declines with age. It is one of the most reliable facts in this entire field.
[00:07:37] So a very reasonable person might assume that if you are trying to read blood sugar off of heart rate variability, you had better first correct for age or the age related decline will muddy your signal. The authors tested six different methods of adjusting for age and not one of them improved the model's performance. Not one that is a genuinely useful negative result. It suggests that, at least in this data set, the age related decline in heart rate variability did not confound the relationship between heart rate variability and glucose. The signal, such as it is, was not being smuggled in through age. Now, I want to be careful not to overread. This absence of evidence is not evidence of absence, and a larger, more diverse data set could yet reveal that age normalization matters in some subgroups. But it is a useful reminder that Our intuitions about what to control for when we try to extract meaningful signals from data are often assumptions rather than facts. Test the assumption before you build the correction into your product. Let me be clear about the limitations because they are substantial and the authors would be the first to name them. 43 people are a small sample and all of them were men, so we cannot assume any of this transfers to women whose autonomic physiology differs in important ways. There are documented differences across the menstrual cycle and the balance between parasympathetic and and sympathetic influence, and in how those balances change with age and metabolic disease, and none of that has been probed in this data. Everyone in the study already had type 2 diabetes, so this tells us nothing about detecting the earliest drift toward dysglycemia in people who are still healthy, which is arguably the population for whom a non invasive screening tool would be most useful. And because this is a cross sectional observational analysis, we are strictly in the language of association. We can say heart rate variability features were associated with glycemic markers. We cannot say that blood sugar causes these heart rate variability patterns or or vice versa, or that intervening on one would affect the other. What this study really gives us is a piece of methodological conscience, a reminder that the way you validate a model matters at least as much as the model itself, and that a fashionable correction is not worth adding unless you have checked that it actually helps. That is a lesson I would like to see. Travel far beyond this one paper. Every time you see a headline announcing that a wearable can now detect some condition or another, hold it up to the standard this study sets. Leave one subject out. Validation pre processing Kept honest, the researcher is actively trying to see whether their own assumptions hold and ask whether that headline has earned the same kind of care. Our second study asks us to lengthen our sense of time considerably. It comes from a large and distinguished group of researchers Kolade Oluvag, Bemigun, Dan Ziegler, Alexander Strom, Margit Heyer, Gidanje Bonhoeff, Michael Roden, Wolfgang Rathmann, Christa Meisinger, Annette Peters, Stephanie M. Hauk, Agnes Petrera, Moritz F. Sinner, Stefan Kabb, Barbara Horind and Christian Heritor. Their work is drawn from a long running population study in Germany and what they have done is follow a large group of people for more than 20 years to ask a question that only time can answer. How do resting heart rate and heart rate variability relate to the risk of dying from cardiovascular disease and from coronary heart Disease, specifically, I love a study like this precisely because of its patience.
[00:10:26] So much of what we cover on this show is a snapshot, a measurement taken once in a lab on a particular afternoon. But cardiovascular risk is not a snapshot phenomenon. It unfolds across a lifetime. To see it clearly, you have to be willing to enroll people, measure them carefully, and then wait for years, for decades, to see what happens. This is an observational cohort study, which means the researchers did not assign participants to any intervention. They measured people as they were and watched what followed. So everything we say here lives in the language of association and risk, never in the language of proof or cause. With that understood, let us look at what they found. I want to underline the associational point one more time, because with a study of this length and complexity, it becomes very tempting to slide into causal language. When a paper follows people for more than two decades and finds that a particular measurement predicts who dies of what, the natural instinct is to say that the measurement caused the outcome. But it did not, and the researchers know it did not. What they showed was a statistical relationship consistent across many years between a set of baseline physiological measurements and the risk of certain deaths later on. That is powerful. It is also carefully bounded. The measurement is a marker of something happening in the body, and that something may be part of the causal chain, a downstream shadow of a different cause or or share a common upstream driver with the outcome. A well designed cohort study is our best observational tool for spotting these relationships. It is not and cannot be a randomized trial. They examined resting heart rate, simply how many times the heart beats per minute at rest, which I will refer to as resting heart rate throughout, alongside several indices of heart rate variability. Those indices included measures from the frequency domain, an approach that breaks down variability and heartbeat timing into rhythmic components at different frequencies. One of these is very low frequency power, abbreviated vlf, which captures the slowest of these rhythms. They also looked at a time domain measure called the root mean square of successive differences RMSSD, which reflects rapid B2B changes strongly tied to the parasympathetic nervous system, the branch that calms and restores the body. And they used a more sophisticated nonlinear measure based on poincheurier plot analysis, a way of charting each heartbeat interval against the next to reveal patterns that simpler measures miss. Here's the shape of what they found. Across the whole group, a higher resting heart rate was associated with a greater risk of dying from both cardiovascular disease and coronary heart disease. That is consistent with a great deal of prior work, and it is worth repeating because it is so actionable. At the population level, a persistently high resting heart rate is a signal worth taking seriously. Resting heart rate is arguably the most accessible autonomic measurement we have. Anyone can put a finger on the wrist and count and when your resting heart rate is drifting upward year over year without a clear explanation such as a new medication or a change in fitness, it warrants a conversation with a clinician. That basic finding has stood up across many populations and many decades of research, and this new work adds another substantial piece of long term evidence to that stack.
[00:12:57] But the more interesting findings emerged when the researchers divided people by glucose status, separating those with normal glucose tolerance from those with prediabetes and established type 2 diabetes because the relationships were not the same across those groups. In people with normal glucose tolerance, very low frequency power was directly associated with cardiovascular mortality. In people with type 2 diabetes, the relationship between the root mean square of successive differences in cardiovascular mortality was not a simple straight line at all. The authors describe it as an S shaped nonlinear relationship, which is a fascinating and slightly unsettling result because it means that more heart rate variability was not uniformly better across the whole range and in people with prediabetes. One of the nonlinear penturier derived measures was inversely associated with cardiovascular mortality, meaning higher values tracked with lower risk. What I take from this is that the metabolic context changes the meaning of the measurement. The same heart rate variability index does not carry the same message in a person with normal blood sugar as it does in a person with diabetes. This is a theme we will return to today, so hold on to it. The number is not the finding. The number in context is the finding. And I would go a step further. This study is a quiet argument against a certain kind of oversimplification that keeps creeping into the wellness conversation around this metric. You will sometimes hear the phrase higher heart rate variability is always better. It is a comforting shortcut. But a shortcut is exactly that. In a population with normal metabolism, higher variability may indeed track with lower cardiovascular mortality across most of the range. In a population with type 2 diabetes, the that same relationship became nonlinear, an S shape meaning that at certain points in the range going higher did not confer additional benefit. It might even have signaled something else was going on. The body is not a linear system, and once you accept that, you start looking at your own readings with more appropriate curiosity and less anxious ranking. The authors are careful about the limits of their work, and we should be too. When you slice a cohort into subgroups by glucose status, each subgroup is necessarily smaller and statistical power drops. Some of these associations rest on smaller numbers than the headline cohort size might suggest.
[00:14:51] A very long follow up is a strength for capturing rare outcomes like death, but it is also a subtle weakness because a single baseline measurement taken more than 20 years earlier may become an increasingly weak predictor as years pass and people's lives change. Someone's blood sugar status at baseline is not the same as it is two decades later. Someone's resting heart rate at baseline is not the same as their resting heart rate 20 years later. So the researchers are in a sense using an old snapshot to predict an outcome that unfolded across the life the snapshot could not have anticipated. That is not a flaw specific to this study. It is a feature of long cohort work everywhere, and it is why we should read these findings as directional rather than as precision estimates. The researchers also noted that the association between these autonomic measures and non fatal cardiovascular events was weaker than that with mortality. It suggests that the signal these measures convey may reflect the severity of underlying autonomic disruption rather than specific cardiovascular events per se. And in a note of real practical humility, they observed that some of the more complex calculations they used, the sophisticated nonlinear analyses most likely exceed what could realistically be done in an ordinary clinical setting today. So this is discovery science pointing toward possibilities, not a recipe you can carry into a clinic tomorrow, but as a demonstration that heart rate variability carries long range prognostic information and that its meaning shifts with metabolic state. It is a genuinely valuable contribution. Our third study is smaller and quieter, and I find it quietly profound. It comes from Miles S. Evans, Caterina Dimarchi Asenson and and Vishwanath Sagi, and it examines what heart rate variability does during a very ordinary medical test, a short term recording of the brain's electrical activity called an electroencephalogram eeg.
[00:16:26] When someone is undergoing an electroencephalogram, small sensors are placed on the scalp to record the brain's electrical activity, usually to look for signs of conditions such as seizures. What this research team realized is that during that same recording, there's also a channel capturing the heart's electrical activity, and that channel is a hidden opportunity. While the neurologists are reading the brain, the heart has been quietly telling its own story the whole time, and almost nobody has been listening. This was a retrospective single center study, meaning the researchers reviewed recordings already collected at one institution rather than collecting new data. Prospectively, they examined 103 electroencephalograms that had all been Read as normal. That is the crucial detail in adults. And they extracted heart rate variability from the single channel of electrocardiogram data recorded during the first five minutes of wakefulness. So picture the situation. Here are 103 people whose brain recordings came back completely normal by the usual standard. These are reassuring results. Nothing to see here, except there was something to see if you looked at the heart. When the researchers calculated heart rate variability measures, including the root mean square of successive differences, again reflecting rapid parasympathetic driven changes, and the standard deviation of the normal to normal intervals written as sdnn, which is a broad measure of overall variability across the recording. They found that these values were significantly lower than the published normal values for healthy people in their subjects. The root mean square of successive differences averaged around 28, compared with a typical reference value of about 42. That is a meaningful gap. These were people with normal brain recordings whose hearts nonetheless showed reduced variability, that is reduced flexibility in the autonomic nervous system. And I want to sit with what that might mean clinically. These are people who are referred for a brain recording because someone you know, physician, or the patient themselves, you had a concern about neurological function. The brain test came back reassuring. The natural next step in most clinics is to communicate that reassurance and move on. But this study suggests that the very same recording carries additional information about how the nervous system as a whole is functioning. And that information might be relevant to the concern that brought the person in in the first place. Not a diagnosis, just another piece of the picture available essentially for free. The researchers also observed the age related decline we keep meeting. Heart rate variability was lower in older subjects, exactly as the broader literature would predict. And they noted that the different stages of consciousness during the recording shaped the heartbeat. As people drifted from wakefulness toward deeper sleep, the intervals between beats lengthened. The larger conclusion they draw is that heart rate variability, captured essentially for free from a recording that is already being made, is a reliable and sensitive marker of autonomic function. And that it can reveal what they describe as subclinical autonomic dysfunction. Dysfunction that is real, but below the threshold of obvious symptoms, even when the brain recording itself looks perfectly normal. Now, the limitations. This is a retrospective study from a single center, which means we should be cautious about generalizing to other populations, equipment and recording methods. Because the analysis relied on existing records, the researchers worked with whatever had been captured, rather than a protocol designed from the start to answer this question. And a five minute window from a single electrocardiogram channel is a brief, narrow slice Of a person's autonomic life. Five minutes is enough to compute standard short term heart rate variability measures, but it is not enough to see the slow rhythms that unfold over an hour or a day, or to see how the nervous system responds to a series of demands and recoveries. We should read this study as a proof of principle, not as a definitive account of what these people's autonomic function actually is. And once again, this is observational and associational. We are seeing that reduced heart rate variability is accompanied by normal brain recordings in this group. We are not establishing that anything caused anything. There could be selection effects. The sorts of people who end up getting a brain recording in the first place are already, on average, dealing with some sort of concern.
[00:20:00] And those effects may travel with the results. But the idea at the center of this study is one I keep turning over. It is the idea that a test can be normal on the dimension it was designed to measure, and still be quietly carrying information on another dimension entirely, if only we bother to read it. There's a kind of humility in that, a reminder that the body is always saying more than we are asking.
[00:20:19] And there is also, I think, a practical reminder. Much of what medicine currently measures is already being paired invisibly with a heartbeat signal. If we get in the habit of listening to that signal too, we may find that a great deal of clinical data has been waiting to speak to us all along. And that feels like the right place to pause for just a moment before we turn to the more critical medical situations that make up the second half of today's research.
[00:20:40] This show is brought to you by Optimal hrv. And I want to take a minute to tell you why I actually use it, and not just because they make this program possible.
[00:20:48] Everything we talk about on this show, the parasympathetic branch that calms you, the autonomic flexibility that lets you meet stress and then recover from it, the slow drift of these numbers over months and years, all of that only becomes real in your own life, when you can actually see it. It is one thing to know in the abstract that the nervous system has these branches and that they respond to how you live. It is another thing entirely to watch your own numbers move in response to a hard week or a stretch of consistent sleep, or the reintroduction of a practice you had let slip. That direct feedback is what turns knowledge into insight and what makes this measurement useful rather than merely interesting.
[00:21:22] And that is what optimal HRV is designed to do. The foundation is a simple morning measurement. You Take a short reading not long after you wake, at the same time each day in the same way, so that you are comparing like with like. That consistency is what turns a single number into a signal. One morning reading on its own tells you very little. The same reading taken faithfully across weeks begins to tell you the story of your own nervous system, how you are absorbing training, how you're handling stress, how you are recovering from illness or a hard stretch of life. Optimal HRV is built around that longitudinal tracking so that instead of chasing a good score on a good day, you get to watch your own trend line and learn what actually moves it for you. And I want to underline that framing because it connects back to today's research. Every study we discussed today in one way or another was arguing that a single reading matters less than a pattern. The cardiovascular study followed people for more than two decades. The multiple Sclerosis study examined how heart rate variability changed during a task, not just at rest. The brain injury study cared about the coupling of two signals over time. A single measurement is a data point. The measurement done consistently over weeks, months and seasons is a story. Optimal HRV is designed for that story and it does not stop at measurement. The app includes biofeedback tools, guided paced breathing that helps you find the breathing rhythm at which your heart rate variability responds most strongly. So you're not just observing your autonomic nervous system, but gently training it. This is not exotic technology, it is a way of putting a well studied practice in your hands with feedback that lets you learn what actually works for your own body. That combination honest measurement paired with practical tools to act on what you see is exactly the spirit of this show. If you would like to bring what we discuss here into your own daily practice, everything you need is in the show notes or you can go straight to WB www.optimalhrv.com Alright, let us get back to the research into the most fragile patients we will discuss today. The second half of this episode carries a different weight than the first in the studies we are about to discuss. Heart rate variability is not being measured on a person going about their ordinary life. It is being measured in a hospital bed, in a treatment room in the presence of serious illness. And that changes what the measurements can mean and what our responsibility is and how we speak about them.
[00:23:29] Our fourth study takes us into the neurointensive care unit and it is for me one of the most striking pieces of research in this whole episode. It comes from Valentin Gabaudo, Gwendane Percival, Samuel Garcia, Hugo Ardellan, Nathalie Bonviso, Clement Minouet and Baptiste Balanza, a team working across several research and hospital institutions in France. And it is about something so basic that most of us have never once thought about it. You know the way your heartbeat is coupled to your breathing.
[00:23:54] Here's the thing your body is doing right now without any effort from you. As you breathe in, your heart rate tends to speed up a little. As you breathe out, it tends to slow down. This gentle rhythmic rise and fall of heart rate in time with the breath is called respiratory sinus arrhythmia. And the researchers refer to the beat to beat heart rate variation tied to breathing simply as respiratory heart rate variability. It is one of the clearest, most elegant expressions of a healthy autonomic nervous system. It reflects a whole coordinated system, the brain stem sensing the breath and and adjusting the heart from moment to moment through the vagus nerve. When that system is working, the coupling is strong and timed correctly, with the heart speeding on the in breath. What this team did was compare that coupling in 55 brain injured patients being cared for in a neurointensive care unit, the intensive care unit or ICU dedicated to patients with neurological injury, against 31 healthy volunteers. And the difference they found was not subtle. In healthy volunteers, the amplitude of respiratory heart rate variability, essentially how much the heart rate varied with each breath had a median of around six beats per minute. In the brain injured patients, the median amplitude had collapsed to roughly one beat per minute. The coupling between breath and heartbeat, so robust in a healthy person, had very nearly flattened out in the injured brain. And then there is the detail that genuinely stopped me. In many of these patients, the coupling had not merely weakened, it had inverted. Recall that in a healthy person, heart rate peaks during the in breath. In brain injured patients, the peak heart rate tended to occur during expiration. Out breath, the rhythm had reversed its phase. That is a profound sign because it suggests that the normal central control of this coupling, the brainstem circuitry that orchestrates the timing, has been disrupted at a deep level. The system is not just turned down, it is playing the music backward. I keep coming back to that image. Think of two dancers who have practiced together for years whose steps interlock without either of them thinking about it. That is what your breath and your heartbeat are doing right now. Now imagine that partnership breaks so completely that one dancer starts stepping when the other is meant to pause. The rhythm is not just weaken it. It has become internally contradictory. That is what the reversal in these patients represents. And it is the kind of finding that stays with you. The researchers also examined what else was associated with lower respiratory heart rate variability, and two findings stand out as immediately physiologically plausible. Patients whose breathing was being fully driven by a machine, those on mechanical ventilation, where the ventilator triggers each breath rather than the patient, showed reduced amplitude, which fits because so much of this coupling depends on the patient's own respiratory drive engaging with the cardiac system. And deeper levels of sedation were also associated with reduced amplitude, which again fits because the medications that quiet an agitated or injured brain also quiet the very circuits that generate this coupling. The author suggests that respiratory heart rate variability could serve as a physiological marker to help track autonomic and neurological recovery in these patients. A way of listening at the bedside for signs that the brainstem's coordination is returning. We do need to hold this carefully. This is a comparison between a group of very sick patients and a group of healthy volunteers. And and the two groups differ in numerous ways beyond the brain injury itself, including, as the study highlights, sedation and mechanical ventilation, which are part of caring for critically ill people. So we cannot cleanly separate the effect of the injury from that of the treatment, and we are again in associational territory rather than in the realm of causal proof. It is entirely possible that some portion of the flattened inverted coupling reflects the medications rather than the injury itself. The study's own analyses suggest that both contribute the sample is modest in size, as studies in this extraordinarily difficult setting almost always are, and measuring this kind of coupling in an intensive care unit amid all the machinery and medication is genuinely hard. There are practical problems of signal quality, of coordinating with clinical teams, of getting comparable healthy control data. Any team that pulls this off deserves respect for the discipline alone, but as a demonstration that the coupling of breath and heartbeat can flatten and even reverse when the brain is injured, and that this might give us a bedside signal of how the nervous system is faring. It is a beautiful and sobering piece of work. It takes something most of us have never noticed about our own bodies and shows us its value precisely at the moment it is lost.
[00:27:30] If you have never sat with the fact that your breath and your heartbeat are gently coordinated together right now, without your asking, let this study be the invitation. It is one of the small miracles of a well functioning nervous system, and this research helps us see what its absence looks like. Our fifth study widens the frame from a single measurement to an entire philosophy of measurement. It comes from Shi Gong Yun Chu and Xiaoting Wang, and it is a review article rather than A new experiment, which means that rather than reporting fresh data from a specific group of patients, the authors have gathered and synthesized the existing evidence to make an argument. Their subject is autonomic nervous system dysfunction in critically ill cancer patients. And their argument is one I find genuinely persuasive that when the body is this sick, no single instrument is enough to understand what the autonomic nervous system is doing. Let me set the stage they are describing. A person with cancer who becomes critically ill is under assault from several directions at once. The malignancy itself can disturb the body's regulation. The treatments we use against cancer, chemotherapy, radiotherapy and the newer immunotherapies can directly damage the autonomic nervous system and can trigger widespread inflammation throughout the body. The authors place all of this within a framework they call the host organ unregulated response. And the picture that emerges is of a person whose self regulatory systems are being pulled in multiple directions simultaneously. The autonomic nervous system, abbreviated ANS and comprising the sympathetic branch that mobilizes the body and the parasympathetic branch that restores it, sits at the center of that dysregulation. And when it falters, the consequences manifest as hemodynamic instability, my unstable blood pressure and circulation and organ failure. So how do we see it? This is where the review makes its central move. The authors argue that we should assess the autonomic nervous system through several complementary windows, rather than just one. Heart rate variability is one such window, a valuable one, offering both time domain and frequency domain views of autonomic function. But they place it alongside blood pressure variability bpv, which examines beat to beat and longer term fluctuations in blood pressure. They add baroreflex sensitivity, abbreviated brs, which measures how well the reflex that stabilizes blood pressure is actually working. That reflex being the loop by which a rise in blood pressure prompts the nervous system to slow the heart and lower the pressure again, and its sensitivity being how briskly and how completely it responds. They advocate for what is called multimodal monitoring, abbreviated as mmm, in which several high resolution physiological signals are captured and synchronized so that you are watching the system as a whole rather than as a collection of separate readings. And they point to the growing role of artificial intelligence, abbreviated AI, in making sense of these complex layered signals, using machine learning to find patterns and to help stratify which patients are at greatest risk of the next crisis. Which is precisely the kind of question a busy intensive care team most needs answered. The reason this resonates with me is that it speaks directly to a temptation we all share, and one that I have to watch in myself. Precisely because this show is named for a single metric, it is tempting to treat heart rate variability as the answer, as the one number that captures autonomic health. But heart rate variability is one voice in a choir. In a stable, healthy person, that one voice may tell you most of what you need. In a person who is critically ill with cancer, whose regulation is being disrupted from many directions, you want the whole choir. The authors even discuss how autonomic dysfunction manifests across different organ systems cardiac, pulmonary, cerebral, renal and gastrointestinal and sketch management approaches, including a strategy they abbreviate as RemyPro DE centered on goal directed sedation, analgesia and calming and overactive sympathetic response. Because this is a review and not a primary study, its limitations are of a particular kind. A review reflects the choices of its authors, which studies they gathered, how they weighed them, where they placed their emphasis. It does not generate new experimental evidence, and it cannot on its own establish cause and effect. It inherits the strengths and the weaknesses of the underlying literature. It summarizes.
[00:31:04] Different authors working from the same body of evidence could construct a somewhat different synthesis with different accents. And a framework as ambitious as multimodal monitoring paired with artificial intelligence is easier to advocate for than to implement at every bedside. Given the cost, the complexity and the expertise it demands. Machines that can synchronize multiple physiological signals in real time exist, but they are not universally available. Algorithms that can extract meaning from those signals in real time are advancing quickly, but they are also opaque in ways that clinicians rightly find uncomfortable. So the vision the authors sketch is aspirational and the path to it will be uneven. But as a corrective to single metric thinking and as a map of where the assessment of the autonomic nervous system in critical illness may be heading, it earns its place in this episode. It is the study that reminds us to stay humble about our favorite number, and it does something I appreciate. It treats the sickest patients not as edge cases at the margins of research, but as the very people whose care most demands sophistication in our measurement and in our thinking. Our sixth and final study brings us to a chronic disease and to a symptom that is easy to underestimate if you have never lived with it. It comes from Guadalupe Garris, Karina Sanders Sandersfeld, Andrea Hildebrandt and Helmut Hildebrandt and asked whether heart rate variability can serve as a biomarker of autonomic activity, reactivity and fatigue in people with multiple sclerosis. Multiple sclerosis, which I will refer to in full, and which is a disease in which the immune system attacks. The protective covering of nerve fibers in the central nervous system brings many challenges, but fatigue is one of the most common and most disabling, and it is a fatigue that is not simply about being sleepy.
[00:32:25] It is a deep depletion, a sense that the reserves are gone and it does not reliably respond to rest.
[00:32:30] The researchers pulled data from two clinical trials giving them a combined group of 78 participants, and they applied a statistical technique called exploratory factor analysis, abbreviated efa, to a set of heart rate variability parameters. Exploratory factor analysis is a way of asking whether a large set of measurements is driven by a smaller number of underlying dimensions, and what they found was that their heart rate variability parameters organized themselves into two factors. The first factor was composed primarily of parasympathetic measures, the standard deviation of the long term variability from POINSARERE analysis labeled SD2, together with the standard deviation of the normal to normal intervals. Again, the SDNN we met earlier, reflecting the calming restorative parasympathetic branch abbreviated pns. The second factor was made up primarily of measures tied to the sympathetic branch abbreviated sns, including the ratio of short term to long term poincheur variability written as SD1 over SD2 and the ratio of low frequency to high frequency power written as the LF over HF ratio, where low frequency and high frequency refer to those rhythmic components of the variability signal. So the first thing this study offers is a kind of tidy structure. The many faces of heart rate variability in these patients can be understood as two underlying dimensions, one leaning toward parasympathetic and one toward sympathetic. But the researchers went further because these two trials also involved biopsychological interventions, allowing them to examine how different practices affected these factors. A practice they call self alert training was associated with decreases across heart rate variability parameters consistent with sympathetic activation. A revving up deep breathing was associated with an increase in heart rate variability, particularly in the sympathetic related parameters and progressive muscle relaxation. The practice of systematically tensing and releasing muscle groups fell somewhere in between, with intermediate effects on both factors. Because participants took part in these practices. This part of the study carries an interventional flavor rather than being purely observational, though the numbers are small and the authors frame it as exploratory. Then comes the part that speaks most directly to the lived experience of fatigue. The researchers distinguish between trait fatigue, a person's general enduring level of fatigue, and state fatigue, the fluctuating fatigue you feel at a particular moment or during a particular task. That distinction matters because the two do not always move together. Someone can feel relatively alert at a specific moment, yet be deeply worn down in daily life and vice versa. Different mechanisms may drive each and they found something specific and interesting. When people performed a sustained attention task, those who showed a smaller change in the standard deviation of the normal to normal intervals during that task tended to report higher trait fatigue. In other words, a nervous system that flexed less that did not adjust its variability as much as the demand for vigilance rose was associated with more enduring fatigue. For momentary state fatigue, no significant associations with heart rate variability were found, which is worth noting. Heart rate variability may be tracking something about the underlying regulatory capacity of the nervous system more than the moment to moment felt experience of tiredness. The larger picture the author sketches that reduced autonomic adaptability, a diminished capacity to shift between the parasympathetic and sympathetic branches as circumstances demand, reflects the kind of chronic fatigue that characterizes multiple sclerosis and that higher disability tended to track with a reduced high frequency heart rate variability response. The limitations here matter, and the authors are appropriately measured 78 participants pooled from two separate trials is a modest sample, and pooling trials that were not originally designed to be combined introduces its own complications. The two source studies may have differed in their inclusion criteria, the specifics of the interventions, the methods used to measure heart rate variability, and the questionnaires used to assess fatigue. Combining them increases statistical power, but it also reduces precision. Exploratory factor analysis is, as its name says, exploratory. It generates structure and hypotheses rather than confirming them, and a different sample might resolve into somewhat different factors. It is important not to reify the two factor structure the analysis produced. It is a useful frame, but it is one lens among possible lenses. And while the intervention findings are intriguing, they are based on small sample sizes and should be read as early signals rather than settled conclusions if you are living with multiple sclerosis, this is not a study that hands you a prescription. It is a study that hands you a possibility worth exploring alongside the clinicians who know you. But the central idea is a powerful one, and it closes our medical week on exactly the right note. And it is the idea that fatigue may be in part a failure of flexibility, that what wears a person down is not only a low resting level of anything, but a nervous system that has lost some of its ability to rise to a demand and then recover from it, that reframing matters for anyone who works with chronic illness or lives with it, because it moves fatigue out of the category of a static deficit and into the category of a dynamic failure of regulation, which is one we might actually be able to change. So let us step back, because the point of gathering these six studies under one theme was was never to collect six separate facts. It was to let them speak to one another. And when they do, I think three ideas emerge that are larger than any single paper. The first idea is that the meaning of heart rate variability depends profoundly on context. We saw it most explicitly in the long Cardiovascular study, where the same index conveyed a different message depending on whether a person had normal blood sugar, pre diabetes, or diabetes, where greater variability was not uniformly better, and the relationship could even bend into an S shape. But the theme runs through everything today. In the glucose study, an age correction that should have helped didn't, because the confounding variable everyone assumed was present simply wasn't. In the data. In the brain injury study, the very same coupling of breath and heart that is a sign of health became, by its collapse and its reversal, a sign of profound disruption. In the multiple sclerosis study, the same measure could indicate either parasympathetic dominance, reduced adaptability depending on how it changed during a task. Heart rate variability is not a thermometer where higher or lower means the same thing everywhere. It is more like a word whose meaning shifts depending on the sentence it appears in. And that should make us humble every time we are tempted to read a single number as a verdict. If you take one habit away from today's episode, let it be this. Whenever you look at a heart rate variability, reading your own or someone you work with, pause and ask, what is the context of this person's life and health? And that gives this number its meaning. The number alone will always be misleading. The number in its rightful context is where understanding begins. The second idea is that heart rate variability is a window onto central control, not just peripheral fitness. It is easy to think of heart rate variability as a measure of how fit or relaxed your heart is. But study after study today pointed upstream toward the brain and brainstem. The electroencephalogram study found reduced autonomic flexibility hiding behind normal brain recordings. The Neurointensive Care study showed that the coupling between breathing and heartbeat flow flattened and reversed when the brain stem's coordination was disrupted. The Multiple Sclerosis study identified fatigue as a loss of adaptability, a failure of central switching between branches of the autonomic nervous system. The Critical Care Review put The autonomic nervous system at the center of a whole body dysregulation orchestrated by the same brain and brain stem circuits. Again and again, the heart was telling us about the brain. That is worth remembering. When we measure heart rate variability, we are, in a real sense listening to the nervous system's capacity to regulate the whole body from the top down. That reframing has practical implications, too. If heart rate variability reflects central control, then practices that train central control, attention, breathing rhythms, the deliberate cultivation of calm and of engagement are not fringe additions to health. They're working directly on the mechanism that the measurement reflects. And the third idea, which the Cancer Review stated outright, but which every study quietly reinforced, is is that heart rate variability is one voice in a choir and we should resist the temptation to treat it as a soloist. The Cancer review argued for measuring blood pressure variability, baroreflex sensitivity, and multimodal signals alongside heart rate variability precisely because in a critically ill person, no single window is enough. But you can hear the same lesson in the glucose study's insistence on honest validation, in the cardiovascular study's caution that its most sophisticated measures exceed what a clinic can currently do, and in the electroencephalogram study's discovery that a normal test can still hide something the heart knew all along. Even the multiple sclerosis study, by reducing many heart rate variability parameters to two underlying dimensions, made a similar point. No single parameter is the answer, and combinations viewed together tell a richer story than any one variable alone. The honest posture across all of it is the same. Heart rate variability tells us something real and something valuable, and it tells us most when we hold it alongside everything else we know about the person in front of us.
[00:40:25] That is a posture that will serve you well whether you are a clinician, a researcher, a coach, or someone tracking your own health at home. Do not chase a single number. Build a picture over time alongside other things you know, and let heart rate variability contribute one important thread to the larger cloth. That is what I hope you take away from this medical week. Not that heart rate variability is more powerful than we thought, and not that it is less, but that it is more contextual, more central and more collaborative than a single score on a single morning could ever capture.
[00:40:52] The number matters, but the number in context, the number read alongside the whole person, is where the truth actually lives. I want to add one more reflection before we sign off, because a theme has been developing across this new format that I do not want to let pass by when we grouped studies together by whatever happened to have been published in a given week. The studies rarely spoke to each other. There was breadth and there was novelty, but there was not always a conversation. Since we started organizing episodes around themes, I keep noticing that the studies begin to answer each other in unexpected ways. Today, the review article on critical illness and cancer felt like it was quietly commenting on every other study in the episode, arguing in effect that no single measurement can carry the weight we sometimes put on it. And the multiple sclerosis study in reducing many parameters to two dimensions was making a related methodological point from a very different clinical context. When you gather the research this way, patterns emerge that no single paper could have shown you. That is the promise of the new format, and I am grateful to everyone who has written in to say it is working for you. Keep the feedback coming. It shapes how we do this show. Thank you as always, for spending this time with me and for thinking carefully rather than quickly. Careful thinking about difficult things is, in the end, what this show is about. We are trying to make honest sense of a growing pile of research on a signal that is neither as simple as some of its enthusiasts suggest, nor as opaque as some of its critics fear. The Heartbeat has been telling its story to anyone who would listen for as long as we have had hearts. We are just now in the position of being able to hear it a little better than before and to test our impressions of it again against the discipline of good measurement. If today's episode was useful to you, the studies we discussed are all linked in the show notes if you want to bring this practice into your own life, you will find optimal HRV there as well, or at www.optimalhrv.com. if someone in your life is dealing with any of the medical situations we cover today diabetes, cardiovascular disease, a chronic condition like multiple sclerosis, or the aftermath of critical illness, the guidance stays the same as always. None of what we have discussed is a substitute for the clinicians who know that person. What we have offered is context, so that if the topic comes up, you might have a slightly richer sense of what the research says and what it does not. Take good care of yourselves, take good care of the people who depend on you, and I'll see you next week when we gather around a new question and do this all again. I'm Matt. This has been this week in heart rate variability and.
