 Good afternoon. I'm delighted to welcome you today to the McLean Center series on ethical issues and end-of-life care. Next Wednesday, let me mention that Professor Dan Brudney from the Philosophy Department. Is Dan in the audience? Dan, where's Dan? Oh, there. We'll be speaking on the topic of democratic legitimacy and end-of-life decisions. Now, it's important that you note that our talk next week will not be in this room, but will be in room H103, so that there will be a change in room. We'll have a sign-up about that, but if you can just note it. Today, it's my pleasure to introduce our speaker, Dr. Bill Meadow. Bill is a professor of pediatrics and the director of the Neonatology Fellowship Program here at the university. Bill earned his MD at the University of Pennsylvania, completed residencies at the University of Chicago and Children's Memorial, and then did a fellowship in infectious disease and neonatology here at the university and then a fellowship in clinical medical ethics at the McLean Center. Dr. Meadow's research focuses on medical ethics in the field of neonatal epidemiology. He has examined issues of resource allocation and informed consent in the NICU. Bill has also been a major contributor, maybe the major contributor to our knowledge about prognostication for children in the NICU setting. His crystal ball study, which has worked both in the pediatric neonatal unit and in the medical ICU, I think with Jesse, collaboration has been a groundbreaking breaker in terms of prognostication for patients who are hospitalized. Bill has published many articles and book chapters and the book that he co-authored with Dr. John Lantos, a book entitled Neonatal Bioethics, The Moral Challenges of Medical Innovation is a classic in the field. Today, Bill will speak to us on the mathematics of morality in the NICU. Please join me in giving a warm welcome to Bill Meadow. Right, where's the monitor? I can put out another 20 bucks, try to win my 20 bucks back that I lost this morning in talking to the fellows. Anyway, the title of the talk is the Mathematics of Morality. This is your standard ethics thing that nobody pays, ethicists anything. And I won't be talking about meds. This slide often appears at the end of talks. I like to put it at the beginning of talks. These are many of the people who've helped me in my life in Chicago, which is a very strange place, but one of the things that it does do well is that it has great people to help. These are three people who are not at the University of Chicago in Jean Vieille. He's a neonatologist in Montreal. Joanne Legada is at the Medical College of Wisconsin. She was here both as a resident as a fellow. And John Lantos, as Sieg's mentioned, is, I think, without question, unless, Laney in the audience? So if Laney's not in the audience, I'll say that John Lantos is unquestionably the most evident pediatric ethicist in the world today. And we'll go from there. If suddenly you were forced to think hard about extremely premature infants, what would you want to know? What would you want to know if you were a doctor, a health policymaker, or a parent? Well, if you were a doctor, you might want to know how frequently babies like this were born, how frequently they lived and died, how long they stayed in hospital, how they did if they survived. These are the sort of standard things that are written about in medical literature in terms of neonatal medical outcomes. Policymakers have two other issues that they talk about. One is how much do these babies cost? And the second is how do these expenses compare to other public health expenses, either in children or in adults? Parents only want to know one thing. They want to know what will happen to my baby, not 100 more or less similar babies, but what will happen to my baby. And if you were a parent, when would you want to know that answer? Well, there are a number of different times. We're going to talk about this in somewhat more detail, but as a general principle, you might want to know before birth and there's a bunch of people in the audience who do this for a living, both as neonatologists and neonatology fellows, and we'll talk about the anti-natal consults that we give. You might want to know in the delivery room in the context of whether or not you should begin or continue neonatal resuscitation. What about after several days in the NICU with what you can learn from a trial of therapy? And finally, at the time of NICU discharge, we'll talk a bit about the pluses and the minuses in terms of counseling at the time of NICU discharge. So here's my central claim. It's a claim that I brought up a little bit when Joel Freider, who many of you were here last week, talked on pediatric end-of-life care. It is, I don't know if it's an outrageous claim, but certainly a claim that I strongly believe. And we talked a bit about it this morning. I don't think, don't tell Joel this, I don't think there's anything interesting about pediatric end-of-life care. And what I mean by that is that I think it's absolutely identical to adult end-of-life care. But I think neonatal end-of-life care is markedly different. And the reason that I think they're different is easy to articulate. I think that the most salient decision in end-of-life decision-making, both for adults and for older children is, are you going to go back to the hospital to get more care, or are you going to stay at home or a nursing home and pass away comfortably without medical intervention? That to me is the single decision that's most important in end-of-life or hospice or whatever you want to call decision-making. And that's the same in older children as adults, but it's not the same in the NICU. Because in the NICU, you've never been home. In the NICU, you're always there. And so we in the NICU simultaneously give supportive care, medical interventions, ventilation, TPN, all this stuff, and hospice care when it's appropriate. And that, I think, accounts for the fact that when you take all these surveys about end-of-life care and hospice and all this stuff, most physicians in most specialties think that they're under-trained, under-exposed, they don't quite know what to do, they could use help. There are hospice services and palliative care teams, and Joel runs the one at Lurie, and there are people here who do it. And when you survey neonatologists, they just don't care. They're not interested in getting help from the outside for end-of-life care. We almost never call end-of-life consults. We don't care because we think we don't need it. It's what we do for a living. We have 50 deaths a year in the NICU, many more deaths than there are, certainly in all of the rest of Comer. Now, there may be Comer patients who die at home. In fact, many of them, in theory, should die at home. But we see death a lot, and we get used to managing it, and there aren't quite the same kinds of concerns. Now, I'm going to make a lot of fun of Jesse over the course of this talk, but it's easy, he's right there. He runs our NICU, and I'm going to talk about a number of different occasions about distinctions between how the NICU, neonatal intensive care unit, and the NICU, which is the medical intensive care unit, which is for adults. I used to, when I was younger, be much more cavalier in making fun of the NICU, because as I'm getting older and could keel over at any moment, I'm much less comfortable mocking everything that Jesse has done for his entire career because I may need him really soon. Okay, so in our world, end-of-life is the other point to be made, and it's an important point is that NICU end-of-life care is different from all other end-of-life cares because in our world, the end-of-life is not the outcome to be most feared, either by docs or by parents. Rather, devastated survivors are generally considered to be the worst possible outcome. So John Lentos had a lovely poignant phrase which is that before the NICU, before neonatology, all of CP was God's fault. After neonatology, half of CP is our fault. And the single most powerful emotion in the field of neonatology without question is guilt. That is the emotion that we deal with every single day we're on service. And the reason we deal with it is because, unlike in adults where we pediatricians sometimes sit in the back, I try not to tell us the seags and try to hide it from the fellows until they get longer into their months. Okay, but sometimes we pay like little bingo in the back when they hand out the cases, right? And if the first line of the case is 88-year-old patients, right? We're done, right? There's nothing else interesting about that case from our perspective. Now, I know you're not supposed to say it, but in fact it is how we often feel because these patients for the most part are going to die. The vast majority of our ethics consults are about end-of-life decision-making and in our world it just doesn't matter who decides whether they die this week or next week. Nick Christakis, who used to be here and then was deemed, hardest it is to believe, too full of himself to be at the University of Chicago. So he went to the only other institution in America where you can be even more full of yourself and comfortable, and that would be Harvard. And here I used to teach lecture together in seags's course to the first-day medical students where we would talk about using the D-word, right? Because Nick is famous for saying the physicians are imperfect at counseling patients and telling them they're going to die, that they don't use death and dying frequently enough and that physicians are imperfect at predicting death. And they usually overestimate how long patients are going to live and stuff like that. And I remember vividly a terrific scene where a first-year med student, it was a brilliant question, just interrupted him right in the middle of his talk and said, so do you think it's good or bad for the patients that doctors are overly optimistic? And he was taken aback. He said, you know I've never thought about that. And I thought it was a terrific question by this medical student because it got to the heart. Now I didn't make fun of him then, I didn't make fun of him at many other times, that you see in our neonatology world all of these adult issues pretty much don't matter because the patients are all going to be dead in six months. Whereas for us they matter because these patients may be alive for decades. And so assessing that is our biggest end-of-life issue. And with that we're going to press forward and talk a little bit more about data. There are these four distinct data-related issues as they impact the morality of NICU care. There's money, there's outcomes, there's prediction, and there's worth. And today, ready for this, is this like, huh, is that like awesome PowerPoint skill? You like this? First it's all dark? Then huh? I know how to do this. I remember the Cahokia Indians crossing over the, that's another story, I'll tell you later. Okay, anyway, today I'm only going to talk to you about money and prediction. Another time perhaps if you'd like, we can talk about outcomes and moral worth of patients. So I'm going to talk about money and prediction both in the NICU and in the adult ICU and give you a feel of both of the kinds of data that we gather to give you a feel for how you can talk about medical ethics from a database relationship. That's the kind of thing that youngsters might learn and want to do for field to give you a perspective. Okay, so I'm going to talk about money first. Here's the numbers for perspective. We're going to split the room right down the middle. CEEX, you can count the numbers. You guys can be one team. You guys can be another team. This is audience participation. I learned from my children's kindergarten classes you can't be like team A and team B because you feel bad or team one and two. So you guys are the sons. You guys are the moons. Ready? Go. In the United States every year, how many babies are born? Oh, all right. So the moons, one. CEEX, pay attention. You don't know the answer to anything. So four million is the answer. What is it? What is it? It is a bad question. Okay. How many people die in the U.S. every year? Good. Oh, we have some ringers here. These health study types. But you're right in the middle, so I don't know how to count you. Okay. Right. So I counted as three million, but between two and a half and three million people. Okay. Die every year. Now, how many babies are born under a thousand grams? Good. Oh, Mike Misali. The moons are stacked here. Well done, Mike. Okay. So I counted a little less than that, Mike. I counted 40,000. But it's about that. It's about 1% of all the babies born. And there's really about 4.2 million born. So it's a little more than that. Okay. Now, they get a little harder. How many babies die in the U.S. every year? Not eventually, right? Eventually we know. But during the neonatal period. Right. Yeah. You should know this. Who said that? Well done, sons. So that's about right. So it's about 0.6%. So it's about 24,000 babies die every year. Next question. How many babies under a thousand grams die? Now, there are two ways to get the answer to this. One is, of all the babies who die, 24,000 or so, how many die from complications of prematurity? Okay. There are only two reasons in America that babies die. What are they? Trauma. No, not trauma. See, just stop. Just keep counting. You're not going to get any of these answers. Right. So one is complications of prematurity. And the other is genetics, congenital anomalies. Mike is like a ringer planted in this talk. This is great. Yes, congenital anomalies. Okay. Here's an interesting question. Over the years. So first of all, how many die, roughly what's the percentage of die from prematurity versus congenital anomalies? Yes, it's about 50-50, which is surprising. Most people think there's many more for premies, but it's not. Over the years, how do you think that ratio has changed? More. Yeah, you might think so. But in fact, the ratio has been the same for about 40 years. So we're better caring for premies, but also fewer children with congenital anomalies die. Why is that? So you might think we're better at taking care of them with surgery and ECMO, but that's not the big reason. Right. The big reason is anti-natal termination. Okay. So those numbers have stayed the same for about 40 years. So babies under 1,000 grams die. Well, one way to say it is about half of all the babies who die die of prematurity. So it's about 12,000 or a third of all the babies who were born under 1,000 grams. So here's another question. This is like double round here. What percentage of NICU resources are devoted to ELBW that's extremely low birth weight, so that's under 1,000 grams, non-survivors as opposed to resources devoted to ELBW babies who survived the discharge? So babies are born over 1,000 grams. There are 1% of 4 million, so there's 40,000. Some of them die. Some of them live. We spend money on both. Of all the money we spend on those babies, what percent goes to babies who are non-survivors? 10 cents of every dollar, 50 cents of every dollar, or 90 cents of every dollar on non-survivors? Yeah. So very good. 10 cents of every dollar. On this slide, these data from our institution, most of the data I'm going to show you from our place. I'll tell you when they're not. So this is birth weight on the x-axis, mortality on the y-axis. The gray bars are percentage of babies who die, but the blue bars are the percentage of dollars spent on non-survivors. And you can see that at the high birth weights where mortality is quite low, well obviously we don't spend much money on dying babies. But look over here. Even when mortality is high, we spend very little money on dying babies proportionately. How can that possibly be? Yes. Because dying babies die quickly. The smallest and the sickest die the quickest. And survivors stay in the NICU a long, long time. So neonatal intensive care from that perspective is well apportioned to survivors as opposed to non-survivors. And now this is the first time of several that I can make fun of Jesse. Okay, so we did the same study in our adult ICU. And here are the data. On the top of the NICU data, split into high-risk, medium-risk, low-risk, and the bottom of our adult data, high-risk, medium-risk, low-risk. So look at the babies. Even the highest-risk babies where 70% will die, 10% of the dollars are spent on non-survivors. And in every risk group, the blue bars are lower than the gray bars. But look at the adults. At every risk group, you spend relatively more money on dying adults than non-dying adults. And by the highest risk group, 70 cents of every dollar spent in the adult NICU on the highest risk patients are spent on patients who are going to die before they get out of this hospital. That's an important distinction between NICU care and adult care from a financial perspective. Here's a slide I like to show. This is part of my Rambo 1 escape slide. So this is the likelihood of survival with every passing day for NICU patients versus adult patients. The NICU is in gray. So even if you take a high-risk group where on Day of Life 1 only 25% will survive with every passing gray, look at the gray bars. If you don't die in the NICU, with every day, you're more and more likely to go home. Now look at the blue bars. Those are Jesse's unit. I don't want to personalize it, okay, every day you don't get out of the ICU, you're less and less likely to ever get out of the ICU. So here's the summary about money. We spend 10 times more on dying NICU patients than dying NICU patients. And there are 100 times more dying adults than dying infants. We already decided that 2.5 million people died, 25,000 of them are babies, and the other 2.4 something million are adults. Yes, we spend relatively more on dying adults because it's disproportionately more than dying ICU adults than dying, than living ICU adults. Yes, yep, yep. So there are no credible financial arguments against neonatal intensive care. So if you're ever at a cocktail party and people say, oh, you spend way too much money on those babies, okay, just stick to your guns. JD, if you should like disall the people at the cocktail party and just mock them and say, if ICU costs savings are to be desired, they should be found at adult ICUs, not in NICUs, right? There is one politician in America who has embraced this theory vividly. Yeah, anybody know who it is? It was a governor of Colorado about 20, 30 years ago, and actually his name escapes me now, but I'll think of it. Yes, that's right. Yeah, you knew this? Yeah, it's true. And so he said this, and he was ex-governor by the next election because premies don't vote and old people do. And so it turns out you can make some quick calculations and David Meltzer has done this among many others. So what percentage of adults are hospitalized in their ICU in their last year of life? That's high, 70% is high. Some people say 20% or 30% of adults will go to their ICU, but 50% will go to some hospital and about 20% or 30% will go to, okay. And so if we could cut the costs even of that subpopulation, okay, by attempting to predict who's going to die successfully, we could save a lot of money for healthcare costs. And so the next question that arises is, can we predict successfully? And this we're going to talk about now. Okay. I'm going to talk about two things in predicting outcomes, timing and positive predictive value. I'm not going to spend forever on this. Just very quickly for some of the young ones, positive predictive value is a forward-looking statistic. I stand here in the present. I make a prediction that I walk to the future. I turn around and I say, was my prediction right? How many times out of 100 was my prediction right? That's different from sensitivity, which is often found in the medical literature, which I'm not going to talk about today. And timing obviously is when it matters. So stick with me. If you looked at your predictive value data and you only started making predictions when patients were severely braided, cardiac, hypoxic and had no blood pressure, your positive predictive value would be excellent, right? But probably ethically irrelevant. And so what we're trying to do is be a little better than that. Okay. This is a terrific slide, Joanne Lagada invented this slide. I'm going to spend a little time on it. This is the outcome prediction in the urethalgy proposed timeline. There are at least four different times when you can talk to parents or other physicians and decide and discuss what the likely outcome is going to be. The first one is anti-natally. This is what my fellows do. There's some fellows here today. And what they do is they are called by the obstetricians to please go... I'm just going to grab a little water here and ask Brayton to go to Jesse's unit. If I'm nasty to him, he won't take me. Mmm. If I drink it fast, I'll ask Brayton. Okay. Anyway, so anti-natally, you get called by the obstetricians, please come and talk to this mother. She's threatening premature delivery. And the big question that is asked for babies who are born around the threshold of viability is should we resuscitate them at all? The alternative is to not resuscitate them at which point they would die with comfort care. The question is, what do we know when we do anti-natal counseling? And these are sort of the four, there's actually one more, things that we use when we do anti-natal counseling. Gestational age is important, whether or not the mother got anti-natal corticosteroids, which mature the fetus. Twins, twins is a general principle, are more at risk than singletons. Small, undergrown babies do worse. And gender. So girls do better than boys at most gestations. Indeed, many women in the audience would say for the indefinite future, girls do better than boys. And I can certainly understand their argument. Okay. So we use these data and we'll talk a bit about how well these data can help us predict what's going to happen to the babies. Okay. Sometimes you can add a little information in the delivery room. Well, mostly what you add in the delivery room is how the baby looks has two parts to it. One is, can you confirm the gestational age is about what we expected? And the other part is, can you see if the baby looks vigorous or not? In the single best way we found to quantify how vigorous the baby is is by the AFGAR score, which you're all familiar with. It's a score of basically vigor of an infant. And so you get all the things that are down their gestational age and twins and stuff like that. And then you can think about how much physicians look at the AFGAR score around the time of deliveries to influence whether or not they should continue their resuscitation even though the data are that the AFGAR scores in the first few minutes are pretty useless as long-term predictors. The third time you could think about how to gather data to predict is in the NICU during what we call a trial of therapy. And here's an important point and it's important even when we compare our data to Jesse's data. In the NICU a trial of therapy can be considered to be any time from time one until the day before the discharge. But that's not what I'm talking about. I'm talking about while a baby is on a mechanical ventilator. And the reason that we say that is because then there's an ethical alternative. You could extubate them, withdraw the mechanical ventilation and the presumption is the best majority who are on a ventilator need to be on a ventilator if you stop the ventilation, they would pass. So redirecting to comfort care when a baby is on a ventilator. Once they're off a ventilator then you have a very much more a very different task. Because it becomes much more difficult to withhold other things. It's very difficult in our world to withhold feeding and nutrition because most even healthy babies get fed from the outside. It's not like adults. And so while a baby is on a ventilator can we learn something about how they're going to do? And there are at least a couple of ways to do it. What is an illness severity score? We use the SNAP score for neonatal acute physiology. It's actually that we inherited them from the adult ICU from the Apache scores. There's other things we can do. We'll talk a little bit about clinical intuitions and head ultrasounds as well. Finally there's the time of NICU discharge. Now at the time of NICU discharge you know a lot. Because you see they've been at the NICU the whole time. They're off the ventilator now. The vast majority of babies don't go on a ventilator so they're off the ventilator. So the problem is although you know a lot it's too late. You can't do anything other than refer them to Mike Massal because you're not allowed don't... And so in that context I didn't say that. But the idea is that in that context all you can do is refer them to early intervention. You can love them. You can hug them. You can support the parents in their difficult times but you don't have any alternative to keeping the babies alive. And so although the data are excellent it's too late from an ethical perspective. And so now what I'm going to say is that this time you like this? This is a square rectangle. 50% transparency. These are the kind of things you learn in talks. You learn PowerPoint and there's another 50% transparency. I'm going to say that these are too early and we don't know enough and this is too late and although you know more it's not helpful and that this is the time that we should be doing most of our counseling because we'll learn more and yet there's still alternative. So now I want to talk about predicting mortality while a baby is on a ventilator in the NICU this is because it's an IBM if it were a Mac it would say 1 and 2. This is what it was. Anyway there's algorithms and there's intuition. So an algorithm is easy. From the Apache you get points if you're sicker. So if you're hypotensive you get points. If you're sodium sucks you get points. If you're on cataclysms you get points. If you're on ventilator you get points. The sicker you are the more points you get. And so therefore it ought to be easy. We showed a long time ago and the Apache people showed too that at a time of entry to the NICU or to the NICU you're more likely to die than if you're less sick. That's not a shock. But let's press it further. Well sicker patients should get sicker and sicker and sicker before they die whereas the survivors should get healthier and healthier and healthier while they live. So not only should day of life 1 illness severity scores be useful in distinguishing two groups but serial illness severity scores are very specific. They ought to be the way we can distinguish babies who are going to live and babies who are going to die and so when you look at algorithms that's another slicky thing. Here's what we showed. We were completely wrong. When we found these data we couldn't believe it so we went back and looked again. Day of life is on the x-axis the y-axis is on the severity score. The red line is for non-survivors the gray line is for survivors the reds and the grays are clearly significantly different at the time of birth and over time the distinction narrows. It doesn't get wider. How can that possibly be right? And the answer is because all survivors stick with me, survive. And so the N of the gray bars is the same at every day of life but the sickest non-survivors die. And so the N on day of life 1 and the day of life 3 and the ones with the highest SNAP scores are gone and day of life 7 still. And so serial illness severity scores are useless in distinguishing babies who are going to die in the future. And that was a big surprise. This is a slide I just want to show you I get points for having made this graph. I made this graph in cricket graph it's really hard to do. Cricket graph is not supported by the way although it was a wonderful thing. Anyway the x's have to do with babies who are normal and who are impaired. And what you can see is very quickly by day 3 and day 5 there are as many x's up top at the highest scores as there are triangles and so the prognostication becomes less and less good. Well maybe I don't need no stinking algorithms maybe I just know this kid's going to live this kid's going to die maybe I know that. And what we do is we walk around that every day we identify all the babies in mechanical ventilation and we ask them the docs, the nurses, the NNPs the attendings, the fellows, the residents one question do you think this baby is going to survive to discharge or die in the NICU? For the methodologists in the crowd the most important methodological insight is candy. If you walk around with a bucket of candy you give them a piece of candy it's really cheap if you don't have candy the people won't come and they won't listen. Apparently I just learned this morning that the adult ICU people take more than one piece of candy for each I just learned that from my informants in the NICU whereas in the NICU they're more polite and they'll take only one piece of candy to give you information. So anyway, the point is that you get intuitions about what people think this kid's going to die if this kid's going to live. So here's a quiz question, this is for those of you just to give you a feel for how the NICU works what percentage of ventilated NICU babies under 1500 grams are never predicted to die? So you're sick enough to be in the NICU on event and yet nobody ever says you're going to die make a guess good, good guess, 50% so it's actually a little higher but it's not much higher so being on event in the NICU is no big deal a lot of our kids are on event, they're easy to manage they're like interns cases we don't worry much about them and here this distribution is about a third of the third of the other half of increasingly stringent of predictions of die the blue one 15% of the patients one person said I think the patient's going to die the green one is two people on the same day said I think the patient's going to die and the yellow one is everybody on the team that day said I think the patient's going to die so how well does the predictions of die actually predict death well 60% of all the babies were never predicted to die how did they do yeah, good for you, they lived so almost all survived stick with me, if 60% were never predicted then 40% were predicted to die you can do that math I'll give the suns an extra point for that so how did these kids do half of these survived as well so imagine this conversation I go to a mother I say ma'am, I'm Dr. Meadow I'm in your attending position you know your baby's really sick in fact I think your baby's going to die other people in our team also think your baby's going to die you should know that the University of Chicago we do research on this question and I'm wrong half the time so it has a different kind of feel to it than other ethical kinds of discussions would have with parents and it suggests that there may be a flaw or a problem in with our using this data set for any kind of counseling here's a cross-cultural tidbit that has to do with the MIQ okay so in the MIQ we learn quickly that being on a vent wasn't that big a deal, you could stop intensive care for MIQ patients whether they're on a vent or not in a way that doesn't happen in the MIQ so we studied all MIQ patients and in fact although we don't tell people so we don't go back and like spill the beans here we only analyze MIQ patients who've been in the MIQ for more than three days and the reason we did was we and others have shown both in justice units and other units that MIQ populations can quickly be divided a lot of patients who get admitted to the MIQ the adult ICU get out in three days, in fact more than half the patients are discharged in three days and almost all of them live they are not ethically problematic, they come in for whatever reasons you guys do you do a great job and they're out so we're only interested in MIQ patients who are there from 96 hours four days or more so what percentage of MIQ patients who were predicted to die by docs, by nurses, by attendings, by fellows actually died in the MIQ died before hospital discharge so they were predicted the patients were going to die before hospital discharge what percentage actually did yeah you'd think so zero, it's half, it's just the same but look what's different here's the crucial difference between the MIQ and the MIQ we did a second study that Jesse's team and I just recently published with in the last year 95% of the patients who were predicted to die are dead by six months so whether they die in the hospital or they go to an LTCH or they go home to hospice or they go to a nursing home or they go somewhere they die but in the MIQ that's not true half of the patients almost all the patients who survived MIQ discharge will not die quickly the post MIQ discharge death rate in the first year of life is less than 10% so the vast majority of MIQ survivors will live for at least a year or many years whereas in the adult unit they're all dead by six months which is why I now am showing you data to suggest that we pediatricians or neonatologists are right to mock all of these consults about end of life decision making in adults because it doesn't matter it just doesn't matter okay so now I want to talk a little bit more about some more subtle things we've done and then get to sort of a little bit of ethical nub before we're done so maybe you guys we didn't really suck when I said we were wrong half the time maybe I was looking at the wrong outcome maybe a better outcome would be the combined what we call an adverse outcome which is either you're dead or you're significantly impaired at two years of age we can have a debate about what significant impairment means for the rest of this talk it's that your MDI and your PDI which are your Bailey scores are less than 70 which is more than two standard deviations below the mean at two years corrected age we can talk about whether that's an appropriate use later Microsoft is in the audience has terrific data that maybe that's not a great thing but it is by far the most common one that's what we're going to talk about for the next 15 or 20 minutes we're going to talk about that later oh no I don't know why this is one and two and the other one wasn't we might use intuitions or algorithms I'm going to tell you for free algorithms were terrible for predicting death they're even worse for predicting death or impairment so we're not even going to talk about the algorithm data although I have them we're going to do just intuitions it's the same question we walk around, we ask people is this baby going to live or die but then we look at a different outcome we added one thing this is Joanne Ligata's fellowship thesis we added head ultrasounds and head ultrasound is easy, it's just an ultrasound of the best majority of you know what an ultrasound is it's sort of a video of the inside of the head and we look to see bleeding, that's what we're looking for for the most part and look what we found these are some of the coolest data in the neonatal prognostication this is not a surprise as your head ultrasound gets worse going to the right likelihood on the y-axis of either death or impairment goes up so normal head ultrasound still it's not so wonderful 40% of little babies with normal head ultrasound all kids under a thousand grams from our institution have an outcome of either death or impairment going up to severe where it's about 80 to 85% that's not a shock but look what we gained with your, our intuitions this is what she gained the gray bars here are just the same head ultrasounds in the light blue one person on one day thought the baby was going to die the green one is more than one person we had to corroborate it two people at least thought the baby was going to die on at least one day and the red bars are everybody on the team thought the kid was going to die and look what we learned we learned that for every degree of head ultrasound it's not just that the predictive power look what else we learned it's not just that the predictive power is better it's terrific so that if you have a look at the red and the green bars together for mild, moderate, severe any degree of abnormal head ultrasound so now we have ma'am I'm Dr. Meadow you know your baby is really sick in fact I and other team members think your baby is going to die if you look at these data and the likelihood that your baby is going to be alive and neurologically normal when the baby is two is four percent do we want to keep going or do we want to redirect care that I would argue is a profoundly different conversation to have with the mother than a conversation about it's 50-50 indeed that's why we're doing this because the 50-50 stuff is just imperfect when you do 20 career 24 weekers it's about 50-50 hi ma'am do you want to start intensive care on your baby take a coin and flip it that's what we know about your outcome that's just not good enough okay how much better are these predictors than just gestational age when they walk over to the OB side and talk to the mothers these are some way cool data so these are 177 kids and the black bars now are 23 and 24 weekers and the striped bars are 25, 26 and 27 weekers mortality or abnormal MDI is on the y-axis and here's the first thing you find okay that this is what you knew already if you're born in 23 and 24 weeks you do worse than if you're born in 25 weeks and more okay that's not a shock to anybody but look what we do learn if you have no events by that I mean nobody predict you're going to die and your head ultrasound is not abnormal then gestational age disappears the likely outcome is independent of gestational age if you simply use a trial of therapy as your predictor and so this we would argue is the way to counsel parents it's much better than what we knew here which is that it's 40 or 50 or something like percent of the bad outcome here's what Bree Andrews has done in the last couple of years it's terrific stuff basically what you do is you go online and my fellows do this one before they go over to the OB side and they say okay so we know the estimated gestational age we know the estimated birth weight we know whether the baby's a boy or a girl we know whether or not the mother's got an antinatal steroid you punch those numbers in and you get a number out this is the likely survival unimpaired and sometimes my kids will use that indeed when we've surveyed re-surveyed how many neonatologists use it and basically all neonatologists use it and what I'm trying to tell you is they're wrong to use it and I'm gonna tell you why these are 19 patients in our place who died before NICU discharge the purples are what we predicted after a trial of therapy and the black crosses or diamonds are what was predicted by the Tyson calculator before they were born and the point to be made here is that they're a pretty good fit there are a couple of kids who we predicted were gonna do well who didn't, they died but for the most part the fit here is quite good the NICHD calculator the black diamonds and Breeze trial of therapy indicators which is just simply a prediction of died by the caretakers and an abnormal head ultrasound of about 35 kids who survived with normal outcomes the purples the purple squares that's what the trial of therapy said the trial of therapy said these kids are gonna do pretty good and with only two exceptions they did good but look at the black diamonds a lot of them had really low predictions of likely good outcome before they were born and these are the kinds of data we'd like to show to say hey you know maybe a trial of therapy is a better way to individualize accurately your predictions of how patients are gonna do better than anything we know about flipping a coin before birth so here's a nice analogy and I like to show this because my son is a co-author on this paper how many people here know anything about sports betting some of you know about sports betting okay so I'll give you a hint in sports betting where the rules is you have to place the bet before the game starts right and what if you didn't what if I let you bet continuously during the game what if you knew that the Yankees were up by three runs before Mariano Rivera was coming in at the night don't you think you'd be more accurate in predicting that the Yankees were gonna be winning at the game that if you had to bet before you'd become better at predicting the winner as time passed and here's data these data were published in ACTA my son did this because he's got nothing to do with his job so he looked at I think 400 some odd baseball games and the likelihood of winning if you're leading after the inning so if you're leading after the first it's about 50-50 but if you're leading in the eighth it's about 94% that you're gonna win the game and then he had nothing else to do 135 professional soccer matches okay and here the likelihood of winning is down to about a third at the beginning because you can tie the soccer match but by the end if you're winning in the 90th minute you're 94% likely and obviously if you've won the game at the end of extra time you've won the game so you're a better predictor while the game is going on if we let you bait on baseball or soccer while the match is going on you do a lot better if we match line why don't we do the same for the parents of extremely low gestational age babies why don't we let the babies declare themselves and we'll know much better with 96% accuracy better than 50-50 so what do we learn if your outcome variable is die in the NICU or be an impaired survivor then you don't have to decide at birth time will help sort things out now the worth of dying versus being an impaired survivor is not settled indeed this is where I argue a lot of neonatal and perhaps adult ethics ICU research needs to go survival from the NICU neurologically intact usually everybody values that that's sort of what everybody's hoping for survival from the NICU neurologically impaired that's what the neonatologists fear but when you actually survey the children or their parents and Seroge Stegel has been doing this for 40 years now the children and the parents seem much happier with their lives than either the docs or the nurses ever would have anticipated which is interesting weird many people don't believe it but if you don't believe it you have some obligation to demonstrate it because Seroge is done go ahead so that's a terrific question yes being vegetative absolutely right Tracy and that's so let me just finish because there's only one or two slides and that could be the first question because it's a terrific question definitely NICU after initial resuscitation so Marin is sitting in the back doing this as part of her fellowship project you're in the NICU you get a lot of things done to you you're on event for a long time a couple of days, a couple of weeks, a couple of months and you still die how do the parents feel about that are they miserable because the docs have tortured their baby or are they grateful that the docs have spent and the nurses have spent so much time trying to help them that's a really interesting question and finally death without resuscitation or comfort care we have no data on whether there's anything like buyers remorse whether or not parents a year or two later feel guilty that they didn't try we just don't know so the final ethical exhortation for the young docs in the room process matters, don't abandon your patients that's what it means to be a dentist so there are questions but let's talk about Tracy's question which is a terrific question what do you want to say about various levels of neurologic impairment because at some level it's a key question for neonatalysis so there are a bunch of things I can say and I won't keep anybody who has to leave just feel free to leave so Annie Jean Vie has done some terrific work where she has sent surveys to docs to nurses, to medical students to anthropology graduate students basically any group of people she can get to answer these questions and she gives them 8 different scenarios and the scenarios are graded by likelihood of survival is 50-50 for 4 of them and 5% survival to 95% die for the other 4 and then the likelihood of neurologic impairment is also 50-50 for 4 of them and 5% normal and 95% impaired for others and she has pretty thick descriptions of what they're going to be like a 7-year-old with CP non-ambulatory can't go to school blind and deaf who now is in a car accident or a 24-week or just born or a full-term baby who has brain hemorrhage or a 3-month-old with meningitis or an 80-year-old adult with recurrent malignancy or an 80-year-old person with repeated strokes so there are 8 stories like that and then she asks people if they all came to your ER at once which one would you resuscitate first which one last and she asks if the parents came to you and said don't resuscitate would you accede to their request or would you ignore their request and do you think the resuscitation is in the patient's best interest so it's pretty straightforward and what she finds is in Annie's words that everybody hates premies that everybody hates premies so that a 24-weeker is 50-50 likely to be surviving and 50-50 likely to have any degree of impairment at all versus your MDI and your PDI are both greater than 70 and yet they were ranked second to last only the 80-year-old with multiple recurrent strokes lower and the 7-year-old with pre-existence CP and neurodevelopmental impairment who is now in a new car accident was ranked like in the middle so that it looks as if premies don't have the same claim on us as a population that older children do that adults do and it's not until you get the end of your lifespan that adults begin to lose that claim on us that we should continue to try resuscitative so the data on what percentage of NICU survivors are really terribly vegetative and the answer is about 5% or 10% of the survivors the most common NICU neurologic complication for NICU survival is 10 points of chance of bonus and it's actually ADD so it's executive function disorders and many parents who have children who have ADD at times feel frustrated with them but as a general principle they don't vote to kill them and so it's probably thought even by the parents in the best interest of those children to be alive so the question is how bad a cognitive disability do you have to have before parents transform their vision from one of rooting for the kid to rooting against the kid and then the NICU the outcomes are 90% at a minimum of NICU survivors won't be that bad and so with that you can go for that other questions we have a little time but I don't know people have to leave the one, feel free okay Dan I thought you were going to talk more about predicting neurological, devastating neurological consequences because that you set up that was something you wanted to avoid and apparently it doesn't occur very often well not devastating but complications are real so it depends on how much of your practice is filled with ADHD well I'm not talking ADH so if you're talking really devastated with my ex-graduates but think of how small your PICU is compared to either my NICU or think about the fact that there are many many more NICUs in the world than there are PICUs in the world I mean most community PICUs are just filled with asthma patients but not yours right but it seems it seems that you would get very different answers if you ask parents up front you know you're pretty certain about a devastating neurological consequence would you want to continue and depressive care versus five years later you have this child and are you pleased with the outcome and so I think and the question is should we allow people early on if we know that they'll probably change their mind in five years because you know they love this child or should we let them decide early on I want to stop now because I can't I don't want to have to deal with this so that both parts are terrific questions and I'll take them both in order the first one is how good are we at predicting really really bad neurologic outcomes and the answer there is we're not great but there are so few that I can't tell you how bad they are because it's only 10% of this so there's a couple of hundred kids in this group and we've done it again so we now have let's say 400 kids from our own institution so of the survivors so that you know there aren't very many kids who are really badly impaired and you'd have to sit and talk with Mike Masal a bit about how bad do you have to be before it crosses either the doc's mind or the parent's mind that we shouldn't have done this okay that's the guilt in the inatologist's field it's not the guilt that Segel finds that parents feel now the other part of your question was a terrific question and it's important for the fellows to hear it it has to do with sort of this when people don't like our data they don't like it for two reasons one is they think it's too soft intuitions are sloppy they're malleable and so what I say to people is fine you don't have to like our predictor find one of your own the point to be made is the trial of therapy is a better way to do this than anti-natally let the baby declare themselves you pick one pick a prostaglandin H5 level or something like that and figure it out for yourself you could feel free to do that but it's important that you do it but the other part of your question is this bonding issue that if the parents let the baby live and then bond to the baby they'll be less likely to stop than before they started and everybody's nodding because it seems to make sense and my question to you is implicit in your question was this idea that that's bad that you shouldn't let people bond to a baby and then not let the kid go the way they would have been willing to let the kid go before they started and the more you think about that the weirder that argument is this idea that people can't have ethical growth that their lives can't change that people are bound by what happens before they start the classic example is the guy who played Superman who was a horseback guy and he knew that it was a risk that if he jumped horses he would fall on his neck and break himself and he wrote all kinds of anti ex-anti things if I fall on my neck don't resuscitate me I don't want to do this and he did fall on his neck and he became a quad and of course everybody ignored him and they ventilated him and he came back and he lived for about seven years or something like that and once his life filled with oh my goodness you should have let me die and the answer was no it wasn't he became a spokesperson for quads and stuff like that and so this idea that people can't change and somehow their earlier ethical intuitions are more valid than their subsequent ones strikes me as weird I'm not saying that we should take one over there I'm questioning then should we give people the choice at that moment or should we inform them that even if you'll probably learn to love them that would be terrific and I had a fellow a long time ago his name is Bob Hips who was a fellow with us and then got out of the field because he couldn't he couldn't bring himself to deal with exactly the question you're asking which is how do you accurately inform families of what it's like to have an impaired child do you take them to back when Mantino had you know homes for impaired children and you say hi here's what your kids going to look like or do you show them a downed kid on TV and say here's what your kids going to look like it's virtually impossible to do and we're not so good at predicting the really really bad ones from the not so bad ones who are still not normal it's an impossible question for us at the moment to answer although it's an important one I completely agree yep it's a terrific question Jesse yes so it's a terrific question and certainly we've experienced it both on Jesse's side and NRR side as well so the short answer the sort of the cop out ethics answer is it's fine with me I'm giving the most accurate data I can they can make whatever decision they can and I'll support it now that sounds a little blasé there's a part of it that's true but think and Jesse's clearly the person to both engage in this and tease about this um what's your alternative well you have a couple of alternatives one alternative is to give them the numbers and just be okay with whatever they say the other is to Will Young but he's learning these tricks already so I told you this morning you shouldn't lie but you can be somewhere between deceptive and ambiguous about how we counsel families so there's no hope well do you really mean there's no hope or there's just a really little hope and so you yourself like we do as well frame these discussions with some degree you shouldn't you write this down with a pen don't explicitly lie that's like bad even Siegs would agree with that okay but having said that there are ways of framing discussions and yet if you're asking me do I think that it's better for parents to know that it's 96 versus 4 compared to this 50-50 for a likely outcome yeah I think that's helpful now do I think that it's going to be determinative and that everybody who gets a 96 versus 4 is going to say oh that's bad odds I'm not going to do it no I think there's going to be people who are going to hang on to the 4% and what I say there is in my NICU that's fine I got them I can name them in our NICU today I won't but I could yeah you know what happens at some level 4% of the time and that's okay it's not too much money okay it's not like the adult unit Javad when you say end of life care I think understand it until I try to understand each word and I wonder if you could as much as you could clarify end of when is the onset of end of another word you decide now it is the end of onset the second thing a little more difficult and you said it life some of the kids you and I saved with interest as I follow them up I didn't see much of a life in this children it was a biological life and it was as if it was the end of life of the mother another word they like that child that their life was gone and second when you say care what do you mean by care you mean send them home and let them die you mean you are going to continue do some kind of care what does care mean if you these three words a little bit clarify I appreciate sure so the care part I'll take first Javad you Javad you may know it's a neurosurgeon an eminent one and has been doing neurosurgery the vast majority of our NICU survivors are not in your world they need care in a very different sense they need love from their parents they need support they may need some meds they need early intervention but they don't need a lot of medical stuff some of them have lung disease they need some oxygen weeding and maybe some diuretics but they are not overwhelmingly needy of medical intervention even the ones like with BPD the number of kids we have will go home on I don't know two every two years or something like that and the kids who have shunts yeah we have some but we don't have that many now for the mothers who feel that it's that their lives are over the data are that when you ask the mothers there aren't very many most of the mothers love these kids come everybody is welcome to come in September third week third Sunday in September the NICU reunion you know we have six seven hundred people come back most of these people are pretty happy now maybe it's a selected group maybe all the ones who are miserable don't come back but the one so JD was just here that the the emotion that all of the attendings have is at some level one of puzzlement because we're looking at these kids and they come back and they're 12 or 13 years old and they're clearly impaired they limp or they have CP or they're cognitively imperfect and the parents come up to us and they're ecstatic doc I'm so grateful look at how well Johnny or Jamie is doing and we're thinking to ourselves we're ecstatic and we're happy and comforting and cheering with the parents and we're thinking wow people are really resilient that they can be so loving of a child that nobody into pregnancy hoping for and so what I'm trying to do here is to give you guys a feel for we are docs we are medical professionals this is what we do for a living the easy ones are the well babies and the normal ones these are the hard ones and the question is can we gather some data to help parents in their hard decision making but that'll stop I can sit down here for a bit