A surgeon whose popular blog is named Skeptical Scalpel put up an interesting post recently about those online physician rating services. In theory, of course, it’s a great idea for people to give their opinions and rate their experiences with any goods or services they buy. The notion goes back long before the internet with publications like Consumer Reports, which is now itself online. Sites such as Angie’s List are very popular and can give you testimonials about providers of everything from home renovations to car repair to daycare. So it’s not surprising several entrepreneurs have started physician rating sites. The Skeptical Scalpel wondered about how useful these are to patients and families. He thinks not much. A huge issue is the sample size is so tiny; many physicians get only a single review. Also, the reviews are anonymous and there have been several well known examples of stalkers trashing a physician’s reputation out of some animus to the person. It’s also well known perhaps a fifth or so of reviews at sites like this are fake anyway.
So I wondered. I’ve been practicing for 35 years and cared for thousands of patients. I have moved a few times and now work at several hospitals, so perhaps I’m hard to track down correctly, but maybe not. Also, rating pediatric critical care providers for patient choice reasons seems silly on the face of it because nobody shops around and plans for things when their child is suddenly critically ill or injured. You just go to the nearest place and are treated by whoever is there. Still, I thought it would be interesting to look. I would imagine I was a parent looking to get the inside scoop on that Johnson guy. The sites generally contain some biographical information about the physician — address, insurance accepted, where he or she trained, if he or she had ever been disciplined by an official body. I wondered how accurate that would be for me. So here we go.
US News and World Report: This site listed me. It had my correct address and correctly listed my specialty, medical school, and years of practice (although it only said 21 plus). It was correct in all my training data and even listed a couple of awards I’ve gotten. It looks as if they obtained my CV from Doximity, a physician site. They did miss several of the hospitals where I have credentials, including the one where I mainly practice, but they did get my various medical licenses correct. On the other hand, they didn’t rate me at all. So if I were a parent checking up on me I’d only have my credentials to go on.
Health grades: This one got my main hospital correct, unlike US News. It also got my years of practice correct. But there were no reviews of me at all.
Vitals: I’m listed on this site. The details of my training are skimpy but what is there is correct. They also got my main hospital affiliation correct. I got 1 review — 5 stars (WooHoo?). I’ve seen hundreds of patients at that hospital, yielding 1 review.
MD.com: This one was hopelessly wrong. It stated I practice primary care pediatrics, which I don’t and never have, and it had an address for me that was 10 years out of date. It had no reviews of me. I suspect this website was just abandoned by its owners.
UCompare: These guys had me listed. The address was wrong and there were no reviews.
RateMDs: I wasn’t listed.
ZocDocs: I wasn’t listed.
WebMD: This is a huge site. Most people use it for information about diseases but it also lists doctors. It listed me, but there were no reviews.
So there you have it. I’ve cared for thousands of children over the years and many hundreds since these doctor rating sites have been around. I had 1 review. It was a great one, but really, I can’t see that any parent would have found searching for information about me to be useful. The great majority of Google hits were to this website, the one you are reading now.
There was a recent sobering editorial in the New England Journal of Medicine that caught my eye. Its title was “Alternative science and human reproduction” and its author is R. Alta Charo, professor of law and bioethics at the University of Wisconsin. It’s been clear now for decades that significant numbers of prominent Republicans either believe or profess to believe things about human reproduction that simply aren’t true. Examples include the debunked notion abortion is associated with increased breast cancer risk or hormonal birth control works by causing repeated abortions, that birth control pills are “abortifacients.” In the words of the editorialist:
Human reproduction has become the victim of alternative science, rife with alternative definitions of well-understood medical conditions and characterized by rejection of the scientific method as the standard for generating and evaluating evidence.
Several recent appointments to the Department of Health and Human Services are disturbing in this regard. The new deputy assistant secretary of HHS for population affairs has openly stated the government has no role in the very thing her job is. She also asserts the absurd idea hormonal contraception is ineffective in spite of overwhelming scientific evidence to the contrary. Another Trump appointee, Charmaine Yoest, has this in her curriculum vitae:
Yoest, former head of the antiabortion advocacy group Americans United for Life, helped to develop the strategy for a Texas statute that was so filled with obstacles to abortion services, presented in the guise of protections for women’s health, that the U.S. Supreme Court abandoned its usual degree of deference to state legislatures and struck down the law because its underlying factual claims were patently false.
But there’s more. Katy Talento, another new appointee, claims contraception causes infertility and miscarriages, claiming contraceptives are “breaking your uterus.” These are the people now in charge. They don’t understand science. Worse, they don’t want to understand science if it conflicts with their existing beliefs. They have the power to do immense harm to the public health, particularly the health of women.
I swiped this editorial cartoon by Steve Sack from the redoubtable Dr. David Gorski’s blog, who goes by the nom-de-web of Orac. Recent epidemiology shows reducing the fraction of vaccinated children in the population rather promptly leads to a resurgence of the diseases vaccines protect against. This is the concept of community or herd immunity. Epidemiologists debate the concept around the margins but overall its importance is well accepted. People who deny the effectiveness of vaccines or even think vaccines are dangerous don’t accept it, though, and you can find many examples of this around the internet, some sort of reasoned, some not. Although I’m trained in pediatric infectious diseases, a field that includes a lot of epidemiology, I’m not an epidemiologist. So I’m not going to chew on the whole herd immunity thing. I’m going to write about a particular form of concern trolling common among vaccine denialists: the claim that they would fully support vaccines if only vaccines could be shown to be fully safe and effective, using their own special definitions of what that means. In effect they erect an impossible standard to meet, which is of course how concern trolling works.
A common claim is that, although individual vaccines may be safe, the safety of combinations of vaccines together has never been shown. It’s not always clear what the demand is here, but it often appears to me to mean they want a trial of all the many possible combinations of vaccines compared. If you just do the math on how many vaccine combinations are possible you can see this demand is absurdly impossible to meet. We actually do have ongoing surveillance of vaccine safety happening all the time and the results show them to be the safest medical procedure we have, with around 1 complication per million doses administered. The only thing safer is homeopathy, which does nothing but harms nothing. Vaccine denialists elide this fact by redefining what a vaccine complication is through including nearly anything that happens to a person afterwards, even years afterwards, as vaccine-caused. I’ve read posts by many adults who claim, for example, that their fatigue, difficulty concentrating, and “metabolic problems” stem from vaccines they received as children.
And then there is the mythical unicorn vaccine denialists claim is the only acceptable answer: a direct comparison of vaccinated and unvaccinated children, searching for differences in outcomes. Key here is the frequent claim unvaccinated children, on average, are healthier than vaccinated ones. To someone who knows little of clinical research this seems like a perfectly reasonable demand. Just compare vaccinated to unvaccinated children, see who got illnesses or complications and who didn’t. I just saw this one on my Twitter feed today:
Also note this tweet includes the common fallacy the Amish don’t vaccinate — most do. Anyway, there have been some terrible studies of this sort reported, such as this one, which highlight the fallacy of doing such a simple comparison without controlling for any confounders. Fundamental to any proper study like this is that the two groups being compared, in this case vaccinated and unvaccinated, must differ only in the variable being tested. A common way of handling the question is a case-control study, in which each case is matched with one or more controls that, as best as can be determined, satisfy that requirement. But vaccinated and unvaccinated children, by parent choice, are hopelessly self-selected right out of the gate. There are other confounding issues, such as blinding, but that’s the main one.
Well then, as some have actually demanded, we must have a randomized controlled trial (RCT), the gold standard of clinical research. RCTs use random assignment of subjects to one group or the other, in this case vaccine or a placebo (fake vaccine), and ensure both the subjects and evaluation team be blinded to who got what. Think about this for a minute. They are demanding parents agree to subject their child to a trial in which they have a 50/50 chance of getting a fake vaccine. All this to satisfy the concerns of vaccine deniers. It would be incredibly unethical to do such a study, and no institutional review board (aka human studies committee) would ever approve such a thing. For such trials there must be reasonable uncertainty about which group is getting the better treatment and in this case there is none. The bottom line is any vaccine skeptic who demands proof like this is being massively disingenuous. It’s akin to demanding a randomized controlled trial of parachutes.
The enduring mystery in this perennial chestnut of a topic is that vaccine deniers demand a level of safety and certainty from vaccines that they demand from no other medical procedure or treatment. Absolutely every treatment I can think of is riskier than vaccination. Some are far, far riskier. I suppose it’s partly owing to a visceral resistance to injecting something into a healthy person, but vaccine denial in general has deep, deep historical roots.
This post isn’t specifically about children; it’s about the elderly, Medicare population. Even so I found this article newsworthy. It’s long been known air pollution is associated with increased asthma rates. I’ve written about asthma in children. Decreasing air pollution leads to reduction in asthma, an encouraging trend. There have also been long-term studies showing a correlation between air pollution and overall mortality among all ages. The offending agents in the polluted air are ozone and fine particles, those with a diameter of less than 2.5 microns, a size too small to see with the naked eye. So it’s established chronic exposure to polluted air is bad. But what about exposure to any air pollution at all, to air we wouldn’t ordinarily think of as polluted? In other words, is there a safe threshold?
Air pollution of course varies from place to place across the country. Here is an instructive map about that. The top panel shows the fine particles, the bottom one ozone.
The investigators used an enormous database consisting of the entire Medicare population of 61 million persons studied over 12 years (2000-12). It totaled 460 million person years. Ten person years, for example, is 1 person followed for 10 years or 10 persons for 1 year. That’s huge. The authors’ goal was to find out if lesser degrees of exposure to air pollution, values half or less of those usually regarded as dangerous, were associated with increased mortality. They found they were, at least in the elderly. The graphs below display what they reported. Again, the top panel shows particulates, the bottom panel ozone. The left axis shows the hazard ratio for death, with a value of 1.0 meaning no hazard.
What the data show is what is called a dose-response curve. This is an important part of any study of potential cause and effect because, if the association of something potentially causing something else is real, you should see an increased effect if the dose is increased. Sometimes such curves plateau if maximal effect is reached, but these graphs show steadily increasing risk of death with increasing air pollution.
I think this is significant research. It suggests there is no safe level of polluted air. The editorial accompanying the paper agrees: its title is “Air Pollution Still Kills.”
As the kids say, it’s complicated. Practicing physicians are seeing an ever increasing list of protocols and pathways coming their way. These arrive in several forms — order sets for medications, guidelines in how to proceed for various conditions, when to do this, when to do that, and when not to do either one. They generally are the product of various committees trying to synthesize what these days we call “best practices.” I have no doubt these aids are useful. They can focus the mind, streamline care and reduce medical costs. They also have the potential to improve care on average.
Checklists, one form of the phenomenon, have been shown to make care safer for several things. The model is simple, following what airline pilots do. If you’ve ever glanced to your left as you filed onto an airplane you’ve seen the pilots methodically going over a series of lists as they prepare for takeoff. Nothing is assumed to be correct and functional until it’s been checked. The practice of anesthesiology in particular has been made safer by the concept; everything is checked and confirmed to be ready and in place. Nowadays we even make sure we’ve got the correct patient by checking the identification band on the person’s wrist. Surgeons verify and mark in advance the spot where they’re going to cut. A popular recent book by Atul Gawande is all about medical checklists.
Treatment pathways are not as clear-cut as checklists. We have several of them in the PICU — for asthma and bronchiolitis, for example. My colleagues in adult practice have many more. There is of course an obvious danger in these things. Although they can help streamline and standardize care according to best practices, medical cannot be practiced well strictly using cookbook lists — every patient differs, sometimes by a lot. The longer I practice, now over 35 years, the more I appreciate this fact.
What set me thinking about all this was the publication in the New England Journal of Medicine of an article analyzing the effect of sepsis pathways on patient outcomes in septic shock. I won’t go into the details of what sepsis is because you can read more if you like about this serious and often lethal condition here, here, and here. Essentially it’s a cascade of complications usually initiated by invading bacteria. Sepsis pathways have become a big thing. They have done some good, I think, by focusing on doing the few things we know help early in the disorder. Mainly they make us think of possible sepsis before it’s obvious, and early treatment improves outcomes. But this careful analysis showed good clinical judgement was just as good as predetermined cookbook measures on a sepsis pathway list. Not surprisingly, strict use of the cookbook bundle was much more costly than were tests and treatments driven by experienced judgement. I’m not surprised at all.
I think pathways are particularly helpful for young physicians and residents. Experience is a great teacher, but all of us have to start somewhere. Pathways and guidelines jump start learning, remind us of what’s known, and focus our minds on what to do. But a pathway in the hands of someone who doesn’t know enough is dangerous, providing confidence where there should be a measure of anxiety.
Randomness in life is inevitable because the universe is a pretty random place, although the extent to which you believe that depends upon your own value system. This notion comes into play almost every day of my practice in the PICU because many of the critically ill and injured children I care for experienced a sudden onset of the difficulties that landed them in the PICU. I do see a fair number of children with significant chronic problems for whom a PICU admission is, averaged over a year or so, nearly inevitable. But most of what I see is random. Some families find this highly disturbing.
Humans seem programmed to explain events in some way. All of us have things happen to us, the cause for which we attribute to something that came before. Often this makes sense, but at least as often we want it to make sense even if it doesn’t. We don’t like significant events to be random. Of course that is the basis for the well-known logical fallacy of post hoc, ergo propter hoc, Latin for “after that, therefore because of that.” Another way to put it is the principle that correlation does not necessarily indicate causation. Vaccine opponents are famous for applying this fallacy to whatever unrelated event happens after their child has received a vaccination. Parents of children in the PICU often search their minds for something they did or didn’t do that led to their child’s problem. Often there isn’t anything — it’s just random.
Some families find this randomness distressing. I have heard more than a few parents actually express anger that all the good and proper things they did appeared to have no effect. For example, some appear almost to think breastfeeding or generally following a certain diet, taking vitamins and probiotics, and exercise makes their child bulletproof to infections. All those things help at the margins, but events are still random.
Cancer is the entity that really rankles with its randomness. On average, just over 40% of us will get cancer sometime in our lives. It is a disease of aging, and our population is aging. So, even if overall death rates of many cancers are falling, more and more of us get it as we age. The hardest cases to accept are when children get cancer, and they do. The overall incidence is about 18 cases per 100,000. Cancer is fundamentally a genetic disease, in that the genetic material in a few of our cells becomes altered in ways that lead to uncontrolled growth. Few, if any of our body’s cells live as long as we do. Our persona, our essential being, is really a huge community of cells that is constantly turning over and passing on their duties to a new generation of cells. The genetic mutations of cancer arise in cells as they replicate their genes and produce more cells. Mutations like that are a well-known statistical inevitability because the process of gene replication makes errors. Not many errors, but a known number. If one of those errors involves a gene that regulates cell growth cancer can arise, such as the breast cancer cells shown in the figure above. It is widely known there are risk factors for this to happen. Some we are born with (family history), some are environmental exposures, such as tobacco smoke or ultraviolet light and radiation. But many cancers, probably the majority, result from random bad luck. Predictably, tissues characterized by high cellular turnover, and hence a lot of gene replication, are more prone to develop the uncontrolled growth of cancer.
This notion received a lot of attention from a couple of recent research articles that tried to estimate just what proportion of cancers are from such bad luck, not due to any environmental or inherited cause. The articles, discussed here and here, estimated around two thirds of cancer are attributable to such random bad luck. That research generated great controversy, but I think it illustrates a great truth: bad stuff happens irrespective of no action or non-action on our part. We already easily except that principle with natural disasters and other catastrophes, but it’s just as true for many of our and our children’s health events. I see and share the bewilderment and anger of parents in the PICU over that, but there’s no helping it.
Anybody who has spent time on internet message boards, whatever their focus, has encountered a species widely known as the concern troll. Discussion threads on Twitter are also loaded with them, although the structure of Twitter makes the threads more difficult to follow. Twitter also is more likely just to have a back-and-forth of hurled invectives in capital letters because the 140 character limit per tweet lends itself to this technique.
A concern troll is someone who presents himself/herself (itself?) as genuinely interested in other points of view but has “concerns” about it. This posture is often presented as “just asking questions.” The troll is simply interested in learning, it says. Yet if you naively answer all the troll’s questions, it just leads to more questions. The troll never asks for further information, receives it, thanks the person who offers it, and proceeds on. The troll always has a “yes, but . . . ” comeback. Concern trolls are famous for the trick of moving the goalposts. You answer one question adequately, but this just leads to more questions from the troll. When you run out of patience with this dance the troll declares victory, often by pointing out how unreasonable you are — not thoughtful like the troll.
I dabble now and then in the interminable online vaccine debates, if debate is really the correct word. Some anti-vaccine combatants patrol the internet with their shared bullet points and links to discredited or just plain wrong publications. They have a standard toolkit of images which they believe are conclusive winners in any flame war. They also have a standard list of insults: paid shill (for Pharma in general, vaccine makers in particular), liar, fake doctor, and the like. They generally insist all information from all reputable sources that support vaccine science is false (or paid for by Pharma) . . . because they support vaccines. Yes, it’s circular. Many insist there is a vast conspiracy afoot to foist vaccines on a helpless public.
Those are the strident ones. I find the concern trolls more interesting. They generally come across as reasonable, at least initially. They have concerns about vaccines, they are just asking questions. They portray themselves similar to that unicorn of political reporters — the mythical undecided voter. But the more you answer their questions, the more they have. They keep bringing up this, and that, and then another thing, and another. The goalposts recede toward the far horizon.
Vaccine trolls claim they would support vaccines, but only if the safety and efficacy meet their exacting demands, standards which are not applied to anything else in medicine. If they applied similar “concerns” to daily life they would never get in a car, get on an airplane, or eat in a restaurant. Since their demands are impossible to meet, they regretfully, with sadness, must reject the whole of vaccine science and epidemiology. For myself, I think vaccine concern trolls act the way they do because they think of themselves as reasonable, thoughtful persons. But really they just are against all vaccines. Concern trolling allows them to manage their cognitive dissonance.
Anyone who knows anything about medicine seems to have heard the famous line primum non nocere. You could call it medicine’s Prime Directive (apologies to Star Trek fans). It’s Latin for “first, do no harm.” This simple concept is also one of the four basic principles of medical ethics, that of non-maleficence, the others being beneficence, autonomy, and justice. So we shouldn’t harm our patients. That seems simple, but working out exactly what it means in practice is complicated because often, in the short run, we do things to our patients that run the risk of considerable harm. The key calculus in sorting this out is considering the risk/benefit ratio of an action. We also need to consider the wishes of the patient (or the family for a child) because people vary greatly in their willingness to tolerate risk.
Of course risk tolerance is a sliding scale for most people. If a person has a life-threatening condition, say a leaking aortic aneurysm, the alternative to a high risk procedure to fix it is an even higher risk of death. On the other hand, if the condition itself is trivial and unlikely to cause serious harm, say a simple rash, the risk of treating it with high dose intravenous steroids is higher than the condition itself. An even more interesting example to me is vaccination. Here we have a situation in which we are giving a therapy to a healthy child, something we do for two reasons: to protect the child from future disease and to protect society from having a pool of unvaccinated persons who can harbor and transmit disease to others. The risk of vaccination is infinitesimally small, but it is not zero. Of course unvaccinated persons also have a tiny risk of contracting the disease, but that is not zero, either. My own view is the good of society trumps the question in this case, tipping it toward vaccination. But it reminds us to take questions of risk versus benefit, of primum non nocere, seriously.
As I’ve written before, people are quite illogical, even delusional, in how they regard risk, particularly medical risk. They particularly fear an extremely unlikely yet dramatic thing while tolerating the real possibility of a significantly bad thing without worry. Physicians make this worse at times. Have you ever signed a consent for a surgical procedure, even a minor one? The language is quite frightening, often including the possibility of death. It’s there even though the statistical probability of death could be similar, or even less, than that of a tornado striking the hospital during the procedure and wrecking the operating room.
The important principle to me in all this is to lay out the risk/benefit calculus when explaining what you’re proposing to do to patients. Tell people what the risk of doing something is weighed against the risk of not doing it. If it’s scan or x-ray of some sort, the risk is not having the information the scan will give or of finding some unexpected yet ultimately trivial thing that leads down a rabbit hole of more tests, the so-called incidentaloma (plus radiation for CT scans). In theory we’re supposed to do this all the time, and procedural consent forms even include that language, the risk of not doing the procedure, on them. But in my experience that doesn’t seem to get understood well. High risk procedures or tests demand a high risk of harm if they’re not done; after all, desperate times call for desperate measures. But everything we do, including the simplest of blood tests, carries risk. Make sure it’s worth it.
I occasionally dip my toe into the constant internet flame wars over what the generous term vaccine skepticism, the less generous vaccine denial. If you’re looking to see one of these quickly, Twitter always has some ongoing vaccine fireworks. The character limit of Twitter tends to compress the exchanges into hurled invectives, only occasionally punctuated by futile pleas for calm. Many quickly devolve into exchanges between posters of what they believe to be crushing evidence in the form of an image or two. A huge variety of message boards and blog comment streams can be just as vitriolic and the posters are not confined to 140 characters. It’s a swamp, but it can be a fascinating swamp. Of course full disclosure requires I say I am a stout defender of vaccine efficacy and safety, which is not surprising since I am a pediatrician with subspecialty training in infectious diseases and critical care. One particular talisman vaccine skeptics (I’ll be kind here) obsess over is the package insert found in the vaccine boxes. I’m often challenged on Twitter with an image of one and told to “read the insert, stupid.” Here’s a fairly benign example of what one sees:
That one’s just kind of ominous. There are plenty more that make it sound as if vaccines are highly dangerous — if you just read the insert. Here’s what can easily result:
Are you terrified yet? No? Well, if you’re not, perhaps it’s because you have a broader understanding of what drug package inserts are and how they got there. The inserts are produced by the manufacturer of the drug and they have long been required by the FDA to include them. Every drug has them, and you can actually see a huge compendium of what’s in them by referring to a book that has been around for many decades: the Physicians Desk Reference, or PDR. For many years every licensed physician received these huge tomes in the mail every year, although I haven’t gotten one in a long time because now it’s on the internet. Mine always went straight into the trash bin anyway. Why? Because the package insert isn’t really used by anyone in using any drug. We use published scientific data and expert opinion in deciding how and when to use a drug. The PDR site has a pretty interesting disclaimer in fine print at the bottom:
PDR.net is to be used only as a reference aid. It is not intended to be a substitute for the exercise of professional judgment. You should confirm the information on the PDR.net site through independent sources and seek other professional guidance in all treatment and diagnosis decisions.
The package inserts are, in my opinion, mostly designed as an attempt to protect the manufacturers from liability for untoward outcomes. They throw in absolutely everything that seems remotely possible or has been anecdotally reported. And you should know the reporting system is not filtered at all, merely compiled. I’ve seen lists of side effects that include opposite things: constipation or diarrhea, high blood pressure or low blood pressure, and quite a few others. Vaccine package inserts are no different in this respect from those for a host of other drugs. Even the insert for normal saline, which is salt and water, sounds kind of scary. Saline is one of the cornerstone treatments we use in pediatric critical care, yet the insert warns: “Safety and effectiveness of Sodium Chloride Injection, USP 0.9% . . . in pediatric patients have not been established.” This is pure nonsense. We inject it as quickly as possible all the time for severe shock. I’ll end with this useful image, since vaccine skeptics do love to pepper their posts with dramatic images:
No, we don’t. That, as the kids say, is the tl;dr (too long, didn’t read). But if you are interested in why I think so, read on.
Here are some background statistics. In 2017 nearly 36,000 new doctor applicants competed for about 32,000 resident positions in all specialties. For pediatrics, there were 2,738 first year residency slots spread among 204 different training programs. This is about 9% of the total positions available, a figure that has remained constant for some years. And, consistently, around 7-9% or so of graduating medical students choose pediatrics for their career. Since most first year residents go on to finish their residency that means we have around 2,500 new pediatricians completing training each year. How many of those choose to continue training in some subspecialty? In 2017 it was nearly 1,400, meaning around half of new pediatricians. The flip side is that half choose to practice primary general pediatrics, a much higher proportion than is the case for internal medicine.
Of the new pediatricians choosing to subspecialize, to train beyond general pediatrics in something like neonatology or cardiology, how are we doing in pediatric critical care? If you look at the results for the subspecialty match, you see critical care was tied with pediatric hematology for number three in popularity; neonatology is perennially number one, followed by pediatric emergency medicine at number two. In 2017, 188 pediatricians chose to continue their training to become pediatric intensivists. Nearly all of them matched to a training program. Here are the trends for the past 5 years:
Note there hasn’t been much change, maybe 10% from year to year. So the popularity of pediatric critical care has been fairly constant. Most of these fellows finish their program, so roughly 175 or so new intensivists enter practice each year. A few of the graduates may choose to return to their home country if they are not US citizens; we don’t know how large that number is. We need to staff our current PICUs and replace those intensivists who leave practice. Do we have enough people?
The American Board of Pediatrics has been certifying pediatric intensivists since 1988, the official dawn of the formal subspecialty. As of 2015 the board has certified 2,377 intensivists over that entire time span. (My certificate is #163, making me an old coot in the specialty.) How many of these intensivists are still practicing, and how many PICUs do we have in the US that need their skills?
In 2015 American Board of Pediatrics published a detailed analysis of the workforce in pediatrics, including historical trends. Among other things, it showed the total number of pediatric subspecialty trainees in all fields had doubled over the past 15 years. Critical care, for example, has gone from 99 entering trainees in 2000 to the 179 noted above. How many of these 2,377 pediatric intensivists are still practicing in the US? A few clues help decide how many of my colleagues have left practice. We know that, by 2015, 399 of the 2,377 did not renew their certification, a pretty solid indicator they are no longer practicing. That’s a lot of people — 17% of the total. If we add back the 2016 finishing fellows we are left with around 2,200 practicing pediatric intensivists at most. The workforce data show something else: how old we are. Around half of currently practicing intensivists are in their prime career age of 35-50. A quarter of us are over age 55. I’m one of only 81 still practicing beyond age 65.
What about PICU numbers? There are just over 400 PICUs in the US at last count. They vary greatly in size from 4-6 beds at smaller community hospitals to 40 beds or more at the nation’s 200 dedicated children’s hospitals. There are around 4,000 PICU beds in total throughout the country. Now we need to match up intensivists with PICUs.
On average, there is one pediatric intensivist per 35,000 children in the US. This varies quite a bit state to state, from a maximum density of one per 17,000 in Vermont to one per 225,000 children in Montana. Wyoming doesn’t have any, but they don’t have any PICUs, either. So we have 400 PICUs to work with and about 2,200 intensivists to distribute them to. Right out of the gate that’s 5-6 intensivists per PICU. But large PICUs at academic medical centers have many more than that, often 8-10 at least. This is of course partly because these units are large. In addition, their intensivists typically have other duties such as teaching and research. My totally but I think educated guess is there are about 50 of these large units, potentially soaking up around 500 pediatric intensivists. This leaves maybe 1600 for the 350 remaining PICUs, or perhaps 4 intensivists for each one. All of this assumes everybody is working full time, which is not the case, and that the workforce is evenly distributed, which it isn’t. To me the numbers look pretty tight. My own anecdotal experience bears this out; hardly a month goes by, often more frequently than that, without somebody somewhere calling me and recruiting me. And they know I’m 65 years old so, although I have no plans to retire yet, I would just be a short-term fix for their staffing problem. Many of my colleagues around the country are working short-staffed and have been for years. Of course this overwork can contribute to burn-out and make the problem worse. My off the cuff estimate is that there are about 2 intensivist positions for every working intensivist.
So my bottom line is that we don’t have enough pediatric intensivists and, looking at the training statistics, we are unlikely to get significantly more. I don’t think the number of PICUs is too high; closing a bunch of them would be an unsatisfactory solution. The standard of pediatric care has advanced over the decades and every critically ill child deserves an appropriately skilled physician. I think the number of people we are training at best matches the number leaving practice. Occasionally I learn of new PICUs opening, but I think we currently have roughly the number we will have for the foreseeable future. Of course some may close owing to various reasons, primarily financial (PICUs generally lose money) or staffing-related.
So what should we do? I think we need to modify our staffing model. Some have advocated using advanced practice nurses, physician assistants, or other what are termed mid-level providers. This might help, if supervised by intensivists. This can work in large PICUs, particularly ones that have significant numbers of patients with the same types of problems, such as dedicated cardiac units. But smaller PICUs treat a wide variety of complicated issues, things not well suited to practice pathways and protocols often used by mid-level providers in other settings. All manner of things roll through the PICU door. So I don’t think that’s the answer. I think our future lies in the rapidly growing pediatric hospitalist movement, pediatricians who work exclusively in the hospital and can rapidly become fairly proficient in managing straightforward, less dire critical care issues if they have backup from pediatric intensivists if needed. For many smaller PICUs, a blended practice of intensivists and hospitalists can work; I know of several PICUs that use this model successfully. One thing seems clear: we never will have sufficient numbers of intensivists to care for all these children by ourselves.