If it were easy...

A few weeks ago I was chatting with a graduating resident who was trying to make a difficult choice between two job offers. One job was a particularly challenging hospitalist job and the other was a safer choice. She’d recently had a baby and since I’d done the same thing (had a baby in my third year of residency - not recently) I understood the temptation she faced, as a new mom, to choose the less exhausting path. We then began to talk about job satisfaction for hospitalists and the attrition she’d seen in young faculty. I started to describe how tough the first few years as a hospitalist were for me and she said, “but it must be so much easier now.”  Well, yes and no. Hospital medicine remains a tough job.  And there are parts of it that never get any easier. Sometimes a week on the ward can feel like a week in the spin cycle of your washing machine - you exit completely wrung out.  

The first rule of pediatric hospital medicine is that having a child hospitalized does not make people happy. Hospitalization is almost never a good thing for a family and you always have to keep in mind that families can be under intense stress, even when the child’s illness is not life-threatening. Many families don’t have enough ready cash to afford to feed a parent staying with a child in the hospital or to make the drive back and forth from home. They often struggle to find a safe place for other children to stay when the parents are at the hospital.  Sometimes they are in danger of losing their job due to missed work.  And worse, all the sympathy in the world won’t defuse certain situations that we get into as hospitalists. How, exactly, do you make the best out of telling a parent their child is having a pseudoseizure? How about when the parent is suspected of being the problem, as with child abuse?  Those are never going to be positive interactions and nobody expects them to be, but we are still human beings first and professionals second.  Cut us and we’ll bleed.

Sometimes the more overtly adversarial interactions are less emotionally costly, if only because you’ve steeled yourself for it.  The cases that can really get to you are what I call the “hot potatoes”.  You know the type, the medically complex patient with a large psychosocial overlay complicating care. There are always multiple consulting services involved but nobody really owns the whole patient. The patient gets passed from doc to doc like a hot potato because everybody is avoiding the real issue.  Sometimes that issue is pursuing aggressive interventional care in a patient with a really poor prognosis.  Sometimes it’s simply that the parents have unrealistic expectations about what medicine can actually achieve. Often the parents are angry and often with good reason. Often you are angry as the doctor, because any number of people could have or should have addressed the issue, but somehow it fell to you.  And, if you are addressing the issue, you are likely doing it solely out of a sense of duty or professionalism or even pity for a suffering child and it is costing you personally to deliver the bad news or to push the difficult question when you could just pass the buck instead.  When you catch the hot potato, it burns.

And then, there are your mistakes. We all define mistakes differently and punish ourselves to varying degrees, but let’s not pretend we don’t make them.  Medicine is full of uncertainty. You can overtreat or you can undertreat, but getting it just right is mostly fiction. It’s impossible because most of the time the “truth” in the illness is only revealed after the fact, if ever. Is it better to try early transition to oral antibiotics and risk a treatment failure or is better to expose a child to sedation, a central line and risk a complication of prolonged access?  Is it better to take a chance on an unproven or potentially risky therapy or to persist in watchful waiting? We know there are no completely right answers in most cases but we are tormented by the feeling that somebody else might have done it differently….and might have been right when we were wrong.

And finally, the job is just not easy, even when you get it mostly right and when the parents and the patients are mostly happy.  There is something I call “decision fatigue” that hits me after a few days of sustained high-acuity or high-volume inpatient work. Every decision comes with risk, even in the least ill of hospitalized children. Weighing the risks, weighing the evidence and weighing patient and family understanding and preference all take time and effort, sometimes an astonishing amount of time and effort. By the end of a week, I am occasionally so tired of making decisions that I can’t even choose what I want for dinner off of a restaurant menu.  Even the simple questions can send me into a circular spiral of angst over evidence, risk and benefit.  How many days of treatment? Further imaging? Dex or Pred?  Do you want fries with that?

So, yes and no, it gets easier but it never gets easy. Doing the job can be a source of tremendous satisfaction. It’s a meaningful job and I’m glad I do it.   Sometimes, it’s even fun. But it’s never, ever easy.

Physician as Pathogenic Agent?

“Don’t’ just do something, stand there….”

Is it incongruous for that particular saying to be the personal motto of a hospitalist? I’ve been asking myself that question for more than a decade and I still can’t decide. Like most hospitalists I enjoy managing acute illness and I love a good procedure (some who know me well might even consider me a little bit of a procedure-hound). I chose an area of practice based on these proclivities, and yet, I continually find that the very “best” things I do clinically are almost always characterized by the absence of any doing.

 

I had a rough week on the ward a while back….one where I came very close to facilitating an unnecessary invasive procedure on a basically well child. As can be common in these situations, the physician at the bedside (often a hospitalist) will bear the brunt of the family’s anger. It’s what we signed up for after all, but it’s still emotionally costly. And, it always provokes in me a certain peevishness, a curmudgeonly harrumphing at the way medicine is practiced in the US. How did it come to this?  The question is even harder to ponder when the history includes caring parents and good doctors, as it often does. The answer: gradually, one step at a time. Step one: we empirically trial a “harmless” medication we know is not really indicated for a self-limited or minor condition. Step two: we order a “few tests” or we consult the subspecialist “just for reassurance.”  Step three: repeat steps one and two for other minor illnesses or normal variants. Step four: get out of the way of the runaway train. And, as every hospitalist knows, anybody who steps in front of a runaway train is sure to get flattened.

 

So, of course, I went to the medical literature to help me cope with my angst. Perusing the early release articles online at Pediatrics I came across an article with a title that I could not resist: “The Hazards of Medicalizing Variants of Normal,” by William Carey. Dr. Carey’s article turned out to be a commentary on a study about the impact on parental desire for medical intervention based on the diagnostic labeling of infantile reflux as a disease, i.e. the impact of the use of the term GERD.  But I was less interested in the study than the commentary to be honest. I was absolutely charmed by Dr. Carey’s prose; consider for example, this comment about the use of acid blockade for infant regurgitation:

"One must ask whether it is adherence to the principles of evidence-based medicine to recommend use of an ineffective drug for a condition that has not been confirmed by appropriate evaluation. We are informed that this is happening frequently.”

 So, when I noticed that Dr. Carey referenced some of his own work from the 1970s in the commentary I absolutely had to track it down, which is how I came to read his article: “Avoiding Pediatric Pathogenesis in the Management of Acute Minor Illness.”  There are many insights in this article, most of which I really can’t improve upon, though I can share a few of my favorites. Here is what he has to say about the negative consequences of “overmanagement” of acute illness:

“unnecessary fear may be generated in the family and child, who think of the child as being sicker or more vulnerable than he really is. Inappropriately extensive or repeated diagnostic measures may reinforce the importance of the symptom in the mind of the parent and child, making the physician a “pathogenic agent.” Unnecessary harm may be done to the child with excessive use of modern diagnostic and therapeutic agents, exposing the child without justification to iatrogenic physical disease…like acute diarrhea from antibiotic treatment….Dependence on drugs may be encouraged.”

 I don’t know about you, but that sounds like a typical day at the office.

In the end, despite the fact that I serve a different patient population, I believe that Dr. Carey’s observations apply to my practice just as aptly as they applied to pediatric practice more than forty years ago when he wrote the piece.  His observations seem fresh and his recommendations for future research still seem pertinent. Two of his final questions struck me as the two that I am continually preoccupied with in my own practice.  The first: “how can a doctor find out if he is making these errors and learn to change his ways” speaks directly to the lack of meaningful feedback on pertinent quality measures in most pediatric health systems, and more pointedly – to the thorny problem of how we might promote physician behavior change.  Have we made progress on this question since Dr. Carey posed it in 1972?  The second: “If pediatricians have thought so little about it, what are they teaching…?” speaks to our role as medical educators and our obligation to have difficult conversations with learners about practice variation in academic medicine.  Again, any progress? 

No doubt things are different than they were in 1972, but I propose that they are similar enough to make reading or re-reading Dr. Carey’s observations to be worth the effort, particularly if you happen to have had a day where you worked exceptionally hard to do absolutely nothing. Or rather, I should say, a day where you worked exceptionally hard to do your job.

 

Carey WB. The hazard of medicalizing variants of normal. Pediatrics. Published online April 1, 2013.

Carey WB, Sibinga MS. Avoiding pediatric pathogenesis in the management of acute minor illness.  Pediatrics. 1972;49:553 -562.

Moral Hazard and Morality

Like seemingly everyone else in the country, I spent the last couple of weeks reading (it is a really long article, after all) and discussing Steven Brill’s recent article in Time, "Bitter Pill: Why Medical Bills Are Killing Us.”  It’s a tour de force by all accounts. Personally, I loved it.  It’s compelling and it’s got a real Upton Sinclair vibe which strikes me as appropriate to the gravity of the situation. Others were highly critical, which should not come as a surprise based on our inability to discuss healthcare rationally in this country. A frequent criticism was that it was nothing we didn’t already know (true, but doesn’t diminish my admiration for his saying it better). I also heard the frequent lament that he failed to offer solutions despite making a strong case for the fact that healthcare is not a free market in the US.  Personally, I think he did offer the solution but he knew better than to state it outright, given the fact that doing so would have marginalized the impact of the piece, which did run in Time after all, and not Mother Jones.  And, the piece really was all about emotional impact. You can’t read it and not viscerally feel something.  Brill is telling stories and stories move people.

In perusing reactions to the article, I was struck by the continued insistence from the business crowd that the solution to outrageous pricing in healthcare is actually little more than moral hazard. Holman Jenkins, in the Wall Street Journal, was derisive, and asked us to read prior, more cogent commentaries.  One he suggests is by Duke scholar Clark Havighurst, who was quoted: "The market failure most responsible for economic inefficiency in the health-care sector is not consumers' ignorance about the quality of care but rather their ignorance of the cost of care, which ensures that neither the choices they make in the marketplace nor the opinions they express in the political process reveal their true preferences."

As a little cog in the big machine that produces the commodity that Havighurst is analyzing (indeed, the cog that signs the orders that generate all of the bills), I couldn’t disagree more.  However, it occurred to me that if you replaced the word “consumers” in Havighurt’s sentence with “doctors,” - well, then you might actually be on to something.  What I am getting at here is a constant tension I feel when providing care in a cost vacuum.  Many physicians care deeply about evidence-based medicine and they feel intense pressure to provide high quality care. Nobody has escaped the push to measure quality in healthcare and we are all struggling to define what it means to be a “good” doctor.  We think we know a lot about what might define quality, but unfortunately we know very little about what constitutes value in healthcare because we have no idea what any of it costs. This is partly true because it’s actually illegal for us to share information about what it costs across practice locations because (it was feared) this could result in price fixing.  Cost benefit analyses and comparative effectiveness research are increasingly popular in academic circles, but they aren’t part of regular medical training and they rarely make their way into the daily considerations of the doctor who pushes the first grain of sand that starts the avalanche of bills.  Medical education is completely off the rails at the moment for so many reasons, but a further major failure is the complete absence of economics in the curriculum.  So, I respectfully disagree with the business school elite who suggest that moral hazard will cure our ills. If the doctors themselves can’t really define quality all that clearly and fail miserably at defining value, then the consumer has little chance of doing so in my opinion.

Why?  Consumers buy a lot of stuff they know absolutely nothing about, right?  And somehow a free market seems to allow things to all come out in the wash.  Unfortunately, in healthcare, the details are devilish. Foremost to me, is the problem of uncertainty in medicine.  Much of the time we are not even close to certain that our diagnosis is accurate.  Let’s say I get the diagnosis wrong but not so wrong that any harm comes to you. How would the consumer know he’s been misdiagnosed? It takes a very long time to rack up the credentials that allow me to misdiagnose you after all and very few illnesses that Americans seek treatment for have immediate consequences of the treatment. Misdiagnosis and overdiagnosis (pretty much considered to be rampant in the US) dramatically complicate things. In reality, one reason (and not even the best reason) that healthcare can't reasonably be analyzed as a commodity is this uncertainty. It’s not a fixed thing and you just can’t pin it down. On the second trip to the doctor’s office, heartburn is sometimes unstable angina and sometimes it’s still heartburn. Oh, and the doctor really has no idea how much it will cost to sort this uncertainty out. So now we are going to make a free market out of an uncertain diagnosis with a doctor who can’t give you advice on the cost-benefit of any of the testing. Better to take your chances with an apple a day.

I’ve never really understood why economists fetishize the idea of healthcare as a commodity. I guess, from the outside, it looks like you could show up, pay your fee, punch the diagnosis into a computer and get a price estimate for three competing institutions on how to fix your problem.  It actually might work with certain obvious things (I’d venture to guess that we are reasonably certain a woman is pregnant.) But these are precisely the things we have dramatically overcomplicated in the current system due to low tolerance for risk and the ready availability of services (again, pregnancy works really well as an example).  But from the inside, healthcare looks like religion. My definition will not match yours, even if we happen to be in the same specialty and sit right next to each other in the pews on Sunday (or Saturday as the case may be). 

In the end, the religion analogy may be a little more apt that I would like. I couldn’t help but react to Brill’s article with significant moral discomfort. Something vaguely immoral is going on in healthcare today and the people Brill describes are bearing the brunt of it. And, by extension, I am actually guilty by association. I have personally generated part of the billing for outrageous amounts for uninsured children which will have undoubtedly bankrupted families. I know I have because I’ve had the conversations. I remember the last one quite vividly. I was taking care of a teenage girl who’d suffered severe Guillain-Barre Syndrome. So severe that she had respiratory failure and an ICU stay. I was caring for her as she approached discharge and we were debating rehab vs. going straight home. It soon became clear that rehab was not going to be an option, despite the fact that I worked in a religiously affiliated, not-for–profit hospital at the time, rehab beds were in short supply and they were cash on the barrelhead or insurance only. During that discussion her mother broke down and we talked for a long time. She explained that the family had decided to forego health insurance about a year prior when costs just became untenable. They ran a small construction business and could no longer insure their employees either, so they just let it all drop. However, they had a few assets, all of which would need to be sold off to pay the 6 week/million dollar hospital bill, and then perhaps bankruptcy.  It was an incredibly depressing prospect for a mother taking home a child who still couldn’t really walk without assistance and who was left with lingering neuropathic pain. I was honestly near tears, and then my patient’s mom let loose with a stream of invective directed against the president’s healthcare reform bill (which was still being challenged in the courts at the time).  For a moment, I thought hard about trying to defend the bill, which clearly had nothing to do with this family’s plight and might actually have attenuated it to a major degree. Instead, I just hugged my patient’s mother and told her not to worry, and that we’d figure it all out in the end….something I did not believe at all.  Moral hazard, right? A premeditated choice to drop your insurance policy, roll the dice, and pay the price when you lose.

 I guess…..but how could anyone analyze all the factors that went into this choice in a rational manner and make a “good” decision? Is there a good way to analyze the risk of catastrophic illness hitting a family (unless you happen to be related to an actuary) and then calculating the costs?  Is there really an option to forego healthcare for a catastrophic illness? Wait, now add in the fact that the patient was a child. In the end I’m awfully tired of the term – moral hazard. It’s a vaguely nasty idea cloaked in sheep’s clothing. Everyone wants to be moral, right?  But the morality involved isn’t your own in this case, it’s mine as the physician.  Am I supposed to stand by and fail to provide care when my patient can’t pay (apparently I am if I work for MD Anderson, which you’ll learn if you read Brill)?  Many physicians won’t sign up for that, but many of us work for institutions that will do the next worst thing – ruin the patient’s life with the bill.  Now I am not that naïve, I completely understand that a hospital can’t run without sending out bills and I don’t really know if an economic entity can truly be moral…..but I do know when my own particular morality feels the squeeze.   And finally, I think that was where Brill was going in this article, past terminology and partisanship, and into the hard truth of what it looks like to make a person an offer they can’t refuse.

Brain on Fire - Book Review (sort of)

Lately I’ve had a bee in my bonnet about anti-NMDA receptor encephalitis. So, when I picked up a recently pulished illness memoir by the journalist Susannah Cahalan called Brain on Fire and it turned out to be about the disease, I really had to read it.   Luckily the book is much more than a medical procedural so I haven’t ruined it for you by giving away the diagnosis.  My clinical experience with the disease is limited to peripheral involvement in a single case…as far as I know. However, If you read the book, I have no doubt that you will ask yourself how many people have had the diagnosis missed and spent the rest of their catatonic lives in a psychiatric facility.  Cahalan spends some time pondering this question as well, since she initially appears to be suffering from psychosis and just misses a trip to the psychiatric hospital. She calls it survivor’s guilt, but the doctor’s feel it too.

Although it was discovered as a paraneoplastic syndrome in adult women with ovarian teratomas, anti-NMDA receptor encephalitis is turning out to have a significant pediatric predilection.  There have been at least two pediatric multi-center case series published to date - Florance in 2009 and Armangue in 2012. From these series, it seems that paraneoplastic disease is much less common in children, and the female predominance wanes in the youngest age groups. In general, children recover in the vast majority of cases but the treatment is aggressive immunomodulation and the response is not rapid.  As a hospitalist, I am mostly interested in not missing the diagnosis and, peripherally, in wondering when I might have missed it in the past.  MRI imaging may be normal until the disease has progressed and the symptoms may be indolent and subtle early on, and without clear correlates on physical exam, hence the mislabeling as psychiatric. The constellation of abnormal movements, language dysfunction and psychiatric symptoms seems to be a “classic” early triad, with seizures and catatonia also common as the disease progresses.  Of further interest for pediatricians, a laboratory based investigation from the California encephalitis project found that anti-NMDA receptor encephalitis was the most common diagnosis in a cohort of patients <30 years of age  in samples submitted from clinicians treating encephalitis of uncertain etiology , surpassing enteroviruses, herpes simplex, West Nile, and varicella. Imagine that!  It was the most common etiology and we were completely unaware of this diagnosis until five or six years ago.

About the same time this disease was being elucidated, I was taking care of a truly unfortunate patient with herpes viral encephalitis. He was a school age boy who was diagnosed and treated in a fairly timely manner, but developed the described complication of choreoathetosis afterward.   We all worried about recurrence but in the absence of any evidence to that effect, he was eventually shuffled off to rehab to try and make the best of it. He had severe language dysfunction with outrageously disruptive coprolalia and a debilitating movement disorder that didn’t respond to any treatment. It was one of the most depressing cases I’ve ever been involved with because I watched his family fall apart in response to this devastating illness.  And, now, as I perused the Armangue case series, I was struck by one of their cases which was eerily similar to my former patient; a child with post-HSV movement disorder failing to demonstrate HSV persistence in the CNS who went on to have NMDA receptor antibodies discovered in the CNS.  Their patient improved on immunosuppression, mine was unimproved after months of rehab and chronic acyclovir. What if I’d heard of anti-NMDA receptor encephalitis in 2006?

But back to Cahalan’s story. As a pediatrician I was struck by the clear picture she presents of what it feels like for a family to experience a life-threatening illness in their child. Although she was 24 when she fell ill, she was basically reduced to a childlike state and her family had to band together to be her advocates and protectors.  Time and again, through her illness, key pieces of the history were not available to the treating physicians because she couldn’t provide them. The illness impaired her ability to communicate and she was terrified of telling the doctors about some of her symptoms for fear of being labeled “crazy.” Early on, she sought care for persistent numbness and tingling on one side of her body, but with a “normal” MRI and exam, the symptoms are written off as “stress.”  The incident will, of course, make you think of all the strange complaints that people report and doctors can’t explain.  One doctor, in particular, illustrates our natural and sometimes fatal tendency to explain things away rather than just admit we don’t really know. He is her initial neurologist, and in the absence of objective findings he is the source of “stress” as an initial diagnosis- a not-unreasonable suggestion at that point in the illness. Unfortunately, after she suffers a generalized tonic-clonic seizure, he decides she’s likely suffering from stress and alcohol withdrawal. He clings to this diagnosis long after it is reasonably plausible to do so and even in follow-up he does not seem to have learned anything from having missed the diagnosis so profoundly.  This doctor is intended by Cahalan as a lesson in humility for our profession. I take him that way as well, and as a reminder that it can be much more productive to admit that you really just don’t know.  

In the end I’d guess that my fascination with this disease has to do with all of the diagnostic and therapeutic dilemmas I remember - the previously healthy children who show up out of the blue with devastating and progressive illnesses while we stand by helplessly and wonder….if only.  If only we could find the right clue… if only there was a better option... if only we were smart enough.   Anti-NMDA receptor encephalitis continues to fascinate and  our newfound ability to diagnose this disease will hopefully decrease the ranks of the “if onlys,” for us,  if only by a little bit.

Florance NR, Davis RL, Lam C, Szperka C, Zhou L, Ahmad S, et al. Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis in children and adolescents. Ann Neurol 2009;66:11-8.

Armangue T, Titulaer MJ, Malaga I, Bataller L, Gabilondo I, Graus F, et al. Pediatric anti-NMDAR encephalitis: clinical analysis and novel findings in a series of 20 patients. J Pediatr 2012.

Gable MS, Sheriff H, Dalmau J, Tilley DH, Glaser CA. The frequency of autoimmune N-methyl-D-aspartate receptor encephalitis surpasses that of individual viral etiologies in young individuals enrolled in the California Encephalitis Project. Clin Infect Dis 2012;54:899-904.

The Sinner's Opinion of the Sin

The most interesting article in the January issue of Pediatrics (for hospitalists, at least) was the piece on preventability of pediatric readmissions by Hain et al.  In this study the authors assembled a team of four physicians to review 200 randomly selected 15 day readmissions to an urban children’s hospital over a two year period and classify them as to their preventability. The authors begin by pointing out that the preventability of readmissions in children has not been well studied and that assumptions that patterns would mimic adult readmissions are largely untested. Nevertheless, there is a growing pressure to hold the discharging hospital accountable for readmissions, be they adult or pediatric.  I am going to admit right up front that I am largely skeptical that pediatric readmissions are a high value target for quality improvement work.  I won’t fault you if you think that attitude is overly self-serving, a bit like asking the sinner to gauge the damage caused by his sins, but I am going to persist in my belief until somebody proves me wrong. And the evidence, as presented in this paper, is in my favor.

Hain reviewed 200 of 2346 readmissions at Monroe Carrell Children’s Hospital at Vanderbilt. The overall readmission rate for the facility for the 2 year period was 8%.  The large majority of the sample (74%) had a chronic illness and more than 80% of the readmissions were for medical (as opposed to surgical) illnesses.  The team rated 20% of sampled readmissions to be potentially preventable which would extrapolate to mean that less than 2% of their total readmissions are potentially preventable. If their methods are sound and their sample is representative, we don't have much to work with here, do we? But, even if neither of those things is true, 50 of the 200 readmissions reviewed were planned readmissions.  Ahem, I know it’s stating the obvious here, but 25% of the readmissions were planned!  I really do hope somebody at CMS notices this study.

Another thing I like about this paper is how well it takes a potential limitation and actually turns it into a strength, of sorts. Inter-rater correlations for classifying preventability were far from perfect. Except for the 50 planned readmissions, the four person panel was rarely universally in agreement when they worked independently.  There was more correlation than disagreement, but nevertheless there was enough disagreement that they had to meet as a group and review competing logic as to preventability prior to final categorization of the readmissions. Generally, this resulted in the readmission being classified as NOT preventable.  Hmmm, what do we think the implications of that particular piece of information might be?  Will a computer program be able to determine if an admission was preventable?  Will a single “functionary” in an insurance office be able to do it?  There really may be a lesson in the trouble taken to achieve correlation in this study.

Some cases are surely easy to classify as unpreventable. Take pediatric cancer for instance. Even if the readmission was not for scheduled chemotherapy, a pediatric cancer program must allow for a certain rate of readmission for febrile neutropenia. We don’t really have the option of going easy on the chemo the next time around to improve the readmission rate.   What about a child with a line infection?  Was that the surgeon who placed the line’s fault, or the nurse who last accessed it, or the home health agency doing the home management?  If it’s a gram negative infection in a short gut patient then you really have to blame the kid himself, right? How much detail are you going to need to determine the preventability of a line infection?  And what about the previously healthy kid with pneumonia who met all discharge criteria but still developed a late parapneumonic effusion? Preventable? What if I chose a ridiculous antibiotic regimen to treat the initial pneumonia, like say azithromycin alone? Finally, consider the largest number of “hospitalizations” in pediatrics, the birth hospitalization for normal newborns.  Should admissions for jaundice or fever really be considered REadmissions? One will need to look very closely at the particulars of the case to determine preventability, and, as it turns out, it may take a four person panel to sort it out.

I’m sure we are all tired of the saying, “children are not little adults,” but it’s hard to resist pointing out that children’s readmissions are not just adult readmissions writ small.  There is, in general, a lack of pediatric expertise in outlying emergency departments and a much lower threshold to admit young children when uncertainty exists.  I don’t really have the ability to “prove” that statement, I simply know it’s true because I’ve spent the majority of my medical career on the receiving end of those admissions.  Nevertheless, even though the universe has its thumb on the scale against us when we weigh readmissions, Hain et al have now given us data with which to argue that there may be as many as 20 preventable adult readmissions for every 1 preventable pediatric readmission.  

As we partner with the hospitals in which we practice, we should advocate for a much more sophisticated understanding of pediatric readmissions. And even if you question the validity of Hain’s methods, the work really drives home the need to consider the large volume of planned readmissions in pediatrics.  I understand that we are all desperately casting about for a metric -  something, anything we can measure to reasonably gauge the quality of inpatient care, particularly on the medical side. Overall, it’s an important search and we should keep trying. However, this sinner is stilll waiting for someone to convince me that limited resources are best spent on the particular sins leading up to readmission in pediatrics.

Hain PD, Gay JG, Berutti TW, et al. Preventability of Early Readmissions at a Children’s Hospital. Pediatrics 2013; 131:1 e171-e18.

 

The Rock and the Hard Place

I keep coming back to one particular article from this year in half of the conversations I have lately.  It’s a JAMA piece by Don Berwick and a RAND health care analyst named Andrew Hackbarth entitled, “Eliminating Waste in US Health Care.”  The piece starts, like most writing about health care in the US, with the dismal realization that health care expenditures are rapaciously chewing up the GDP and we are on a collision course with a giant asteroid. Not news, right? After setting the stage in admirably succinct fashion, they discuss solutions floated by both sides of the political divide, many of which have been focused on some form of cost shifting.  Across the board cuts vs. “rationing” - choose your rock or your hard place, doctor.

Berwick and Hackbarth go on to make the argument that waste makes up more than 20% of health care expenditure in the US, and that we’d do well to address the issue more aggressively prior to asteroid impact.  For those of you who’ve ever worked in a hospital, the idea that there is significant waste in health care  cannot possibly be news, though even if you’ve heard it all before, you’ll want to check out their estimates, if only to get a sense of the scale of things.  Berwick categorizes waste into six large buckets: overtreatment, failures of care coordination, failures in execution of care processes, administrative complexity, pricing failures, and fraud and abuse. He further estimates a low, midpoint and high cost for each category based on the available literature. Oh, and the costs are all in billions, just in case you were wondering.

Quite honestly it’s really just one line in this paper that I am stuck on.  It’s a very BIG line - the one that quantifies the cost of overtreatment.  Of the six categories Berwick and Hackbarth use to classify waste, overtreatment is the largest for almost all of their estimates.  However, they’ve placed it in the middle of the table rather than at the top for some unspecified reason. Granted, I’m congenitally predisposed to dislike overtreatment and have expended a large amount of time on the issue, so maybe I’m just being touchy here, but seriously, why not put it at the top of the list? JAMA, after all, is directed at a predominantly physician readership. The other categories are important, sure, but as a physician one has to admit that overtreatment is the one that we really own.  Berwick and Hackbarth go on to define a proposed approach to addressing the massive problem of waste using proportional attacks on each category -  climbing the mountain one step at a time.  However, the vast majority of us are not policy-makers, administrators or analysts; we’re in the trenches at the foot of the mountain.  What can we do about waste?  I don’t feel empowered when I think about administrative complexity or care coordination or fraud. I feel like a cog in the big machine. The category that resonates with me is overtreatment. As physicians, our choices dictate the existence of the category.  We are entirely responsible. The buck stops here.

Berwick defines overtreatment as: “waste that comes from subjecting patients to care that, according to sound science and the patients’ own preferences, cannot possibly help them—care rooted in outmoded habits, supply-driven behaviors, and ignoring science.” I absolutely love this description. There are things I can sink my teeth into here.  It reminds me of all that I need to do daily in order to do my job well. These are hard things in my opinion. I need to know “sound science,” which means I need to acknowledge and use the available evidence as my shared baseline when negotiating care. I am never done with my medical education because sound clinical science is ever-evolving.  I need to know the patient’s preference. And, in order to know the patient’s preference I need to acknowledge and to be able to communicate clinical uncertainty.  I need to recognize outmoded habits and address them. I can’t do what I did last time, without first asking myself if that course remains sound.  And when it’s not, I have to change my behavior, and, boy, behavior change is hard.  It takes skills we’ve not been taught as physicians to change our clinical behavior.  Finally, I need to recognize my supply-driven behaviors. This is likely the most difficult task of all - along the lines of identifying Rumsfeldian unknown unknowns. So much of medicine is based on social relationships. Do I order certain scans or tests because of subtle pressure to utilize a very expensive investment my institution may have made? Or, do I over-consult specialists employed by my institution because I know they need the clinical revenue?  These are difficult behaviors to spot in one’s self, much simpler to leave them unknown unknowns.  But I may as well tell you that the estimates for the cost of overtreatment are between $158 billion and $226 billion for 2011. The numbers are so staggering as to be almost meaningless to me, like the distance to the sun.

And yet, when it comes to overtreatment, it’s all on us, so we’d better start climbing……..or perhaps I really meant to say, we’d better start standing still.

Berwick D, Hackbarth A. Eliminating Waste in US Health Care. JAMA. 2012;307(14):1513-1516.

Defining Insanity.....Through Blood Cultures

Pediatric hospitalists are intensely interested in the utility of blood cultures. It may be the single most common theme of papers we have under review at HP right now.  In fact, I’d say hospitalists are very nearly obsessed with the topic (myself included, as I’m about to prove).

Here’s a story to put it in perspective. Recently, I admitted a 2 year old with a positive blood culture in the setting of fever without source. He’d been discharged home from the ED after his initial evaluation, but when the blood culture alarmed at just under 24 hours, he was called back for re-evaluation. He’d been quite healthy for the last year, but was born moderately prematurely and suffered a serious infection at birth, then been evaluated for immunodeficiency afterward. Although immunodeficiency was ruled out with extensive testing, uncertainty still clung to the child. A second blood culture was drawn (with some difficulty) 24 hours after the first and he was admitted for observation. He was clinically well appearing without having been treated and afebrile, but his past medical history trumped his clinical appearance in this case. Of course, anyone who has been on a hospital unit for more than a week knows that coagulase negative Staphylococcus grew from the blood culture. No problem, right?  Except the second culture also grew gram positive cocci (aaarrrgggh).  So, instead of going home the patient got to spend a second night in the hospital awaiting culture results, this time on IV antibiotics after the second positive culture. And, yeah, we did define insanity and pulled that third culture.

Hmm…..what are the chances of two contaminants in series? If my lab has an average contaminant rate of 4%, the chances of two in series should be 0.16% (conditional probability).  That certainly seems worrisome for the patient in question, and some might even interpret it as a 99+% chance that we’d found a pathogen (am I missing endocarditis?)  Of course, in this case the second culture also grew coagulase negative Staphylococcus, which incidentally had a completely different susceptibility pattern from the first isolate, thus nicely confirming contaminant status for both isolates. I’d be willing to bet you weren’t surprised by the outcome.

The American College of Pathologists says that a blood culture contaminant rate of 2-4% is acceptable, and in a setting where you only draw a blood culture in very ill patients, that is probably not a bad call. But, what if you draw a blood culture as a screen in most young febrile children? Or, more commonly, what if you don’t see enough kids to feel comfortable deeming them “very ill-appearing” or not?  Don’t forget that not so long ago it was conventional wisdom that febrile children under 36 months of age all needed a blood culture. The rate of occult bacteremia approached 10% in certain populations. Now, in the age of HIB and Prevnar, it is rare enough to be difficult to quantify in a fully immunized 2 year old. Have you ever seen a study addressing the yield on blood cultures drawn from IV catheters at 2am in the ED in fussy 2 year olds?  Even if we had the information, using conditional probability to assess the likelihood of two contaminant blood cultures in series in such a situation is probably not going to put me any closer to taking better care of the patient after the damage is done.  My calculations would predict that I’d experienced a very unlikely event and yet my clinical sense was always that this patient was the poster boy for contaminant blood cultures.  The numbers say I should have been surprised by the whole mess, but just like you - I wasn’t.

Maybe, I was looking at the wrong numbers. The question, when evaluating blood cultures prior to final determination of species in pediatric patients, is probably much more complicated than the overall contaminant rate. It’s really about the likelihood of a positive culture turning out to be a pathogen given the relatively higher likelihood of contaminants.  If we start with stable (and by stable, I mean not septic) patients with gram positives growing from peripherally obtained blood cultures (because that is the hospitalist’s sample), then we can comfortably estimate that a kid with a 1 per 100 (1%) likelihood of bacteremia now gets a test where the likelihood of a positive being a “true positive” is 1 in 4 (25%) on a good day. Humor me here, and let me push this point. Really, we ought to take into account that the overall contaminant rate is drawn from all-comers in your hospital lab (which could include patients who know how to sit still and don’t bite) and make an attempt to account for the difficulty of the draw itself. We could use total number of blood draw attempts as a proxy for difficulty (my patient was stuck four times prior to success on one ocassion).  We’ll also need to consider the issue of drawing the culture off the IV (also done in my patient’s case).   I must say that I lack published evidence for the assertion that this practice is detrimental, but….duh.  Then, we’ll need to take into account the time to positivity for our culture, which we know from the literature has a fairly significant predictive value.  So, to make a long story short, what all hospitalists know is that we have to modify the utility of nearly every positive blood culture by a number of factors, whittling down the test's utility until it often has remarkably little positive predictive value (until speciated, of course).  Even without doing the math, one can see that we have a situation where an initial positive test may not really move me off of my pre-test probability estimate of bacteremia.  Very little illustrates that painful dilemma more than getting two positives in series and still not having real disease.

This is the point where most reasonable human beings begin to tear their hair out.  Something we take for granted as necessarily a “gold standard” test (you can’t have bacteremia without a positive culture, right?) really functions more like a screening test, at least for the first 48 hours. Hence, you now understand the proliferation of research intended to clarify things like time to positivity and appropriateness of blood cultures in different clinical setting.  I’ve also seen multiple quality projects intended to redress rates of contaminant blood cultures.  Without a doubt, these are all worthy efforts.  However, after having torn out my share of hair over the issue, I can’t help but think that the real solution has to include drawing fewer cultures in the first place…..now there’s a really sane idea.