Saturday, October 3, 2015

Good At It

Friday, 4:55 pm. I am getting sign out from a colleague in order to cover the newborn nursery for the weekend. At one point, my young colleague sighs and says, “I just don’t think I’m very good at this.” I brush her off with a casual comment along the lines of: how much trouble can a bunch of normal babies cause? She isn’t satisfied. She is an excellent doctor as well as an excellent colleague, so she goes on: "I have this baby who was just born, with a pretty significant murmur and the nurses have already called me about it, so I don’t want to leave anything hanging over you for the weekend." She describes the situation in further detail. We talk it through and we decide that almost nothing would cause me to need an urgent echo for an otherwise well baby. Eventually we move on.

When we’re done, I come back to her expression of distress and “not being good at it”. I assure her that she is quite good at it. She worries she’ll miss something, a minor anomaly, a harbinger, a clue to impending disaster... I assure her that she will almost certainly do just that, and that we all do. I tell what I consider to be a funny story about a former colleague who missed an imperforate anus prior to discharge from the nursery. I pause to let that sink in ... but she can’t laugh just yet. She’s still too wound up from a busy week. She has already reviewed the prenatal ultrasounds, gotten pre- and post-ductal saturations and assured herself there are good femoral pulses on the baby in question. The baby is fine, but the physician is suffering. She finally says she is concerned about her physical diagnosis skills because somebody always manages to find something on the physical exam that she has missed. I say, oh like what? -the dreaded clinodactly of the fifth finger? Still... no laughter.

Our apparent belief that an obsessive pursuit of minor variations on the theme of normal makes you a good doctor has always been a pet peeve of mine, but suddenly I see how the phenomenon has shaken the confidence of a truly excellent junior faculty member. This matters because it is undermining her ability to say that any infant is normal. Of course, the infant in question wasn’t normal. Her mother had issues - advanced maternal age, a mildly abnormal glucose tolerance test, an “elevated” blood pressure, etc, etc. The infant herself was born at 37.5 weeks which also is not normal. The point being that normal no longer seems to exist. Still, how can I possibly find fault with a careful consideration of all the facts, including clinodactyly of the fifth digit (which is after all an anomaly, and furthermore, three such minor anomalies presenting in the same infant provide a 90% chance that the infant has a potentially significant underlying major anomaly - did I simply make that up? Do you like how I used "potentially" to modify the word significant? I swear I learned this factoid in residency, can you prove me wrong?*** Don't give up now, this story really does have an ending.)

I saw the baby the next morning and found the murmur to be present, but fairly soft and in no way provoking of further concern on my part. However, in the middle of the next night the baby had an “event” while undergoing a routine nursing assessment. Who knows what happened, but the baby appeared pale or blue or just “off” and had low sats, which prompted an ersatz 3 am resuscitation that included a few positive pressure breaths and an inevitable trip to the NICU. There, he received an evaluation for sepsis and an urgent echo…almost certainly because we had written murmur all over the chart. The labs were normal and the echo was normal, which you already knew because otherwise there would be no blog. Oh... but the baby did suffer a small pneumothorax, most likely due to the resuscitation - though there is no way to be certain of that. I am, however, certain that he spent a week in the NICU for culture-negative sepsis.

It is quite possible that the baby in this story really was intrinsically sick. I am aware that infection can be quite subtle in newborns. I am also aware that my own awareness often creates a tautology, a situation in which the conclusion is equivalent to the premise. For what is “culture negative sepsis” if not a circular argument? If you object to my characterization, consider the obverse: If I can never rule out infection in an infant with certainty then I can never be sure that I didn’t make a normal infant sick simply by suspecting that it might be so. I sincerely do not know which it was in this case.

Certainly, it’s necessary and useful to cultivate humility in the face of clinical uncertainty. But, it is equally useful to consider the harm we may do to our patients when trying to allay our own fears created by that uncertainty. And finally, it is also worth noticing, now and again, the harm we do to ourselves. I devote a large amount of time and energy to trying to elucidate the chain of events that lead us to provide unnecessary medical care. I never want that to stop me from providing necessary care. Yet, I can’t fail to notice that many, many times there is simply no way at all to tell the difference. I have days, like my young colleague described above, where I am profoundly depressed by the thought that I am not “good at it.” For me, the real impediment to feeling “good at it” comes down to the daily suspicion that at least half of what I do provides no tangible benefit to the patient. Which half? I simply don’t know.


***If you feel the need to go down the rabbit hole of the history of this factoid in medical education, here is a good start: J Pediatr. 1987 Apr;110(4):531-7. Predictive value of minor anomalies. I. Association with major malformations. Leppig KA, Werler MM, Cann CI, Cook CA, Holmes LB.

Tuesday, September 8, 2015

The Pulse Oximeter is Not the Problem....We Are (Guest Blog by Alan Schroeder)

"Who turned this patient’s oxygen off?”
“Uhhh…. Me?
“What were you thinking - you’re just an intern. You’re not allowed to do that!”
“I’m not? But he was only on 1/4L and his O2 sat was 98%”
“Well, it was 86% on room air a few hours ago!”


Rinse. Lather. Repeat.

If you’re like me, you had versions of this conversation many times during your residency, especially during winter months. The scenario went something like this: a 6 month-old infant hospitalized a few days ago with bronchiolitis is now better by all measures, but there’s that pesky bit of pulse ox data on the flowsheet that you’re being forced to confront on rounds. “Can’t send him home if he’s still needing oxygen. Who’s the next patient?”

For me, as a pediatrics resident in the late 90’s, this was as clear an example of injustice that I can remember. If we were evaluating this smiling, healthy-appearing and currently well-saturated infant in urgent care, we wouldn’t dream of hospitalizing him. But now here we are committing him and his family to another day in the hospital because the saturation dipped below 90% (or the threshold du jour) during a period of otherwise blissful sleep. We might toy with some modifications to the orders: “Do not place infant on oxygen unless saturations are < 90% for greater than 5 minutes despite suctioning and repositioning.” Or, if we are feeling really bold, we might even tolerate a few saturation points lower. But we definitely can’t send him home now.

When I finished residency I went on to an academic general pediatrics fellowship, and I was determined to draw attention to the “prisoner to the pulse ox” problem. We considered several study designs, and ultimately decided we would begin by doing a retrospective chart review and simply describing the problem. We found that pulse ox thresholds were all over the map, and we estimated that ~ ¼ of bronchiolitis hospitalizations are prolonged by at least one day because of a perceived need for supplemental oxygen based on pulse oximetry readings. The study,1 published in 2004, was far from perfect: single-center, small sample size, and no shortage of methodological limitations, but it really resonated with folks who were similarly frustrated by this problem.

A decade or so later, critical voices are starting to unify over the use of continuous pulse oximetry. One of the 2013 Choosing Wisely Top 5 recommendations from Pediatric Hospital Medicine was to avoid continuous pulse oximetry in infants not on oxygen, and the 2014 AAP bronchiolitis guidelines suggest consideration of intermittent pulse oximetry. The recommendations are fascinating, in that they reflect a somewhat surrendering acknowledgement that healthcare providers are simply not to be trusted with data. Instead of arming providers with better ways to react to information, let’s just shield them from it. A sad acknowledgement, really, but one that extends well beyond continuous pulse oximetry use in bronchiolitis.

In an unexpected turn of events, however, two studies published this year did not prove what they were supposed to prove. The first, by Schondelmeyer et al,2 was a quality improvement project from Cincinnati Children’s that attempted to decrease the use of continuous pulse oximetry in infants with bronchiolitis once they came off oxygen. The project was successful, in that continuous pulse oximetry use decreased substantially more on the intervention ward than on a control ward. But length-of-stay (LOS) was no different! This finding was disappointing given that the main point of getting rid of the darn continuous pulse ox was to reduce LOS. The second study, by McCulloh et al,3 was a multicenter randomized trial comparing a continuous vs intermittent pulse ox monitoring strategy in 161 infants hospitalized with bronchiolitis who no longer require oxygen. Once again, no impact on LOS!!!

What happened? In my mind, the link between continuous pulse oximetry use and prolongation of the hospital stay was about as causal as it gets – I have lived it each and every winter. So why was there no impact on LOS in these studies? As we are prone to do when we are confronted with results that do not conform to our preset beliefs, we look for holes. True to form, I think there are certainly some reasons why we shouldn’t dismiss the role that avoidance of continuous pulse oximetry can play in reduction of LOS.

The first thing that jumps out from the studies is that the LOS in each study group was short: ~2 days in the McCulloh trial, and closer to 1 day in Cincinnati (in comparison to 3.5 days in our 2004 study). Interestingly, McCulloh et al based part of their sample size calculation on an expected LOS derived from an older study involving two of the participating sites (sites that contributed the bulk of patients to the current study), where the mean LOS was almost a full day longer. Why did the LOS become shorter over time in these centers, irrespective of the pulse oximetry monitoring strategy? I wonder if part of the explanation may be that a new era of “permissive hypoxemia” emerged in the providers involved in bronchiolitis care in these institutions (where, notably, several of the authors of the Choosing Wisely recommendations served as PIs). Although both studies utilized pre-specified O2 sat thresholds of 90%, there is no reporting in either study of the lowest recorded oxygen saturations. Is it possible that saturations below 90% in the continuous monitoring arm of the McCulloh trial or in the control ward of the Schondelmeyer QI project were tolerated by the medical team? (“If this kid had been randomized to the intermittent arm, we wouldn’t even know about this 87% saturation while he was napping - so just send him home!”)

It’s easy to imagine that increased general awareness of the “prisoner to the pulse ox” phenomenon may have led to an evolution in practitioners’ impulsivity to react, thereby decreasing the probability that an intermittent monitoring strategy would shorten LOS. I suspect that in centers where the LOS is longer and where strict O2 saturation thresholds are enforced, that an intermittent monitoring strategy would have more of an impact on LOS. Additionally, sticking to intermittent monitoring throughout the hospitalization (study currently underway: https://clinicaltrials.gov/ct2/show/NCT01646606), rather than the tail-end, may get kids off oxygen and home sooner.

The authors of both studies are quick to point out that there are other important reasons to avoid continuous pulse oximetry, namely the increasingly described concerns about alarm fatigue in hospitals. Hard to disagree, and this topic is a hotbed of ongoing research. We can also agree that, based on these studies, intermittent pulse oximetry appears to be safe, though neither study was appropriately powered to detect rare serious events.

As the “prisoner-to-the-pulse ox” saga continues to unfold, I find that I myself am evolving in my perspective. I’m increasingly appreciating that some of our anger at the pulse ox may be misdirected; the problem isn’t with the pulse ox, it’s with us. We are simply unable to avoid reacting to the number, even if the number doesn’t seem to be telling the story of the patient in front of us. I’m all for abandoning tests or technologies that do nothing other than to generate a cascade of unnecessary interventions. But wouldn’t it be great if, in addition, we could teach ourselves to be less reactive to minor deviations from normal? With pulse oximetry, these recent studies suggest to me that maybe, just maybe, we’re starting to get there.

1. Schroeder AR, Marmor AK, Pantell RH, et al. Impact of pulse oximetry and oxygen therapy on length of stay in bronchiolitis hospitalizations. Archives of pediatrics & adolescent medicine 2004;158(6):527-30.
2. Schondelmeyer AC, Simmons JM, Statile AM, et al. Using quality improvement to reduce continuous pulse oximetry use in children with wheezing. Pediatrics 2015;135(4):e1044-51.
3. McCulloh R, Koster M, Ralston S, et al. Use of Intermittent vs Continuous Pulse Oximetry for Nonhypoxemic Infants and Young Children Hospitalized for Bronchiolitis: A Randomized Clinical Trial. JAMA pediatrics 2015.

Saturday, August 1, 2015

Better

I don’t know about you, but I've noticed that sometimes things don’t go entirely the way I’d like in my hospital. When I was younger, I used to mutter to myself, in those not-so-rare moments of frustration. I’d think: “if only (insert CEO, CMO, CQO or your authority figure of choice here) knew that this was happening, he or she would make it better.” Of course, what I was only beginning to guess was that no single person really, truly understands or manages most of what happens in a hospital. A hospital is a loose conglomeration of departments and service lines and sometimes it runs and sometimes it doesn’t. So when Paul Batalden famously quipped, “every system is designed to get exactly the results it achieves,” he was, of course, speaking less about design and more about its absence.

As I spent more years in hospitals, I developed my own personal riff on the experience. If I happened to be rounding and a student asked, “why did X or Y happen to patient Z when none of those things are (pick one or more) useful, helpful, necessary, or evidence-based?” I would answer with this; “well, if I teach you only one thing, let it be this – nobody is in charge.” Now, maybe you find that to be an overly pessimistic sentiment, but I always meant it to be empowering. Oh okay, I meant it to be a little bit pessimistic, but mostly empowering. Because the first step on anybody’s yellow brick road to “better” is to come to the understanding that nobody else is likely to fix it for you. The first and only rule at Oz General Hospital is this: if you are willing to own it, then you can fix it. All you need to do is care enough to accept responsibility. The wizard is otherwise engaged.

Much of this was driven home to me recently when I was precepting in a course where the graduating medical students are required to spend a month on a quality improvement project. I’ve taught in several different versions of the obligatory QI course for medical students, and I’ve always felt a little queasy about the experience. Mostly, I’ve worried that the students aren’t getting it, that they are still too much occupied by the idea that there are facts to be learned and all this QI stuff is not really their problem. It often seemed to me that we were teaching them the specifics of QI before they’d really internalized any belief in the need for it, and certainly before they saw it as their responsibility.

My last group of students, however, finally made a believer out of me. I gave them the task of updating a complex and frustrating screening protocol in our newborn nursery, frankly because I’d lost my own will to confront the problem. Early on, they struggled to believe that 1) our current process was not the best possible process and 2) that they were going to be able to come up with a solution that people would readily accept. To paraphrase their thought process: “if only (insert your authority figure of choice here) knew this was a problem, he or she would fix it… Oh, and if they find out you are letting medical students do this, don’t you think somebody will object?”

The first step, I thought, was to pull back the curtain and show them that there is no central intelligence running every detail of the show. This realization liberates us from believing someone is to blame for the problem, but it also frightens us to know that nobody is coming to save us. So, I trotted out the “nobody is in charge” mantra again and laid it on them. I said: fix the problem, or don’t, but don’t make the mistake of thinking somebody else will do it for you. They proceeded to get busy and delivered an improved protocol, helped implement it and even managed to collect a week’s worth of data showing a sharp decrease in unnecessary lab draws for babies as a result of their work – all in four weeks’ time. In short, they really couldn’t have done a better a job. But I still wasn’t actually sure they’d gotten something meaningful out of participating in the course. A bunch of high achievers did a good job on a complex task - not exactly a huge surprise. Again, I’m a pessimist, especially when it comes to the idea that I can possibly teach the things that really matter.

Months later, I got around to reading the essays they wrote as a course evaluation. And I was floored. There it was, exactly what I had hoped to teach. I am not even slightly embarrassed to say that I had tears in my eyes when I read these essays. Excerpts from two different students are illustrative:

“The first major lesson for me was brought to my attention in my small group. ‘No one is in charge.’ … Now, at first this sounded cynical and an excuse to be passive, but as we went through the course I began to see how valuable of a realization this was. The point really was that if any of us wants to make a difference in the future of healthcare, it is one hundred percent our responsibility. That no one person is sitting in the White House, or the American Medical Association, or even in the hospital, whose job is to make these sort of changes”


“When we started in our small group, we were told that we would be changing a hospital protocol that had been in place for several years before our arrival at this institution. My first reaction to this was skepticism. Students don't just come in and change protocols! CEOs change protocols! Department chairs change protocols! … Looking back, I am surprised that all of this (learning) came from such a simple task: “redesign this protocol” seemed relatively straightforward, even as the idea was daunting. Now that we have completed our mission, though, it feels simultaneously endlessly more complicated and vastly more doable than when we began. The phrase “No one is in charge” now means we all are.”

The students are gone. They’ve graduated and started residencies. I don’t even know which student wrote which essay because the feedback is anonymized. The most likely outcome is that I will never see any of them again. But they’re all out there, each one of them understanding that it is “endlessly more complicated and vastly more doable” than you think. I think they will make things better.

Friday, July 10, 2015

Seeing My Way to "No" (Guest Blog by Alan Schroeder)

When I asked my optometrist why I needed a dilated eye exam, his answer was… well, at least factually accurate: “So I can see inside the eye!” I was in his office because I have a nasty habit of losing my prescription eyeglasses and sunglasses, and when trying to get a replacement pair after this most recent occasion, I was informed that over 2 years had passed since I last had an eye exam. Neither my optometrist nor the nearby eyeglass retailer were “allowed” to sell me a new pair unless my prescription was updated. I was perfectly happy with my old prescription, so this struck me as a fairly paternalistic regulation, undoubtedly driven at least somewhat by financial factors. But, I was out of options and I made the appointment. However, eye dilation is a hassle, and from my rudimentary ophthalmologic knowledge I had at least some sense that the probability of benefit was minimal at best.

In an article I recently wrote with colleagues on overdiagnosis in children, we suggest that when a test is proposed, families ask their child’s doctor not just what the test is intended to detect but how such detection will lead to net benefit for the child. Clearly my optometrist hadn’t read the paper. I pressed him. I have no complaints, no concerning medical problems, and no relevant family history. “Doesn’t matter,” he said, “there can still be problems.” Then he pointed to the results from my visual field test (which he had not reviewed prior to stating that I needed a dilated exam): “See, you have some possible defects in your visual field here.” Uh-oh, he had played his trump card and now he was in the driver’s seat. I, the patient, have disease. Timidly, I pressed further, recalling that my effort during the visual field test was half-hearted at best. “What could cause that? Are you sure it’s accurate?” Already upset with me for having gone three years without an eye exam, and for continually refusing the fancy Retina Cam that was offered for a mere additional $30, he sighed, suggested a few unrecognizable conditions, and then added that it could be glaucoma. Now, I’m no eye doctor, but I was pretty sure that you don’t need to dilate the eyes to diagnose glaucoma. But I couldn’t help feeling that I had already crossed the sacred line in the doctor-patient relationship, and that it was time for me to stop being so difficult. For the remainder of the visit, I slipped back into “compliant patient” mode and did what I was told.

My inability to stick firmly with “no” has not been isolated to this visit or this doctor. Most of my visits with healthcare professionals (which I try to keep to a bare minimum) seem to involve some sort of negotiation where they want to do lots of stuff, but I don’t want any of it. Nonetheless, I usually wind up either surrendering or compromising. When I broke my arm, at each weekly visit the clinic staff would take off the cast and perform x-rays before the orthopedic surgeon even laid eyes on me. I wondered why they wouldn’t use the physical exam to guide the x-ray, but never had the courage to ask. Dental x-rays are a constant negotiation – with my dentist and my children’s.

I can appreciate why folks like me have a hard time saying “no” when they’re clueless about the test or intervention being proposed. When my mechanic lists everything that’s wrong with my car, how can I be sure that the problems mentioned won’t result in my car spontaneously catching fire or veering dangerously off the road - tomorrow?! When our contractor says we need a new roof, what do I know about roofs? But when it comes to matters surrounding my own health, I’m a fairly well-informed and skeptical physician. Why can’t I say “no”? Why do I hear stories like mine from other physician-patients, over and over? How are patients ever expected to question or to say “no” if we can’t? Why should we ever have to feel like we’re negotiating at the doctor’s office?

I wish I had all the answers. The standard proposed drivers of overtesting and overtreatment - fee-for-service, malpractice, time pressures, etc. - probably drive overly aggressive screening recommendations and/or doctors’ apparent inflexibility and persistence. But what about me, as a patient? I’m aware of all of these things, so why can’t I advocate for myself? That patients are afraid to “speak up” in the doctor’s office is not new news. A study in Health Affairs in 2012 concluded that collaborative discussion is often hampered by patients feeling compelled to conform to socially-sanctioned roles, fear of being categorized as “difficult”, and by “authoritarian” physician approaches. There are so many deeply embedded aspects of the medical culture that create the “helpless patient” phenomenon. We call physicians “Dr. ______” and they call us by our first names. After we get roomed, there’s this big buildup - “The doctor will be in to see you soon”. When they arrive, we feel lucky to be getting even a few minutes of their time, so best not to waste it by challenging them with questions. And, if I dare say “no”, what is he going to write about me in the chart? How will it affect the rest of my care? As a physician myself, I am not immune to these forces when I am a patient. In fact, perhaps there is a tacit understanding that I as a physician-patient should know better than to challenge my doctor, a professional code of sorts.

As for my dilated eye exam, what’s the big deal? Just do it, and get it over with, right? The problem is that I’ve been around long enough to realize that no test is benign, that any test can lead to a cascade of potentially harmful interventions. Is it possible that there is in fact solid evidence supporting biannual dilated eye exams in low-risk, asymptomatic patients? Maybe, though I doubt it (and is the onus really on me to dig it up?). Any benefit would be predicated on the idea that early detection of disease drives an intervention that is of proven benefit, that these benefits outweigh any harms that might occur either as a result of the test itself, or from false positives or overdiagnosis, and that the number-needed-to-screen is not astronomic. If true, then my optometrist failed, miserably, at an opportunity to make his case. Because of that, I’m in the market for a new optometrist, preferably one who, rather than starting the conversation with “Today we will be dilating your eyes,” instead begins with “Today I’d like to discuss a dilated eye exam.” I may be in for a long search.

References

Coon ER, Quinonez RA, Moyer VA, Schroeder AR. Overdiagnosis: How our compulsion for diagnosis may be harming children. Pediatrics 2014 Nov;134(5):1013-23


Frosch DL et al. Authoritarian Physicians And Patients’ Fear Of Being Labeled ‘Difficult’ Among Key Obstacles To Shared Decision Making. Health Affairs, May 2012; 31(5) 1030-8

Thursday, June 4, 2015

How it Feels to Fail

Recently, I had one of those weeks where you find that you just can’t catch a break. I discharged a patient who came back quite a bit sicker only 48 hours later. Granted, he danced out of the hospital eating popcorn the day I sent him home and I don’t plan to learn any universal rules of behavior from the failure….but it was a failure nonetheless, and it felt bad.

Yet, remarkably, its not that failure that I really want to talk about. I made another mistake that week as well, which I think was the more important one in the end, if only because it was entirely avoidable. I admitted a young child with classic lobar pneumonia who, after three weeks of illness, developed an effusion. The child’s primary care doctor had been doing an excellent job managing the illness outpatient, then simply ordered one too many follow-up films and was told by his reading radiologist that he’d uncovered the kind of thing that “must be drained.” The child had a white count of 30k and was reportedly persistently febrile as well, so I went into automatic mode. I accepted the patient for admission and began planning the next steps before the child even arrived at our hospital. I consulted my surgeons given the duration of illness and the possibility that the patient might even need a VATS. Generally our approach is to follow the APSA algorithm and only do VATS if the patient fails chest tube and fibrinolytics; however this patient was very young and had had a prolonged illness, so I was concerned he was already an outlier. My surgeons are amazingly judicious and I dearly love the way they practice, so I wasn’t surprised when they decided my patient did not need their services….but the patient’s parents were surprised, and deeply unhappy. They’d dropped everything and driven several hours to our referral center based on the impression that intervention was urgent. They’d been dealing with a sick child for three weeks and they were also highly sensitized to the x-ray and laboratory abnormalities, which continued to worsen throughout the course of illness, and now included anemia as well as profoundly elevated inflammatory markers. They’d run out of patience.

So, here is where the rubber meets the road. Now is the time to put up or shut up, right? I make my living pointing people to things like the wonderful work of Eyal Cohen and colleagues in Toronto showing that you really don’t have to drain these things and that a non-interventionist approach achieves equivalent outcomes (1). For heaven’s sake, I had the support of the surgeons! And, what did I do? I put a pigtail chest tube in the child. Looking back on it, I feel like I only looked at the patient himself after the fact. I certainly looked at his unhappy parents and I looked hard at his x-ray and his labs, all of which were problematic. Yet, the patient himself was absolutely fine. He never even had a fever in our hospital despite being admitted for nearly 24 hours prior to the procedure. If you passed him on the street you would not have guessed anything was amiss. After the fact, I’m certain it was an unnecessary procedure. And worse, I'm certain I could have helped his parents arrive at the same understanding.

So why did I do it? There’s the puzzle. It is not my usual practice to overtreat or to avoid a difficult conversation, but I did it this time and I needed to know why…which brings me back to the first mistake. I’d taken a less aggressive approach with another patient and although it was entirely justifiable, it didn’t work out. As I was making decisions about the patient with pneumonia I was impaired by my emotional response to that failure. Looking back, I realize I was not even using my rational brain. Several people even tried gently to stop me, and I missed the cues. I’ve often written about how it feels to stand in front of the onrushing train of unnecessary intervention and hold up your hand to stop it, but I’m not sure I’ve written about how it feels to be driving that train…..simply awful, all the way around.

So, how do I learn from it? Ultimately, I think the lesson is in how we think about thinking. Call it meta-cognition if you want to be fancy - but it’s a skill I was not exhibiting at the necessary time. I needed to stop and think about my own flawed thinking as it was occurring. I needed to be able to step outside myself and see my clinical decision making impaired by my emotional state. The question you may be asking yourself is: can that be done? People were clearly asking me to do it in this case. Instead, I engaged in motivated reasoning to continue to justify my chosen course. I was continually talking myself into it by trying to talk other people into it. I am certain that every physician has felt like this - you start out trying to traverse the thin edge of a decision and you simply slip one way or the other and suddenly you’ve fallen. You wake up at the bottom of the well looking up a distant point of light with no easy way out. So, you get busy explaining why you actually intended to end up at the bottom of the well and how the view from here isn’t so bad after all. That is how it feels to fail. What I hope for from this failure is to recognize the feeling a little earlier the next time around, so that I can at least start the long slow climb out of the well before it’s too late.

1. Cohen E, Mahant S, Dell SD, Traubici J, et al. The long-term outcomes of pediatric pleural empyema: a prospective study. Arch Pediatr Adolesc Med. 2012 Nov;166(11):999-1004.

Sunday, May 3, 2015

No Free Lunch

I just got back from a large pediatric meeting. On the way home, I happened to share a cab to the airport with an industry representative attending the same meeting. We chattered about the weather but I never found out who she worked for….though I had to wonder. This particular meeting attracts mostly academics and yet there are always hundreds of “exhibitors.” I was even regaled by a group of residents about the exceptional quality of the coffee proferred at each of the booths, and we shared a joke about whether the formula manufacturers use any of their product in the lattes. All fun and games, right? Except, maybe not. I remain exceptionally uncomfortable with the influence industry has in our profession, including pediatrics, and I can’t think of any valid reason why we allow it at our academic meetings, except, of course, that it pays a few of the bills. We accept industry involvement as normal, but should we? Advertising pervades our society, but we would do well to remember that there is a reason for the phrase, "there's no free lunch."

Conflict of interest is generally held to occur in a situation wherein a professional’s personal interests are in opposition to their professional obligations. The American College of Physicians ethics manual has this to say about financial considerations which pose a conflict of interest: “physicians must not allow such considerations to affect their clinical judgment.” Okay, good to know, sounds simple enough...except that realistically the only word I can use to describe that particular prohibition is “quaint.” While it would be lovely if it were that simple, it is more realistic to incorporate a modern understanding of how the human mind works - which inevitability requires us to question the idea that conflict of interest can be managed or mitigated.

We’ve learned a lot about human cognition in the past fifty years and it should inform how we think about the topic of conflict of interest. Blanket “thou shalt not” statements like the one taken from the ACP manual look painfully na├»ve in light of what behavioral psychology has taught us about our ability to allow or disallow select considerations to affect our judgment. The discovery that much of human behavior is heavily influenced by rapid, automatic “heuristic” thought systems which often operate in contrast to what we say motivates our behavior is a really big deal. Our two systems of thought (system one and system two, fast and slow, heuristic and logical) do not necessarily lead us to the same conclusions and are often found to be in conflict. In the case of conflict of interest, a large amount of behavioral psychology research now suggests that our thought processes around personal self-interest are very system 1 (rapid and heuristic) whereas our thought processes around our professional code of ethics are very system 2 (cognitively taxing, slow and logical). In general, behavioral psychology suggests that system 1 generally wins out unless a concerted effort to apply system 2 thinking is made, and even when that is accomplished, system 2 mitigates conflicts only incompletely. Many, many experiments now paint a picture of human cognition that leads me to believe that our view of our own ability to mitigate any conflict between personal and professional obligations is quite mistaken. The problem is that system 1 never even lets us in on the realization that we need to be using system 2. In the words of DA Moore (an experimental psychologist whose research focus is COI), “psychological research suggests that biased information processing is not only pervasive, but is typically unconscious and unintentional.” In the same paper, Moore and colleagues state:

“This view is compatible with what Chugh et al. call bounded ethicality: people routinely do
things that dispassionate observers would regard as unethical without intending to behave unethically or even considering the possibility that their behavior has ethical implications.”

Bascially, behavioral psychology suggests that conflict of interest is nearly impossible to mitigate because of the inherent processing characteristics of the human mind. Again, Moore says it best:

“when professional responsibilities clash with self-interest, the two motives tend to be processed differently: self-interest exerts a more automatic influence than do professional responsibilities, which are more likely to be invoked through controlled processing. Since automatic processing tends to occur outside of conscious awareness, its influence on judgment and decision making is difficult to eliminate or completely correct."

Therefore, even situations where potential for conflict of interest is clearly identified, publicly acknowledged and directly addressed, still typically lead to behavior that an impartial third party would consider “conflicted.” Regretfully, the evidence argues that professionals are mostly fooling themselves when we suggest we can mitigate conflict of interest. But it gets worse, the latest research suggests that personal relationships, no matter how casual, are often equally as potent a motivator as financial incentives (think drug reps). Human beings want to be liked and being liked is a potent incentive. Small gifts may even work better than large ones over the wider population. We turn on system 2 if the gift is large enough to get our attention, whereas system 1 obligates us to the giver of the lanyard, pen or desk do-dad without allowing us any overt awareness of the fact.

In the end, I suppose its lucky I never found out what my friendly cab-mate was selling because I’d probably be buying one right about now.


Moore DA, Tanlu L, Bazerman MH. Conflict of Interest and the Intrusion of Bias. Judgment and Decision Making, Vol. 5, No. 1, February 2010, pp. 37–53.

Moore DA, Loewenstein G. Self-interest, Automaticity and the Psychology of Conflict of Interest. Social Justice Research, Vol. 17, No. 2, June 2004


Saturday, April 4, 2015

The Blame Game (guest blog by Ricardo Quinonez)

Recently, while preparing for yet another Choosing Wisely talk, I was reviewing the portion of the talk that focuses on potential drivers of overuse. This time, one slide gave me pause – the one proposing that patient/parent demands are a significant driver of overuse. As with many of these proposed drivers, the topic is understudied and some things simply make the list because they make sense. They have face validity. After all, why wouldn’t patient/parent demands be one of the reasons doctors use unnecessary antibiotics or order unnecessary tests? I know I’ve felt that pressure, or at least used it as an excuse. But as it turns out, there are studies on the topic. For instance, a recent article in JAMA Oncology explored this very question.(1) The investigators reviewed 5050 patient encounters and evaluated patient demands during office visits. Out of these, only 8.7% of patients made demands to their treating physicians for treatments or test not already proposed to them. Of those who made these demands only 11.4% (50 patients), made clinically inappropriate ones. Physicians complied with these inappropriate demands only 14% of the time. In other words, of the total encounters only about 1% of them resulted in an inappropriate patient demand and 0.13% in inappropriate treatments or tests; hardly a poster child for overuse. Yet in my experience every time I’ve given a talk on this issue, the pressure of patient demands becomes a significant point of discussion. The comment prevails that parents, in our case, are demanding treatments or tests and that sometimes, for time’s sake; it’s simply easier to give in. Perhaps pediatrics is different. After all, we have not just the patient, but the advocate for the patient demanding that something, anything be done. So it begs the question: Is parental demand a significant driver of overuse? As is usually the case, someone already asked this question.

Back in 1999, Mangione-Smith et al surveyed physicians and their patients presenting with symptoms of an upper respiratory tract infection about their expectations before and after office visits.(2) Those expectations were then compared to inappropriate antibiotic prescribing by physicians. They also went a step further and studied how the “fulfillment” of expectations related to patient satisfaction. In simple terms: do patients come in with an expectation for specific care, and if they do and the physician fulfills this expectation, does the physician earn higher marks? They surveyed the care of 10 different physicians and 306 consecutive patient visits. Physician perception that the parent wanted an antibiotic was independently shown to be associated with 62% of cases of antibiotic prescription for presumed viral infections. Those physicians who presumed parents wanted antibiotics prescribed, were much more likely to assign a bacterial diagnosis to the patient (despite convincing evidence in the medical record of a likely viral source). However, physician’s perception of the caregiver’s desire to have an antibiotic prescribed did not match parental expectations in the majority of cases and it was this misperception that was associated with low ratings in parental satisfaction; not whether the patient received a prescription or not. Follow up work by the same investigators revealed that physician-patient interaction may be responsible for this phenomenon.(3,4) Although we often encourage patients to ask questions, the questioning of treatment plans by patients seems to be responsible for a misplaced belief by physicians that questioning equals a desire to have an antibiotic prescribed.

This misperception of parental expectations is not unique to pediatric outpatient practices. In a pediatric emergency room treating physicians misjudged parental expectations for antibiotic prescription for viral illnesses 66% of the time in one study and 75% in another and once again satisfaction scores were not tied to antibiotic prescription.(5,6) But patient-parental expectation is certainly not the only proposed driver of overuse that has face validity, but little basis in evidence. One of the most common excuses and one that I certainly propose as a potential driver in my talks is fear of litigation. Once again, it makes sense that doctors don’t want to be sued and in order to avoid being sued we test first and ask questions later. However, when this factor has been studied as a driver of overuse it does not seem to be as significant a cause as we think. Specifically, tort reform, where it has been enacted, has failed to decrease physician overutilization and costs.(7,8)

Other potential drivers of overuse are likely better candidates for blame. Physician’s own fear of uncertainty, supply sensitive care and our antiquated system of medical education are likely more reliable drivers than those we use to shift the blame to others. Certainly patients don’t seem to be the ones at fault. So as with many issues related to overuse, when looking for causes, it seems best to look in the mirror first.

1. Gogineni K, Shuman KL, Chinn D, Gabler NB, Emanuel EJ. Patient Demands and Requests for Cancer Tests and Treatments. JAMA Oncol. Published online February 12, 2015.
2. Mangione-Smith R, McGlynn EA, Elliott MN, Krogstad P, Brook RH. The relationship between perceived parental expectations and pediatrician antimicrobial prescribing behavior. Pediatrics. 1999 Apr;103(4 Pt 1):711-8.
3. Stivers T, Mangione-Smith R, Elliott MN, McDonald L, Heritage J. Why do physicians think parents expect antibiotics? What parents report vs what physicians believe. J Fam Pract. 2003 Feb;52(2):140-8.
4. Mangione-Smith R, Elliott MN, Stivers T, McDonald LL, Heritage J. Ruling out the need for antibiotics: are we sending the right message? Arch Pediatr Adolesc Med. 2006 Sep;160(9):945-52.
5. Karras DJ, Ong S, Moran GJ, Nakase J, Kuehnert MJ, Jarvis WR, Talan DA; EMERGEncy ID NET Study Group. Antibiotic use for emergency department patients with acute diarrhea: Prescribing practices, patient expectations, and patient satisfaction. Ann Emerg Med. 2003
6. Ong S, Nakase J, Moran GJ, Karras DJ, Kuehnert MJ, Talan DA; EMERGEncy ID NET Study Group. Antibiotic use for emergency department patients with upper respiratory infections: prescribing practices, patient expectations, and patient satisfaction. Ann Emerg Med. 2007 Sep;50(3):213-20.
7. Kavanagh KT, Calderon LE, Saman DM. The Relationship Between Tort Reform and Medical Utilization. J Patient Saf. 2013 Oct 7.
8. Thomas JW, Ziller EC, Thayer DA. Low costs of defensive medicine, small savings from tort reform. Health Aff (Millwood). 2010 Sep