Alan Schroeder
1. You have an interest in “overdiagnosis” and this is reflected in your presentations and articles. What do you think is currently the most frustrating example of overdiagnosis in pediatric hospital medicine?
I have gravitated to overdiagnosis in recent years because I believe the topic is generally under-recognized and under-appreciated. The notion that the detection of abnormalities is not synonymous with patient benefit is lost on many patients and doctors. If you obtained a diagnostic test, and you found what it is you were looking for, how can that be anything but good? We discuss and propose a number of examples of overdiagnosis in children in a Special Article in Pediatrics in 2014. One of the earliest proven examples had to do with screening for neuroblastoma – two studies published in NEJM demonstrated that if we screen universally for neuroblastoma with urinary catecholamines, we find a decent amount of asymptomatic stage I neuroblastoma, but by finding these early cases, we do not prevent late-stage neuroblastoma or mortality. So we are finding cases that never would have caused harm – they either remain latent or they regress.
The issue that lit an early fire in me in terms of looking at overuse and overdiagnosis was continuous pulse oximetry screening in bronchiolitis. During residency, it struck me as an injustice that an infant who was smiling, eating well, and breathing comfortably could not go home from the hospital because the O2 sat was 86% at 5AM. There is now abundant evidence that overdiagnosis of hypoxemia in bronchiolitis occurs frequently (Schuh JAMA 2014, Principi JAMA Peds 2016, Cunningham Lancet 2015). This phenomenon has led many organizations and hospitals to try to limit continuous pulse oximetry use, which is probably the right thing to do because there’s also the issue of alarm fatigue. However, increasingly I am realizing maybe we are the problem, not the pulse ox. Why are we so beholden to a specific O2 sat threshold? We tend not to have specific thresholds for RR, HR, temperature, feeding, etc. Part of the solution to mitigating the harms of overdiagnosis in general, and certainly in bronchiolitis, is getting practitioners to be more comfortable with minor deviations from what we conceive of as “normal.”
2. Along the same lines, a lot of recent activity on the listserv has focused on admission of RSV patients based on age and the natural course of the illness with the ability to worsen on days 3-5 of illness. Care to weigh in with your thoughts?
I am always trying to get trainees to unlearn what they think they know about the “classic trajectory” of bronchiolitis. Particularly given our recent understanding that ~30% of subjects with bronchiolitis are found to have 2 (or more!) viruses, I think it’s really hard to try to predict the disease course. I rarely base my clinical decisions on where the patient is in their illness, and I don’t think there’s any good evidence to support such a practice. Regarding age, as a general rule, I do worry about younger infants, so I would probably have a lower threshold to admit a 6 week-old with the same degree of distress as a 6 month old, especially because of the risk of apnea in younger infants with bronchiolitis. However, as we have learned recently, the risk of apnea does not vary by viral subtype, so RSV positivity does not influence my decision-making on that front.
3. You participated in an entertaining debate on pediatric dogmas at PHM 2016. What is your favorite (or least favorite) dogma and why?
It’s funny, I took the “pro” side of testing and treating for Strep throat during that debate. I took the pro side mostly because I do believe that it’s worth trying to prevent transmission of strep and it’s important to get your kid back to school. I titled the talk, “Why I would have my own child tested and treated for Strep (if sick enough),” but I had never had to make that decision for my own children until about a month ago. My 6 year-old was absolutely miserable with a high fever and abdominal pain, and she had an exudate on her tonsils. I bit the bullet and had her tested (+) and we put her on amoxicillin. Interestingly, she was completely back to normal within about 12 hours, but then on day 6 of antibiotics she broke out in a drug rash.
I think if I had to choose one dogma that gets me going the most, it’s the notion that IV antibiotics are superior to oral antibiotics. I am perfectly happy prescribing IV antibiotics to the kid who presents with purpura and hypotension, but when we are faced with a kid who has recovered from his or her bacterial infection, and we deem that an additional x days of IV treatment are needed (often with PICC/sedation), I think it’s baffling that we have so readily accepted such a practice (which carries very relevant risks and costs) without any supportive evidence.
4. You practice both as a hospitalist and in pediatric critical care. How does your role in the PICU help you as a hospitalist?
I love being able to work both as an intensivist and a hospitalist.The benefits of being trained in critical care have some obvious advantages – being well-trained and very comfortable with sedation is one example. There are also some disease entities such as congenital heart disease where we just have a bit more training and exposure, and that can be quite helpful – both in terms of feeling comfortable taking care of these patients but also in terms of being able to explain the conditions and the surgeries to trainees and families. Finally, when we inherit patients from the PICU onto the hospitalist team, it is probably a little easier for me to see the rationale for some of the things that may have happened during the PICU stay. Conversely, I have learned a lot from my hospital medicine colleagues, many of whom have incredible clinical skills, and also have very solid expertise in quality improvement, coordination of care for complex patients, medical education, etc.
5. You co-authored an article in Hospital Pediatrics entitled, “Is Tradition Trumping Evidence in the Treatment of Young, Febrile Infants?” In the article, you write “discussions will continue, as they should, over whether the preferred initial regimen should be ampicillin plus gentamicin or cefotaxime alone, and whether reports of recent Listeria outbreaks should factor into this decision.” What is your personal practice on empiric antibiotics in young febrile infants?
I think smart and reasonable folks can disagree on this one. I generally err towards amp + gent over cefotaxime. That said, people say that amp + gent is narrower than cefotaxime, but I’ve never been given a convincing answer as to whether in this case the choice of two narrower spectrum antibiotics is definitely more “stewardly” than the one broader spectrum antibiotic. There are also increased costs (and potential for error) of administering two antibiotics instead of one, especially if the amp is dosed q 6 hours. But given some of the studies in neonates suggesting worse outcomes with 3rd gen cephalosporins (data that admittedly is almost certainly confounded by indication), and given the sense that cefotaxime may be overly broad, I usually stick with amp and gent. My main beef, though, is routine addition of amp to cefotaxime, where the NNT for ampicillin is unacceptably high.
To me, a much more important question is why we refuse to use ceftriaxone under 1 month of age – I can appreciate this concern in a jaundiced baby, but in a non-jaundiced febrile 3 week old?? If this is the main issue keeping us from moving towards outpatient management in the < 1 month age group, then this is indeed a “dogma” that warrants closer examination.
6. According to your Stanford bio, you have authored 35 papers! What advice do you have for others out there in the field of hospital medicine looking to get more involved in clinical research and publications? Is there a secret to tailoring research to make it more publishable?
ONE: Research questions should be organically motivated. It is imperative that when you are asking your research question, you really want to know the answer. I love doing research. It is incredibly satisfying for me to write a “do file” in Stata, plug in my data, and await the results. I have been keeping a list of “questions in need of answers” (unofficially, my “QUINOA” list) for the last 5-6 years. These are issues that come up on rounds, at conferences, while reading other papers, etc, that I hope to one day be able to study. I think it’s around 50 questions long at this point. I will often review this list with trainees interested in doing research projects with me, and I encourage trainees and faculty interested in clinical research to make their own lists.
TWO: Get used to disappointment. I should make it very clear that I’ve had PLENTY of rejected manuscripts in my time. I don’t think the disappointment that occurs when I get that “We regret to inform you…” email has attenuated one bit over my career. Intermittent disappointment (ideally counterbalanced by the intermittent success) is just part of the process.
THREE: Don’t be sloppy! Be meticulous with your design and execution, and be meticulous when you write the manuscript. I review a lot of manuscripts (for other journals and as an editor for Hospital Pediatrics), and if I am reading a paper that has obvious mistakes or typos, I am going to be very reluctant to recommend anything other than “reject” even if the study question and the findings are important. My rationale is that if the authors were sloppy with the manuscript, then they were probably also sloppy with the study itself, and I will have a hard time trusting the findings. I would guess that the average number of revisions for papers that I have been involved with before the initial submission to a journal is between 5-10.
FOUR: Avoid making misattributions of causality. With the increasing availability and allure of big data, it’s tempting to try to examine these databases to see if X intervention is associated with Y outcome. Breastfeeding studies are probably the best example of this – there is really no need for any more population studies assessing the “benefit” of breastfeeding, in my humble opinion. The problem of residual confounding in these studies just will not go away. Although experimental designs often take more time, expertise, and money, I think it’s well worth it.
FIVE: Get your name out there. When people ask you to author or co-author something, whether it’s a review article, a commentary, a book chapter, whatever, DO IT! The more you write, the easier it gets, and the more your name gets out there, the more folks will want you to write stuff. There are lots of ways to write for academic journals without doing research (e.g. Hospital Pediatrics’ “Bending the Value Curve”, JHM’s “Things we do for no reason”, etc.)
SIX: (this is probably the most obvious) Find a good mentor. I was very lucky to start off my clinical research career with Tom Newman as a mentor, and 15 years later I still lean on him – frequently! Few people will be so lucky as to have a Tom Newman, but it’s important to try.
7. Jack Percelay often ends his list-serve commentary with the phrase “that’s just my 2 pennies.” What are your two cents?
Well, I for one would pay more than 2 pennies for Jack’s advice. In fact, I sought out his advice very early on in my career! I had the privilege of attending the first PHM conference in San Antonio, which I believe was in 2002. The keynote speaker was a protégé of Bob Wachter and Lee Goldman from UCSF. Drs. Goldman and Wachter are often viewed as the grandfathers of the field of hospital medicine, and this speaker passed along their recipe for success in this field. The idea is that our careers are divided into systole and diastole, where systole is our daily clinical grind, and diastole is everything else that we do. In order to ensure having a sense of purpose and avoiding burnout, it is important to have sufficient time (and fulfillment) in diastole. Diastolic activities can include research, med ed, QI, advocacy work, committee work, leadership, etc. I have had the good fortune of enjoying both my systolic and diastolic work immensely, and in many ways I use the systole to drive diastole, in that many of my research questions are derived directly from my clinical work. I pass along this career cardiac cycle model to many of my mentees and trainees, and hope it is useful for readers of the Hospitalist Corner.
I have gravitated to overdiagnosis in recent years because I believe the topic is generally under-recognized and under-appreciated. The notion that the detection of abnormalities is not synonymous with patient benefit is lost on many patients and doctors. If you obtained a diagnostic test, and you found what it is you were looking for, how can that be anything but good? We discuss and propose a number of examples of overdiagnosis in children in a Special Article in Pediatrics in 2014. One of the earliest proven examples had to do with screening for neuroblastoma – two studies published in NEJM demonstrated that if we screen universally for neuroblastoma with urinary catecholamines, we find a decent amount of asymptomatic stage I neuroblastoma, but by finding these early cases, we do not prevent late-stage neuroblastoma or mortality. So we are finding cases that never would have caused harm – they either remain latent or they regress.
The issue that lit an early fire in me in terms of looking at overuse and overdiagnosis was continuous pulse oximetry screening in bronchiolitis. During residency, it struck me as an injustice that an infant who was smiling, eating well, and breathing comfortably could not go home from the hospital because the O2 sat was 86% at 5AM. There is now abundant evidence that overdiagnosis of hypoxemia in bronchiolitis occurs frequently (Schuh JAMA 2014, Principi JAMA Peds 2016, Cunningham Lancet 2015). This phenomenon has led many organizations and hospitals to try to limit continuous pulse oximetry use, which is probably the right thing to do because there’s also the issue of alarm fatigue. However, increasingly I am realizing maybe we are the problem, not the pulse ox. Why are we so beholden to a specific O2 sat threshold? We tend not to have specific thresholds for RR, HR, temperature, feeding, etc. Part of the solution to mitigating the harms of overdiagnosis in general, and certainly in bronchiolitis, is getting practitioners to be more comfortable with minor deviations from what we conceive of as “normal.”
2. Along the same lines, a lot of recent activity on the listserv has focused on admission of RSV patients based on age and the natural course of the illness with the ability to worsen on days 3-5 of illness. Care to weigh in with your thoughts?
I am always trying to get trainees to unlearn what they think they know about the “classic trajectory” of bronchiolitis. Particularly given our recent understanding that ~30% of subjects with bronchiolitis are found to have 2 (or more!) viruses, I think it’s really hard to try to predict the disease course. I rarely base my clinical decisions on where the patient is in their illness, and I don’t think there’s any good evidence to support such a practice. Regarding age, as a general rule, I do worry about younger infants, so I would probably have a lower threshold to admit a 6 week-old with the same degree of distress as a 6 month old, especially because of the risk of apnea in younger infants with bronchiolitis. However, as we have learned recently, the risk of apnea does not vary by viral subtype, so RSV positivity does not influence my decision-making on that front.
3. You participated in an entertaining debate on pediatric dogmas at PHM 2016. What is your favorite (or least favorite) dogma and why?
It’s funny, I took the “pro” side of testing and treating for Strep throat during that debate. I took the pro side mostly because I do believe that it’s worth trying to prevent transmission of strep and it’s important to get your kid back to school. I titled the talk, “Why I would have my own child tested and treated for Strep (if sick enough),” but I had never had to make that decision for my own children until about a month ago. My 6 year-old was absolutely miserable with a high fever and abdominal pain, and she had an exudate on her tonsils. I bit the bullet and had her tested (+) and we put her on amoxicillin. Interestingly, she was completely back to normal within about 12 hours, but then on day 6 of antibiotics she broke out in a drug rash.
I think if I had to choose one dogma that gets me going the most, it’s the notion that IV antibiotics are superior to oral antibiotics. I am perfectly happy prescribing IV antibiotics to the kid who presents with purpura and hypotension, but when we are faced with a kid who has recovered from his or her bacterial infection, and we deem that an additional x days of IV treatment are needed (often with PICC/sedation), I think it’s baffling that we have so readily accepted such a practice (which carries very relevant risks and costs) without any supportive evidence.
4. You practice both as a hospitalist and in pediatric critical care. How does your role in the PICU help you as a hospitalist?
I love being able to work both as an intensivist and a hospitalist.The benefits of being trained in critical care have some obvious advantages – being well-trained and very comfortable with sedation is one example. There are also some disease entities such as congenital heart disease where we just have a bit more training and exposure, and that can be quite helpful – both in terms of feeling comfortable taking care of these patients but also in terms of being able to explain the conditions and the surgeries to trainees and families. Finally, when we inherit patients from the PICU onto the hospitalist team, it is probably a little easier for me to see the rationale for some of the things that may have happened during the PICU stay. Conversely, I have learned a lot from my hospital medicine colleagues, many of whom have incredible clinical skills, and also have very solid expertise in quality improvement, coordination of care for complex patients, medical education, etc.
5. You co-authored an article in Hospital Pediatrics entitled, “Is Tradition Trumping Evidence in the Treatment of Young, Febrile Infants?” In the article, you write “discussions will continue, as they should, over whether the preferred initial regimen should be ampicillin plus gentamicin or cefotaxime alone, and whether reports of recent Listeria outbreaks should factor into this decision.” What is your personal practice on empiric antibiotics in young febrile infants?
I think smart and reasonable folks can disagree on this one. I generally err towards amp + gent over cefotaxime. That said, people say that amp + gent is narrower than cefotaxime, but I’ve never been given a convincing answer as to whether in this case the choice of two narrower spectrum antibiotics is definitely more “stewardly” than the one broader spectrum antibiotic. There are also increased costs (and potential for error) of administering two antibiotics instead of one, especially if the amp is dosed q 6 hours. But given some of the studies in neonates suggesting worse outcomes with 3rd gen cephalosporins (data that admittedly is almost certainly confounded by indication), and given the sense that cefotaxime may be overly broad, I usually stick with amp and gent. My main beef, though, is routine addition of amp to cefotaxime, where the NNT for ampicillin is unacceptably high.
To me, a much more important question is why we refuse to use ceftriaxone under 1 month of age – I can appreciate this concern in a jaundiced baby, but in a non-jaundiced febrile 3 week old?? If this is the main issue keeping us from moving towards outpatient management in the < 1 month age group, then this is indeed a “dogma” that warrants closer examination.
6. According to your Stanford bio, you have authored 35 papers! What advice do you have for others out there in the field of hospital medicine looking to get more involved in clinical research and publications? Is there a secret to tailoring research to make it more publishable?
ONE: Research questions should be organically motivated. It is imperative that when you are asking your research question, you really want to know the answer. I love doing research. It is incredibly satisfying for me to write a “do file” in Stata, plug in my data, and await the results. I have been keeping a list of “questions in need of answers” (unofficially, my “QUINOA” list) for the last 5-6 years. These are issues that come up on rounds, at conferences, while reading other papers, etc, that I hope to one day be able to study. I think it’s around 50 questions long at this point. I will often review this list with trainees interested in doing research projects with me, and I encourage trainees and faculty interested in clinical research to make their own lists.
TWO: Get used to disappointment. I should make it very clear that I’ve had PLENTY of rejected manuscripts in my time. I don’t think the disappointment that occurs when I get that “We regret to inform you…” email has attenuated one bit over my career. Intermittent disappointment (ideally counterbalanced by the intermittent success) is just part of the process.
THREE: Don’t be sloppy! Be meticulous with your design and execution, and be meticulous when you write the manuscript. I review a lot of manuscripts (for other journals and as an editor for Hospital Pediatrics), and if I am reading a paper that has obvious mistakes or typos, I am going to be very reluctant to recommend anything other than “reject” even if the study question and the findings are important. My rationale is that if the authors were sloppy with the manuscript, then they were probably also sloppy with the study itself, and I will have a hard time trusting the findings. I would guess that the average number of revisions for papers that I have been involved with before the initial submission to a journal is between 5-10.
FOUR: Avoid making misattributions of causality. With the increasing availability and allure of big data, it’s tempting to try to examine these databases to see if X intervention is associated with Y outcome. Breastfeeding studies are probably the best example of this – there is really no need for any more population studies assessing the “benefit” of breastfeeding, in my humble opinion. The problem of residual confounding in these studies just will not go away. Although experimental designs often take more time, expertise, and money, I think it’s well worth it.
FIVE: Get your name out there. When people ask you to author or co-author something, whether it’s a review article, a commentary, a book chapter, whatever, DO IT! The more you write, the easier it gets, and the more your name gets out there, the more folks will want you to write stuff. There are lots of ways to write for academic journals without doing research (e.g. Hospital Pediatrics’ “Bending the Value Curve”, JHM’s “Things we do for no reason”, etc.)
SIX: (this is probably the most obvious) Find a good mentor. I was very lucky to start off my clinical research career with Tom Newman as a mentor, and 15 years later I still lean on him – frequently! Few people will be so lucky as to have a Tom Newman, but it’s important to try.
7. Jack Percelay often ends his list-serve commentary with the phrase “that’s just my 2 pennies.” What are your two cents?
Well, I for one would pay more than 2 pennies for Jack’s advice. In fact, I sought out his advice very early on in my career! I had the privilege of attending the first PHM conference in San Antonio, which I believe was in 2002. The keynote speaker was a protégé of Bob Wachter and Lee Goldman from UCSF. Drs. Goldman and Wachter are often viewed as the grandfathers of the field of hospital medicine, and this speaker passed along their recipe for success in this field. The idea is that our careers are divided into systole and diastole, where systole is our daily clinical grind, and diastole is everything else that we do. In order to ensure having a sense of purpose and avoiding burnout, it is important to have sufficient time (and fulfillment) in diastole. Diastolic activities can include research, med ed, QI, advocacy work, committee work, leadership, etc. I have had the good fortune of enjoying both my systolic and diastolic work immensely, and in many ways I use the systole to drive diastole, in that many of my research questions are derived directly from my clinical work. I pass along this career cardiac cycle model to many of my mentees and trainees, and hope it is useful for readers of the Hospitalist Corner.