Go to an inexpensive medical school

May 13th, 2011

Over the last two weeks, I have been engaged in an email discussion with a young woman who is trying to decide where to attend medical school. I shall call her Ms. X. Ms. X has been accepted to numerous schools and has narrowed her options to two: School 1 is a private school in Boston (not Harvard), and school 2 is a public school in suburban New York City in her home state of New York. While the private school in Boston is ranked slightly more highly than the public school in suburban New York City according to US News and World Report (not a terribly reliable source, but anyways…), the public school in suburban New York City ranks within about 20 slots in both the primary care and research categories. What is the not-so-surprising kicker? The annual tuition and fees at the private school in Boston is approximately $22k more than the public school in suburban New York City.

Choosing the private school in Boston over the public school in suburban New York city would cost Ms. X approximately $88k additional in tuition and fees alone over the 4 years of medical school, all other things being equal, which, of course, they are not because the cost of living in Boston is higher than in suburban New York City. For the sake of this discussion, let’s assume that the cost of living would be $3k more per year in Boston than in suburban New York City for a total of an even $100k difference over the 4 years of medical school. Ms. X has expressed one factor that is pulling her towards the public school in suburban New York City: cost. She has expressed two factors that are pulling her towards the private school in Boston: 1) ranking and 2) she really liked this school better. I gave Ms. X the same advice that I give every pre-med whom I meet: If you are paying for medical school yourself, then go to the least expensive school at which you get accepted. With Ms. X, though, I gave the advice more emphatically as Ms. X has stated time and again that she wants to go into primary care.

Why do I tell pre-meds to go to the least expensive school possible (i.e., if there is a large difference in cost between the schools from which they are choosing)? In my circuitous and protracted path from medical school graduation to family medicine residency completion, I have (admittedly anecdotally) observed that how one performs in medical school (and on USMLE exams) is much more important in landing a competitive residency than where one attended medical school (if anyone has empiric data or knows of published studies on this topic, then please let me know… a quick PubMed search didn’t turn up anything useful). Granted, a bottom-of-the-class student at Harvard will likely have more opportunities than a bottom-of-the-class student at a low-reputation (public or private) medical school, but the top students at nearly any school will likely have similar access to the most competitive residencies. It seems to be much more important HOW one does in medical school than WHERE one does it. And Ms. X is not the kind of person who will do poorly in medical school. This is why I don’t pay much heed to the differences in the US News rankings between Ms. X’s two options, aside from the fact that these two schools’ rankings are actually quite similar relatively speaking anyhow.

Regarding Ms. X’s liking the private school in Boston better than the public school in suburban New York City, only she can place a monetary value on that, but $100k is a lot of money, which brings me to my next point. If Ms. X is being honest with me, and I think that she is, then her dream is to be a primary care physician. We all know that primary care physicians don’t make the big bucks, and many of us have heard or read stories of would-be primary care physicians who ended up pursuing more highly-paying specialties largely because of their perceived inability to manage their medical school debt burden on a primary care salary.

Ms. X has been getting a lot of advice from orthopedic surgery residents whom she knows. Their consensus opinion is to go to the BEST medical school that she got into and that the money will take care of itself. While I’m not convinced that performing equally well at a slightly lesser-ranked medical school would close any doors, even for a would-be orthopedic surgeon, their vantage point is completely different from that of a would-be primary care physician. If your dream is to be an orthopedic surgeon where the competition for residency slots is fierce and where you will likely be making a lot of money when you are in practice, then maybe it’s worth $100k to go to a slightly more highly-ranked medical school to give you that tiny edge in achieving your dream. But if your dream is to be a primary care physician, then it seems to me that an additional $100k in debt would actually be a significant deterrent to achieving your dream.

Ms. X is an intelligent, thoughtful, and very motivated young woman. I have little doubt that she will make the decision that is best for her and achieve whatever she sets out to achieve. But medical school can be very expensive, and I believe that for anyone considering becoming a physician, it is important to consider one’s goal and how the cost of medical school might impact one reaching that goal before choosing a medical school.

The American College of Cardiology’s Cardiac Stress Testing Appropriate Use Criteria Answer the Wrong Question

January 16th, 2011

by Adam Rothschild, M.D., M.A.

Dr. William Follansbee is the chairman of the American College of Cardiology/American Society of Nuclear Cardiology (ACC/ASNC) task force on non-invasive cardiac imaging and the director of nuclear cardiology at the University of Pittsburgh Medical Center Cardiovascular Institute. He recently published an editorial in the Pittsburgh Post Gazette in which he criticized the local Blue Cross/Blue Shield carrier, Highmark, for restricting the use of nuclear cardiac stress testing in favor of sonographic cardiac stress testing (a.k.a. stress echocardiography). Dr. Follansbee made several arguments as to why he believed that Highmark’s restriction of nuclear cardiac stress testing was wrong. One of his core arguments is that “patients will be…denied access to appropriately indicated nuclear cardiology tests ordered by their physicians” (emphasis is mine). He (indirectly) references the ACC’s 2009 Appropriate Use Criteria for Cardiac Radionuclide Imaging in support of this argument. This document identifies clinical scenarios where a group of experts reached consensus that nuclear cardiac stress testing was appropriate. Dr. Follansbee fails to mention that the ACC also publishes an analogous document called 2008 Appropriateness Criteria for Stress Echocardiography, which uses the same methodology to identify clinical scenarios where a group of experts reached consensus that sonographic cardiac stress testing was appropriate and which illustrates that indications and test performance characteristics for nuclear and echocardiographic stress testing are virtually the same. That said, neither of these ACC documents explicitly identifies where nuclear cardiac stress testing is preferable to sonographic cardiac stress testing and vice versa.

Appropriateness criteria give a semblance of objectivity as to when cardiac stress testing is indicated, but since there are many situations in which the ACC suggests that both nuclear and echocardiographic stress testing are indicated, they give no explicit guidance on which modality is preferred in a situation in which when they are both indicated. The question that doctors like me actually need answered is two-fold: For a given clinical situation 1) is a cardiac stress test indicated and 2) should I order a nuclear or echocardiographic stress test? Since the ACC cardiac stress testing appropriateness criteria guidelines did not address which testing modality is preferred where they are both indicated, Highmark did. Although Highmark’s answer apparently did not meet Dr. Follansbee’s approval, their preference of sonographic over nuclear cardiac stress testing actually makes a lot of sense.

All other things being equal, a safer test is preferable to a riskier test, and a less expensive test is preferable to a more expensive test. Guess what? With respect to “all other things”, nuclear and sonographic cardiac stress testing are essentially equal; even the ACC’s appropriateness criteria say so: “The overwhelming majority of final ratings of cardiac RNI [i.e., nuclear stress testing] and stress echocardiography [i.e., sonographic stress testing] were concordant for similar clinical indications.” Nuclear and sonographic cardiac stress testing have nearly identical sensitivities and specificities; if anything, the specificity of sonographic cardiac stress testing even seems to be somewhat better. Guess what else? Sonographic cardiac stress testing is safer than nuclear cardiac stress testing because it does not use carcinogenic ionizing radiation, and sonographic cardiac stress testing is significantly less expensive than nuclear cardiac stress testing. As such, Highmark’s preference of sonographic cardiac stress testing over nuclear cardiac stress testing is appropriate.

It’s no longer news that America’s health care costs are high and rising with no improvement in our mediocre quality. Through its justified preference of sonographic to nuclear cardiac stress testing, Highmark has identified a way to safely improve (or at worst not decrease) quality and simultaneously decrease cost. This is win-win and a powerful example of comparative effectiveness research, although not referred to as such. I do not know whose interests Dr. Follansbee primarily had at heart in arguing for less safe, more expensive, and no more accurate nuclear cardiac stress testing over sonographic cardiac stress testing, but it seems to me that it was neither individual patients nor society at large. Regardless, the ACC should issue a guideline that explicitly states when sonographic cardiac stress testing is preferred to nuclear and vice versa. Sure, this would likely anger many cardiologists and nuclear medicine physicians who make a nice living performing nuclear stress testing, but we should practice medicine with our patients’ best interests primarily at heart, not our own.

The Electronic Medical Record System Innovator’s Paradox

December 22nd, 2010

by Adam Rothschild, M.D., M.A.

At Doctrelo we are building an electronic medical record system (EMR) for primary care that we believe is so much better than any other product out there that we don’t even like to call it an EMR system. Instead, we call it the Doctrelo Clinical Command System. Rather, I should say that we WANT to build the Doctrelo Clinical Command System, but some industry veterans tell us that we likely won’t be able to raise the capital that we think that we need to do so.

Why do they think that investors likely won’t fund us to develop our product?  Their main reason is that there already exist many ambulatory EMR products in a “competitive and crowded” market, several from large vendors with mature products and significant resources. I acknowledge that this is true… but only sort of. Yes, several large vendors do indeed have well-supported products that implement a set of features whose utilization thereof has been deemed “meaningfully useful”. I acknowledge that these products exist, but who is USING these products, or, more poignantly, who is NOT using these products? Not using these products are 90% of our nation’s primary care physicians (PCPs) in practices of 5 or fewer physicians. This comprises 120k or so physicians, which is more than 40% of our nation’s PCPs.

So we’ve now established that large companies sell “mature”, meaningful use certified ambulatory EMR products, yet the majority of our nation’s primary care physicians still prefer to use pen and paper. Why? Researchers have studied this topic repeatedly in recent years with cost consistently arising as the number one barrier. Concluding simply that cost is the main barrier, however, misses the point. The main reason why the majority of our nation’s PCPs still prefer to use paper and pen is that existing ambulatory EMR systems suck… or to put it more scientifically, they lack sufficient value. While early-adopting PCPs in small practices perceive sufficient value in existing EMR systems such that they have been willing to purchase them, the other 90% do not.

Value is defined as utility divided by price. If the goal is for all PCPs to adopt EMR systems, then there are two possible ways to overcome this barrier of lack of value: 1) increase the utility of the EMR systems (i.e., build better EMR systems) or 2) decrease their price. When the politico-economic stars aligned such that the government was looking for new ways to spend lots of money quickly (i.e., stimulate the economy), Congress passed the HITECH Act. The (large-EMR-vendor-written) HITECH Act  functionally decreased the price of EMR systems by paying physicians for purchasing an EMR system and jumping through the (admittedly reasonable and well-intended) hoop of meaningful use. In short, the government is subsidizing the purchase of EMR systems that the market, itself, has empirically shown to be of low value. So much for promoting innovation.

The large EMR vendors who wrote the HITECH Act are ecstatic that the government is subsidizing the purchase of their admittedly non-innovative products, but it is reasonably clear that universal “meaningful use” of existing EMR systems will have minimal impact on increasing healthcare quality and decreasing healthcare costs. That’s fine for stimulating the economy but not so fine for solving our nation’s healthcare woes. Along comes Doctrelo (and maybe other startup companies, too) with plans for a product that has unprecedented potential in increasing healthcare quality and decreasing healthcare costs and does so in such a way that physicians actually want to use the software that we want to build, a product whose value is high because its utility is high, not because its price is artificially low due to government subsidies (or ethically questionable practices). This leaves us with the EMR system innovator’s paradox: We can build a better EMR system that will (likely) improve healthcare quality and decrease healthcare costs, but we (maybe) can’t get the money that we need to build it. A corollary paradox: Existing EMR systems (likely) won’t decrease healthcare costs or increase healthcare quality, yet the government is spending a LOT of money to get physicians to purchase them. Again, this is fine for economic stimulus (and big EMR system vendor profits) but not so great for American healthcare.

What are we at Doctrelo going to do? First, we are going to continue with the planned release of our first product, a unique, stand-alone e-prescribing “plus” system called Doctrelo eRx+ in early 2011. Second, we are going to build the Doctrelo Clinical Command System. Thanks, naysayers, for your concern, but I’ve been dreaming about this for too long (I first started working on this during my informatics fellowship in 2003), and my team and I believe in this too much;  giving up is just not an option. If we can raise capital, great. If we can’t raise capital, then we will still build the product. We’re not going to let a few million dollars get in the way of significantly changing American healthcare for the better.

Full disclosure: My research adviser during my informatics training at Columbia was George Hripcsak, who is the co-chair of the Meaningful Use Workgroup of the HIT Policy Committee of the Office of the National Coordinator of Health Information Technology (i.e., the group that came up with the meaningful use criteria). It was with George that I first started thinking about applying the problem-driven workflow to primary care.

The 6 P’s: Who are the Users (and Re-users) of Patient-specific Clinical Information?

November 19th, 2010

by Adam Rothschild, M.D., M.A.


Electronic health record (EHR) systems are big news (and big money) these days. Our government is pushing adoption and “meaningful use” of EHR systems because computerizing patient-specific clinical information can benefit our nation’s health care system. The primary focus has been on the benefit to the care of individual patients, but computerizing patient-specific clinical information can also benefit other parties in the health care systemi. I believe that there are 6 major parties that serve to benefit from the computerization of and access to patient-specific clinical information. I refer to these parties as the 6 P’s.

The six P’s are as follow:

  1. patient
  2. provider
  3. payer
  4. public (as in public health)
  5. posterity (alternatively PhD’s)
  6. product makers (alternatively pharma).


At this juncture in the national discussion of EHR systems, it should be self-evident why patients serve to benefit from the computerization of patient-specific clinical information. I will thus not further belabor the point.


Ditto for provider. By “provider” I refer to all individuals and organizations that deliver care to patients, including but not limited to physicians, nurse practitioners, nurses, physical therapists, hospitals, home health agencies, etc.


By “payer” I refer to third parties that pay for patients’ care. I use the term broadly and include both for-profit and not-for-profit health insurance carriers, government-run health insurance programs such as Medicare and Medicaid, and even the many self-insured corporations that utilize health insurance companies solely for their administrative functions but not for their assumption-of-risk functions.

Payers will ideally be able to automatically monitor the quality of care that their patients are receiving. While many payers currently analyze claims data for this purpose, actual patient-specific clinical data will likely provide a much more precise and useful dataset. Knowledge of the quality of the goods and services that payers are purchasing should help them to design programs to improve value for their covered patients.


By “public” I refer to the population overall, as in public health, whether on a local, regional, national, or even international level. The potential public health applications of having large quantities of computationally tractable patient-specific clinical information are enormous. Examples include real-time detection of emerging infections, simplified identification of locations of environmental hazards, and real-time access to community-wide antibiograms.


By “posterity” I refer to the use of patient-specific information for clinical research. I use the term “posterity” because good research adds permanently to the scientific knowledge base, and, well, I needed a word that starts with the letter P. Since PhD scientists do a lot of research, you can alternatively think of this one as PhDs.

Product Makers

Like it or not, medicine cannot be practiced without products that health care providers use, such as pharmaceuticals and medical devices. Manufacturers of these products, of course, rely heavily on patient-specific data for FDA approval of their products but also—and possibly more relevant with respect to EHR systems—are under increasing pressure to monitor the safety of their products in the post-marketing phase.


While patients and providers benefit mostly from providers’ direct use of patients’ individual clinical information, the other 4 P’s benefit mostly from analysis of many patients’ clinical information in aggregate (de-identified where appropriate). Because patients seek care from providers who will be using disparate EHR systems, this is yet another reason to implement national terminology and communication standards for clinical information sharing; the value contained within EHR systems will likely not be realized without it.

In closing, when you are thinking about who are the users and re-users of patient-specific clinical information, think the 6 P’s.

iIn addition to the benefits of sharing and re-using patient-specific clinical information, the 6 P’s also serve to benefit from yet-unimplemented sophisticated clinical decision support. Further discussion of this topic is beyond the scope of this blog post.