Residents & The Match: Overworked, Underpaid

The Atlantic has a nice brief history of the NRMP Match and an argument for it as a causal factor as to why being a resident is generally terrible. And, in case you didn’t know, the public also wishes you weren’t working so hard:

Medicine enjoys the status of being the most prestigious profession in America, yet the rigor of medical training remains unduly excessive. The American public overwhelmingly supports restrictions on residents’ working hours. A recent poll conducted by an independent public-opinion survey firm found that nearly 90 percent of Americans believe residents’ shifts should be 16 hours or less, and over 80 percent of those surveyed said that they would request a new doctor if they knew their physician was on the tail end of a 24-hour shift.

The Atlantic has been posting a lot of doctor stories recently with the current Republican-ACA collision. One thing Ryan Park’s argument is missing, though, is the fact that the hospitals only sort’ve determine the salaries of their residents. The more than $100k cost of a resident’s salary plus their “training” comes from CMS. Yes, the government, which also sets the number of spots they’re willing to fund. If a hospital were to suddenly improve salaries and benefits, they would lose the “free-ness” of the labor. If they hire more people than are funded (i.e. over the cap) to get the work done, they’re even more in the red. The government subsidizes the cost of training doctors, but as a practical matter, the government is largely subsidizing academic medicine, as well as teaching and county hospitals nationwide. The vast majority of these hospitals aren’t really footing the bill, and their budgets rely on having residents on hand for predictable periods of time churning through the night.

Park includes a reference about how resident training preferences may be a contributing factor to suppressing salaries:

In ranking programs, as Signer of the NRMP points out, most medical students are mainly concerned with prestige and the quality of training, not money. One 2015 study showed, for example, that even without the match, residents would still earn far less than their true market value—which is estimated to be about double what they presently earn—because they, in effect, accept a pay cut for high-quality medical training and a prestigious residency placement.

But of course! The salaries are all terrible. That reference does make that conclusion, but we know better because resident pay is so homogenous (again, paid by CMS with regional COLA). If a terrible program pays a few thousand more per year than a great program, of course no is going to care. Educational factors clearly trump trivial salary differences. If cost was the only factor in all people’s choices, no one would choose to attend private schools. But if a decent nonmalignant program paid twice as much (i.e. a PA salary) as a prestigious misery-factor? It wouldn’t sway everyone, but I have no doubt it would absolutely have a big impact, just like how a lot of very talented people only consider attending their state medical school.

The US has an abundance of patient care to carry out and a growing shortage of doctors, but we’ve both resisted real increases in resident numbers and prevented substantial changes in the training paradigm. In a world where the same Medicare coffers will pay for drugs that cost more than a resident salary while advance practice nurses have lobbied for greater and greater autonomy, the ACGME’s focus on “milestones” and the length of training has serious unintended consequences.

Imagine for a moment that internal medicine, family medicine, and pediatrics were two-year residencies. Without massive budget changes, suddenly we’d be training 33% more generalists per year AND the return on the time/money investment of becoming an internist would improve substantially (likely luring higher performing students). Would there be major negative consequences in the quality of those graduating residents? How long would they last? If so, could they be mitigated by changes in medical school or residency training? Have we even tried? Have we even considered it in the past 40 years?



Medical training return on investment

Some fun (but not new) light reading for those debating whether pursuing medicine was a mistake: UC Berkeley’s Nicholas Roth’s The Costs and Returns to Medical Education.

Overall, of the specialties included, rad onc and radiology topped the scale and endocrinologists bottomed it. The data is from 2009, so some of the assumptions are out of date (as well as pre-dating the imaging reimbursement crunch and subsequent fall in radiology reimbursement, for one). In particular, how student loans are handled makes the data presented significantly less terrible than the reality for some specialties. But it’s still worth reading for several reasons, including:

After Congress passed the 1997 Balanced Budget Act, which capped government payments to hospitals for residents, hospitals added over 4,000 more residents than the government would support. This suggests that market forces are at work as hospitals try to hire residents until the marginal value of an additional resident is zero. It also suggests that hospitals profit from additional residents long after the point when our government stops funding resident education.

I didn’t know that, but it jives perfectly with the narrative all residents believe that hospitals benefit from our cheap labor despite the ludicrous claims that it “costs more” to educate a trainee.

Back to the numbers though, and ultimately, the provided calculated rates of return for the investment in medical school and training is fraught with misleading specificity. Career duration can change the entire calculus (he uses a retirement age of 65). From chron:

Although the overall physician population has grown 188 percent between 1970 and 2008, according to the AMA, the physician population over age 65 has grown by 408 percent in the same period. Economic factors may be keeping many physicians on the job longer, according to data from The Doctors Company, a medical malpractice insurance firm. The company found that the portion of physicians reporting satisfaction with retirement plans has dropped 18 percent since 2006, and the average age at which an internist retired had increased from 62 in 2002 to 70 in 2009.

When considering the costs of becoming a doctor, one must add up the real costs of attending school, lost wages during school, decreased wages during residency, and interest on student loans. Roth uses $36,369.68 as the annual tuition and fees (now it’s probably closer to $47,500 according to the AAMC). He does not count for living expenses, which is fair since everyone’s got to live, but students usually borrow this cost and his treatment of loans is suboptimal. He uses the wages of an intern to approximate lost wages during school (which is probably low for what most doctors could earn in other fields). He uses the average wage of a same-aged person with a college degree to calculate the opportunity cost of not working during school (again, probably low, given that physicians are typically better than average students and likely destined for better than average nonmedical careers overall as well). He also gives us free money:

In 2011, approximately 88.8% of professional degree students received some sort of aid. Of those students, the average aid awarded amounted to $27,500. xxix When weighted, the total aid for all students on average amounts to $24,420 annually.

I know almost no medical students who receive aid anything like that unless we’re grouping student loan “aid” in here. While there are the occasional folks with full rides, most scholarships are small, and most aid given for professional students is actually in non-medicine fields. This probably shouldn’t have made it into the analysis.

I also consider interest payments on student loans. Creditors offer many different types of loans to students, and this makes it very difficult to infer a general loan payment structure. For my purposes, I assume that a typical student accrues $100,000 in interest payments from loans for medical school. I assume that this student pays $5,000 in interest payments annually during the first three years out of medical school and $12,140 annually for an additional seven years.

These sums do not include payments on the original principle; they only include interest payments. These assumptions are similar to a sample repayment schedule presented by AAMC. This repayment schedule assumes an initial Federal Stafford loan of $160,000 dollars with public service loan forgiveness with an assumed $100,000 starting salary after residency

Essentially all loans are DIRECT loans now, but it’s still impossible to handle all loan possibilities with one plan. But this assumes basically a three-year residency followed by PSLF. Given that 44% of graduating students are considering PSLF, assuming that the average doctor isn’t really on the hook for their student loans is misleading. The total loan amount is also now too low, as are the likely annual payments as an attending.

So that’s the dealbreaker. Student loans have changed a lot since interest rates have risen. He assumes no interest payments followed by 100% PSLF adoption, which severely underestimates the cost of attendance to basically 100k even. That’s not realistic for the 60% of doctors who don’t plan to attempt PSLF and would even less useful if PSLF is eventually capped as has been proposed.

It would have been nice to see the payback method used as well (the measure of time to break-even point on an initial investment (it’s intuitive and easier to understand for most folks). But using Net Present Value (NPV) is a great way to present the value of an investment in medical training. Unfortunately, the assumptions are everything: the initial investment cost and the discount rate change the game:

It would seem that the interest payments weigh heavily on the net present values of medical education investments, although these values remain substantively positive.

Specifically, these correlations suggest that physicians receive increases in earnings that overcompensate for the opportunity costs of additional training. This assumes, of course, that additional years of residency and fellowship training result in higher earnings than lesser-trained physicians. The correlation between median earnings and the years of training necessary for the specialties I analyzed, however, is only .4588. While this correlation is significant, it reveals some inconsistencies between further training and earnings. If further training does not result in increased earnings to justify that training, then some physicians may find it profitable to avoid specialization.

Case in point: Infectious disease and endocrinology. You lose three years of attending income only to make less than if you hadn’t specialized in the first place. You are effectively taxed against your academic and clinical interests.

As Roth notes, the assumptions used and the relative costs etc change the number. But in the past several years, the only trend has been more cost to training at overall higher interest rates than were available in the 2000s. This change only exacerbates the cost of choosing a specialty with a bigger duration-to-income ratio.

It’s nice to see a mathematical illustration of what everyone implicitly knows. While Roth’s investment outlook is sunnier than reality, the comparison between different specialties is still relatively meaningful.

The average doctor with average debt is still doing okay. But a doctor choosing a less than average remunerative field with greater than average levels of debt is a different story. That private college + private medical school graduate passionate about rheumatology better be planning on starting their career in academia and hoping PSLF stays just the way it is. The orthopod? They’re just fine.

And if medical schools continue to get more expensive and options for forgiveness are capped, we’re not that far off from the point when some fields will no longer make financial sense at all.

The danger (?) of intravenous contrast media

Another study piling on the mounting evidence that at least modern contrast agents put into people’s veins (and not arteries) for CT scans might not be bad for your kidneys after all.

The biggest single center study of EM patients was just published in The Annals of Emergency Medicine, which studied 17,934 patient encounters and compared renal function across 7201 contrast-enhanced scans, 5499 non-con scans, and 5,234 folks with no-CT.

6.8%, 8.9%, and 8.1% were the rates of AKI respectively. As in, folks who received either no contrast or no CT imaging were more likely to have a significant rise in creatinine than people who got contrast. As in, contrast was protective (statistically). Using different cutoff guidelines for AKI, the three were all statistically equivalent.

Practice patterns here still get in the way. Patients with low GFRs are more likely to get fluids prior to receiving contrast, possibly explaining the pseudo-protective effect of contrast. Patients were poor renal function are less likely to get contrast in the first place, reducing the power for evaluating contrast’s effects on those with CKD. However, controlling for baseline GFR didn’t change the story: there wasn’t an increased risk associated with receiving intravenous contrast in this controlled retrospective study regardless of underlying renal disease.

Historically, randomized controlled trials designed to elucidate the true incidence of contrast-induced nephropathy have been perceived as unethical because of the presumption that contrast media administration is a direct cause of acute kidney injury. To date, all controlled studies of contrast-induced nephropathy have been observational, and conclusions from these studies are severely limited by selection bias associated with the clinical decision to administer contrast media.

Maybe with all this mounting evidence it’s time to do an RCT.

CFPB sues Navient (even the feds aren’t happy with their own student loan servicers)

The Consumer Finance Protection Bureau is a nonpartisan government agency that was created after the 2008 financial crisis to help protect us from evil banks, credit card companies, predatory lending practices, etc.

They announced this week that they’re suing Navient, the biggest federal (and private) student loan servicer for defrauding borrowers and being generally terrible.

You may or may not really know or care about loan servicers, but they’re the middlemen who send you emails about making payments and take your money. Even when you take federal loan money, most people end up shuttled to one of the several private servicing companies who manage the payments. It should come as no surprise that handling that money is profitable and that unsavory companies stand to profit more if people send more money their way.

In short:

The Bureau alleges that Navient has failed to provide the most basic functions of adequate student loan servicing at every stage of repayment for both private and federal loans. Navient provided bad information in writing and over the phone, processed payments incorrectly, and failed to act when borrowers complained about problems. Critically, it systematically made it harder for borrowers to obtain the important right to pay according to what they can afford. These illegal practices made paying back student loans more difficult and costly for certain borrowers.

When people talk about trying to do clever (or even sometimes simple) things with payment timing or distributing loan payments, this is exactly why I would be hesitant to encourage them. It makes great sense to try to have extra payments go toward loans with the highest interest rate (and it’s still worth it). It makes sense to try to “time” extra payments to occur after the application of a REPAYE interest subsidy to get the best of both worlds by getting free government money while also paying down your debt as aggressively as possible (and probably not worth trying, I’d argue). It makes sense to talk with your servicer to get information about your options and make sure everyone is on the same page.

The problem is that the servicers aren’t good at servicing your debt.

I’m concerned enough about the servicers doing the basics of applying IDR-based interest subsidies for borrowers in general, and they purposely make it difficult to confirm what’s happening on their end by poorly if ever demonstrating this information on loan statements, even while they promise “they’re there” when you call. I’m deeply suspicious of any plan that involves the servicers working in borrowers’ best interests.

I’ve spoken with multiple residents steered away from the REPAYE plan through misinformation into IBR or PAYE. I don’t know how much is malfeasance or just incompetence—the people on the phone are customer service reps, not experts—but it’s nonetheless one of the downsides of having a complex convoluted federal system that even the people charged with handling it are unclear as to how to implement its policies.


Atul Gawande with another fun New Yorker feature on The Heroism of Incremental Care:

Rescue work delivers much more certainty. There is a beginning and an end to the effort. And you know what all the money and effort is (and is not) accomplishing. We don’t like to address problems until they are well upon us and unavoidable, and we don’t trust solutions that promise benefits only down the road.

Incrementalists nonetheless want us to take a longer view. They want us to believe that they can recognize problems before they happen, and that, with steady, iterative effort over years, they can reduce, delay, or eliminate them. Yet incrementalists also want us to accept that they will never be able to fully anticipate or prevent all problems. This makes for a hard sell. The incrementalists’ contribution is more cryptic than the rescuers’, and yet also more ambitious. They are claiming, in essence, to be able to predict and shape the future. They want us to put our money on it.

But our free-market insurance only wants to pay for 15 minutes of it, of course.


As an American surgeon, I have a battalion of people and millions of dollars of equipment on hand when I arrive in my operating room. Incrementalists are lucky if they can hire a nurse.


The difference between what’s made available to me as a surgeon and what’s made available to our internists or pediatricians or H.I.V. specialists is not just shortsighted—it’s immoral.

When people think about rationing care, they talk about rationing care to people. About grandma not getting a pacemaker or a new hip. They speak disparagingly about Canada or the UK. What people don’t realize is that we also ration care internally within medicine. We just do it based on RVUs.

Then, at the end, he finishes with some jabs at half-baked plans to repeal the ACA and a powerful somber note:

In this era of advancing information, it will become evident that, for everyone, life is a preexisting condition waiting to happen.