Physicians earn high incomes but those incomes come at a cost of investing between 7 and 12 years of education and training after undergraduate college. This post will examine the most recent physician compensation report and what it indicates about the relationship between income and the years of training required for each specialty.
Determining average physician incomes by specialty turns out to be a lot more difficult that it would seem. There are many physician compensation surveys and each of them reports compensation a bit differently with the result that it is difficult to accurately know how much the average specialist actually earns per year. Some of the most common surveys include:
- AAMC – American Association of Medical Colleges. This annual survey reports physician compensation from 153 U.S. medical schools and > 400 teaching hospitals that serve 124,000 physicians.
- MGMA – Medical Group Management Association. This annual reports surveys 3,400 U.S. medical practice administrators that serve 142,000 physicians and advanced practice providers. These group practices are largely mid-sized groups (typically 6 – 50 physicians).
-
AGMA – American Group Medical Association. This survey represents 380 medical groups from large-sized groups (with > 100 physicians).
- Doximity. This survey is of self-reported total compensation from 31,000 full-time U.S. physicians.
- Medscape. This survey is of self-reported total compensation from 13,000 U.S. physicians.
- Various physician search firms and consultation firms. These are typically of small numbers of physicians and often limited to compensation reports of individual physicians that they have helped with job placement and physician groups that they have consulted with.
I tend to rely mostly on the AAMC and MGMA reports because they sample the largest number of physicians and have stricter methodology regarding what is (and is not) included in total compensation. For academic physicians, the AAMC survey is more comprehensive and generally reports higher incomes for academic physicians than the MGMA survey. For non-academic physicians, the MGMA report provides comprehensive data. For this post, I will use the 2022 MGMA physician compensation report. Total compensation is defined as salary and bonuses as well as physician contributions to retirement plans, health insurance, and life insurance. Notably, the reported compensation does not include employer contributions to retirement plans, health insurance, life insurance, or malpractice insurance.
This is particularly important when comparing academic from non-academic physician compensation since most academic jobs come with lucrative employer contributions. As an example, the Ohio State University contributes about $25,000 per year to their physician faculty member’s State Teacher’s Retirement Plan, life insurance, disability insurance, and health insurance. OSU also pays for medical malpractice insurance – the U.S. national average cost for a critical care physician’s malpractice premium is $20,215 per year. In other words, a typical OSU physician has a total of about $45,000 per year in fringe benefits as an academic physician that they would otherwise likely not have had if they were in a private medical practice. One of the reasons that the MGMA reports that academic physician compensation is much lower than private practice physician compensation is because these employer contributions provided by academic institutions are not included in the total compensation listed in the MGMA reports. If you were to factor in these employer contributions into total compensation, academic physicians’ compensation is closer to that of non-academic physicians.
The MGMA breaks reported compensation into mean, median, 25th percentile, and 75th percentile. For academic physicians, the MGMA additionally breaks down compensation by academic rank: instructor, assistant professor, associate professor, and professor. Other metrics of compensation and productivity are also included such as average total RVUs, average work RVUs, and total compensation per RVU for each specialty. Caution must be exercised when interpreting these data. For example, the mean compensation will include all non-academic physicians in a specialty, regardless of seniority. Physicians in their first years of practice after completion of training are less efficient, less productive, and less highly compensated than physicians in practice for 10, 20, or 30 years. Therefore, a newly-trained physician should not expect to earn the mean or median compensation for a specialty. Conversely, experienced physicians with many years of practice generally earn more than the mean or median. However, for simplicity purposes, this post will focus on the mean total compensation for various specialities for non-academic and academic physicians. The total compensations are summarized in the tables below:

Non-Academic Physician Compensation

This graph illustrates the mean total compensation for non-academic physicians reported by the MGMA in 2022, similar to the table above (to enlarge this graph, click on it to open it in a new window and then click on it again to enlarge). The most highly-compensated specialties were neurosurgery ($947,030), cardiovascular surgery ($829,072), cardiology electrophysiology ($747,947), orthopedic surgery ($715,399), and interventional cardiology ($702,019). At the low end of the compensation spectrum were pediatric specialties: pediatric hospitalist ($237,530), pediatric endocrinology ($239,072), general pediatrics ($252,575), and pediatric infectious disease ($256,364). In fact, of the 9 lowest compensated specialties, all but one (geriatrics) was a pediatric specialty.
Academic Physician Compensation

This graph illustrates the mean total compensation for academic physicians reported by the MGMA in 2022. The most highly-compensated specialties were cardiovascular surgery ($718,802), neurosurgery ($694,605), pediatric surgery ($588,934), thoracic surgery ($581,387), and plastic surgery ($525,215). At the other end of the compensation spectrum were again pediatric specialties: pediatric endocrinology ($184,479), general pediatrics ($189,178), pediatric infectious disease ($201,607), and pediatric hospitalist ($204,661).
In every specialty, academic physician total compensation was lower than non-academic physicians (academic pediatric-internal medicine compensation was not reported). The specialties with the greatest difference between non-academic and academic compensation were cardiology electrophysiology ($293,318), neurosurgery ($252,425), gastroenterology ($244,091), hematology/oncology ($237,720), and orthopedic surgery ($231,973). The large difference between academic and non-academic incomes explains why it has been so difficult for medical schools to keep gastroenterologists and oncologists since they can earn a quarter of a million dollars more per year in private practice. The lure of that much money is just too much for even the most noble of academic teachers and researchers. Specialties with the least difference between non-academic and academic compensation were pediatric hospitalist ($32,869), pediatric nephrology ($44,281), pediatric critical care ($47,283), and pediatric hematology/oncology ($53,152).
Compensation per work RVU

Physician work effort is often measured by the number of RVUs (relative value units) produced. Every physician service and procedure is assigned an RVU value by Medicare and then Medicare pays the physician based on the number of RVUs billed. Currently, Medicare pays $33.89 per RVU. Commercial insurance companies generally pay a higher amount per RVU and Medicaid pays a lower amount per RVU. The RVU is composed of three subunits, the work RVU (wRVU), practice expense RVU, and malpractice RVU. Of these subunits, the wRVU is most commonly used to measure physician productivity. Note that anesthesiology does not use RVUs and anesthesiologist productivity is instead measured by anesthesia units (1 unit = 15 minutes of time).
Physicians who earn a high dollar amount of compensation per wRVU generally require subsidization from hospitals.This is typically done either when the physician performs procedures that are highly lucrative for the hospitals (such as open heart surgery) or when the physician performs a lot of non-compensated work essential to the function of the hospital (such as hospitalists who take night-call). On the other hand, physicians earning a low dollar amount of compensation per wRVU have less (or no) hospital subsidization. These are usually outpatient specialties whose physicians are less often employed by a hospital.
Non-academic physicians with the highest compensation per wRVU are pediatric surgeons ($148/wRVU), pediatric hospitalists ($138/wRVU), pediatric infectious disease ($123/wRVU), neurosurgeons ($113/wRVU), and pediatric hematology/oncology ($112/wRVU). Those specialties with the lowest compensation per wRVU are pediatric/internal medicine ($54/wRVU), endocrinology ($59/wRVU), ophthalmology ($59/wRVU), family medicine ($62/wRVU), and general pediatrics ($62/wRVU).

For academic physicians, the specialties with the highest compensation per wRVU are pediatric hospitalist ($179/wRVU), pediatric surgery ($133/wRVU), internal medicine hospitalist ($123/wRVU), hematology/oncology ($117/wRVU), and infectious disease ($114/wRVU). The high compensation per wRVU for academic infectious disease physicians may reflect the impact of the COVID-19 pandemic when academic infectious disease specialists were called on to perform a great deal of administrative duties (subsidized by hospitals) in addition to their regular clinical duties. Academic physician specialties with the lowest compensation per wRVU are dermatology ($48/wRVU), neonatology ($50/wRVU), pathology ($51/wRVU), radiology ($55/wRVU), and interventional radiology ($55/wRVU). The MGMA survey did not report data for academic pediatric/internal medicine or for pediatric infectious disease.
Compensation per year of residency & fellowship training

Residency and fellowship can be viewed as an investment in a physician’s career. In theory, the longer the period of training, the greater the knowledge and skill of a physician in any given specialty. Residents and fellows do get paid but the average annual income is modest, starting at $61,000 for a first year resident (i.e., an intern) and that amount increases by about $2,500 for each additional year of residency and fellowship. During this time, residents and fellows are also required to start paying back student loans (payments averaging $4,000 per year during residency). As a consequence of residency and fellowship training years, most physicians finally enter the workforce when they are in their 30’s. The total duration of residency varies from the shortest at 3 years (internal medicine, pediatrics, and family medicine) to the longest at 7 years (neurosurgery, pediatric surgery, and interventional radiology). Fellowship training after residency further extends the total duration of training, for example, cardiology electrophysiology requires 8 years of training (3 years internal medicine residency, 3 years cardiology fellowship, and then 2 years cardiac electrophysiology fellowship). Longer residency/fellowship durations also equate to a shorter working career. The general internist with a 3-year residency will typically work 35 years before retiring at age 65 whereas the cardiology electrophysiologist will only work 30 years before retirement at age 65. Thus, the cardiology electrophysiologist sacrifices 5 of their lifetime income-earning years to do fellowship training after their internal medicine residency.
Do more years of residency/fellowship translate to higher incomes? One way to answer that question is to express physician compensation per number of years of training required for that specialty. In a completely free labor market, there would be a direct relationship between income and duration of training: every additional year of training for any given specialty would result in a predictable increase in annual income. In other words, the return on investment in terms of years of training should be constant across all specialties. This turns out to not be the case in reality.
For non-academic physicians, there is a wide variation in compensation per year of training. The specialties with the largest amount of total compensation per year of residency/fellowship are orthopedic surgery ($143,080 per training year), dermatology ($140,439 per training year), cardiovascular surgery ($138,179 per training year), neurosurgery ($135,290 per training year), and emergency medicine ($124,239 per training year). These specialties have a very high return on their investment of training time. At the low end are pediatric endocrinology ($39,845 per training year), pediatric infectious disease ($42,727 per training year), pediatric hematology/oncology ($43,808 per training year), pediatric nephrology ($44,756 per training year), and pediatric hospitalist ($47,506 per training year). These specialties have a low return on investment of training time.

The spread of total compensation per number of years of residency/fellowship training for academic physicians was similar. Specialities with a high compensation per year of training were cardiovascular surgery ($119,800 per training year), emergency medicine ($102,326 per training year), anesthesiology ($101,900 per training year), neurosurgery ($99,229 per training year), and thoracic surgery ($96,898 per training year). Once again, the least compensated per year of training for academic physicians were all pediatric specialties: pediatric endocrinology ($30,747 per training year), pediatric infectious disease ($33,601 per training year), pediatric hematology/oncology ($34,950 per training year), pediatric pulmonary ($35,946 per training year), and pediatric nephrology ($37,376 per training year). The MGMA survey did not report on pediatrics/internal medicine.

Several subspecialties were particularly noteworthy because their total compensation was less than their parent specialties. For example, pediatric hospitalists require 2 additional years of fellowship after completion of a pediatric residency and pediatric endocrinologists require 3 years of fellowship after pediatric residency. However, both non-academic pediatric hospitalists and non-academic pediatric endocrinologists make less money than non-academic general pediatricians who only completed the 3-year pediatric residency. Similarly, to specialize in geriatrics or endocrinology, a physician must first complete a 3-year internal medicine residency followed by a 1-year (geriatrics) or 2-year (endocrinology) fellowship. However, non-academic physicians specializing in geriatrics or endocrinology make less money than non-academic general internists who only completed the 3-year internal medicine residency.
In academic practices, there are even more specialities where subspecialty fellowship results in lower total compensation than the parent specialty. Academic pediatric endocrinologists make less than academic general pediatricians. Academic geriatric, rheumatology, endocrinology, and infectious disease specialists all make less than academic general internists. In these subspecialties, not only does the additional years of fellowship training not result in greater income, but the those physicians are actually financially penalized for their additional years of training by making less money than if they had just stopped after their pediatric or internal medicine residency.
It is noteworthy that there are more factors to consider than just years of training when comparing total compensation between different specialties. Some of the specialties with the highest compensation per year of training are also those with the most grueling on-call schedules, such as cardiovascular surgery, anesthesiology, emergency medicine, and neurosurgery. It is entirely appropriate that the neurosurgeon who has to take trauma call every 4th night for his/her entire life makes a high income. In addition, the cost of medical malpractice insurance premiums varies significantly. The average general internist pays $16,000 per year in malpractice premiums but the average neurosurgeon pays $92,000 per year for malpractice coverage. Once again, it is entirely appropriate that the neurosurgeon has a high income in order to cover the high overhead malpractice insurance expense inherent in that specialty.
What is the solution to these compensation disparities?
In a free labor market, a worker’s income is determined by the supply of workers and the demand for that worker’s services. So, on the surface, it would appear that there is a shortage of heart surgeons and neurosurgeons whereas there is a overabundance of general pediatricians and pediatric endocrinologists. However, American medicine is not a simple free market economy. Hospitals make the most money from procedures and surgeries: the financial margin on a surgery is much greater than the margin on a medical admission. That margin is highest for inpatient surgeries such as cardiovascular surgeries and neurosurgeries. Because of this, hospitals are incentivized to subsidize specialists who perform these high-margin procedures. Furthermore, many of these surgical subspecialists have much more rigorous on-call schedules – a neurosurgeon or interventional cardiologist is much more likely to be called into the hospital in the middle of the night to manage a patient with head trauma or with a myocardial infarction than an endocrinologist or rheumatologist whose practice is largely outpatient and limited to Mondays through Fridays during the daytime. Therefore, in order to provide 24-hour trauma or cardiac care, hospitals must pay these subspecialists substantial on-call pay.
A central problem with physician reimbursement is that it has not kept up with inflation and has, in fact, fallen over the past decades. In 1998, Medicare reimbursement per RVU was $36.69 and 25 years later, in 2023, the reimbursement per RVU had fallen to $33.89. By contrast, if the RVU reimbursement had merely kept up with inflation, then the $36.69 rate in 1998 should be $70.45 today! Physicians have made up for the reduced payments per RVU somewhat by spending less time with each patient in order to see more patients per day but that alone has been insufficient to maintain a constant income. The solution has frequently been for physicians to become employed by hospitals with the hospitals subsidizing their income. This has resulted in physician income becoming untethered from physician work productivity. The effect has been that physician income is increasingly determined by the value of the physician’s specialty to the hospital’s finances more than the physician’s actual patient care work effort.
It has been proposed that the solution would be to pay low-compensation subspecialists more. This would work in a pure free market economy but would not work in our current system of physician reimbursement. Physician services are categorized by CPT codes and then reimbursed by the number of RVUs associated with each of those CPT codes. Non-procedural specialties all use the same CPT codes for the evaluation and management services that they provide. Thus, the endocrinologist or geriatrician bills the exact same CPT codes as the general internist and gets reimbursed the exact same amount per RVU as the general internist. Because of this, the “cognitive” subspecialties of pediatrics and internal medicine (i.e., those without associated procedures) have no chance of generating more RVUs than the general pediatrician or internist. Indeed, the amount of time and effort to see a 10-year old with uncontrolled type 1 diabetes in the pediatric endocrinology office is considerably more than that required to see an otherwise healthy 10-year old with an ear infection in the general pediatrics office, even though the payment is the same for both patients. As a result, for many of these subspecialties, the reward for more years of training is a lower income. Because these pediatric and internal medicine subspecialties do not generate significant margins for hospitals, there is little incentive for hospitals to subsidize them.
It is notable that pediatric subspecialties dominate the low compensation specialties. One of the driving reasons for this is Medicare/Medicaid. Nearly every American over age 65 qualifies for Medicare so older adults are by and large all insured. Children are not eligible for Medicare but are instead covered by CHIP and Medicaid programs (or have no insurance at all!). In most states, Medicaid pays considerably less than Medicare (in Ohio, Medicaid payments for primary care services are only 57% of the Medicare amounts). Consequently, pediatricians of all subspecialties have an inherently worse payer mix than physicians who care for adults. Similarly, pediatric hospitals also have a worse payer mix than hospitals caring for adults.
So, how do we fix this? There are several tactics that can be considered:
- Increase residency positions in some specialties. This will work only for those highly compensated specialties where there is truly an insufficient supply of physicians for current demands.
- Re-align RVUs assigned to different procedures and services. The current RVU assignments have been affected by intense lobbying from subspecialty physician organizations and in many cases, the most RVUs have been given to the loudest lobbyists.
- Increase physician reimbursement for Medicaid and CHIP patients. In an ideal world, a physician would get paid the same for a patient with Medicare, Medicaid, or CHIP. This would help correct the low compensation for pediatric specialties.
- Increase the RVU conversion factor. The current conversion factor of $33.89 per RVU is too low for the vast majority of physicians to earn a living from professional billings alone with the result that most physicians require hospital subsidization. This has eroded free market effects on physician compensation.
- Normalize the relation between years of training and income. It is entirely appropriate that the interventional cardiologist who trains for 7 years has a higher income than the general internist who trains for 3 years. But it makes absolutely no sense that the endocrinologist who trains for 5 years makes less than the internist who trains for 3 years.
- Strategic expansion of advance practice provider utilization. We have to face the reality that income disparities in some specialties will eventually result in fewer physicians entering those specialties. Hospitals should start training nurse practitioners, physician assistants, and pharmacists to perform some of the work done by these specialists. For example, advance practice providers can often effectively replace most of the daily inpatient diabetes management currently done by endocrinologists.
- Embrace AI. The heart surgeon will not do a coronary artery bypass surgery faster using artificial intelligence but AI may allow the general internist to more efficiently evaluate a patient with chest pain. Similarly, AI may speed up the time required for an infectious disease specialist to come up with a diagnosis based on a patient’s presenting history and lab findings. It can help the endocrinologist select the most effective diabetes treatment based on a patient’s co-morbidities. It can shorten note and order-writing time for patients performing E&M (evaluation and management) services. Artificial intelligence has the greatest potential to improve productivity of physicians in cognitive specialities, which are also the specialties that are the most under-compensated.
The forces that affect physician incomes are complex. But if we do not begin to take corrective action soon, we will find ourselves without endocrinologists, geriatricians, and pediatric endocrinologists in the near future. Because of the structure of American healthcare, we cannot rely on free market forces alone to solve this problem.
April 9, 2023