Medical Economics Medical Education

What The 2019 Fellowship Match Tells Us About The Future Of American Medicine

Last week, the National Resident Match Program released its annual data report on the most recent fellowship match for positions to begin in July 2019. The fellowship matches take place between May 2018 and January 2019, with different match days for different specialties. The NRMP releases a summary of all of the specialty match results in February each year. By analyzing the data, you can learn a lot about which specialties are attracting new physicians and which ones are not attracting new physicians. And projecting forward, you can predict which specialties are going to have shortages of practitioners in future years. The 2019 year had the most positions ever offered in the U.S. (10,936), the most fellowship programs ever in existence in the U.S. (4,750), and the most residents who matched into a fellowship program in the U.S. (9,378).

Because the hospital I practice in primarily has internal medicine specialists, I am particularly interested in the outcome of the various internal medicine specialty fellowship matches. This table in blue shows the number of internal medicine fellowship positions by specialty. They vary widely from a low of adult congenital heart disease (9) to a high of cardiology (951). Oncology-only fellowships and hematology-only fellowships generally have very small numbers of available positions because most physicians tend to go into combined hematology and oncology fellowships. Similarly, there are few pulmonary-only fellowships because most candidates go into a combined pulmonary and critical care fellowship.

The NRMP also reports the percentage of available positions that fill with U.S. allopathic (MD degree) medical school graduates, as shown in this table in red. Eliminating oncology, hematology, and pulmonary from the analysis because most doctors interested in those fields do a combined fellowship, it becomes clear that the most popular specialties are hematology & oncology, gastroenterology, and palliative medicine (adult congenital heart disease, interventional pulmonary, and cardiac electrophysiology are only done after a doctor has already completed a pulmonary or cardiology fellowship and are thus sub-specialty fellowships). The least popular fellowships are nephrology, geriatrics, and endocrinology.

Next, lets look at the how all of the various specialties filled when including not only U.S. allopathic graduates but also U.S. osteopathic (DO) graduates, U.S. citizens who attended foreign medical schools (most frequently Caribbean medical schools), Canadian medical school graduates, and foreign medical school graduates. Here, in the table in brown, we see that most fellowships eventually filled. However, nephrology and geriatrics are notable exceptions.

So, why are certain specialties so unpopular? One factor may be salary. Using the 2018 Medscape Physician Compensation Survey data, the internal medicine specialties with the highest salaries also had the highest percentage of available fellowship positions filled with U.S. allopathic medical school graduates (cardiology, gastroenterology, and hematology/oncology). Endocrinology is notable because the annual salary of an endocrinologist is less than a general internist, meaning that for the personal cost of doing 2 additional years of fellowship training in endocrinology compared with a general internist, you get to make $18,000 less than a general internist – not surprisingly, relatively few of the endocrinology fellowship positions were filled by U.S. allopathic medical school graduates.

Another way of analyzing salary and work effort is from the annual MGMA Physician Compensation Report. The 2018 report is the most recent and it reflects 2017 data. There are important differences between the MGMA compensation data and the Medscape compensation data. The MGMA report only includes large group practices, is reported by practice managers, and reflects total compensation. The MGMA report has data on private practice physicians (who make more) separate from academic physicians (who make less). On the other hand, the Medscape report includes any physician, is self-reported by the physician, and represents salary (which may not be the same as total compensation). The Medscape report includes both academic and private practice physicians. The MGMA data is probably more accurate but represents a skewed population of doctors. The Medscape data is probably less accurate but in theory is more representative of the average doctor (or at least those who are willing to take the time to fill out the survey). This table reflects just the MGMA private practice data. From the MGMA report, it is apparent that based on median wRVUs, nephrologists work very hard, in fact, harder than any other specialists except cardiologists. However, they make $140,000 less than cardiologists. When the MGMA data is expressed as total compensation per wRVU, it appears that nephrologists make the second lowest amount of money per wRVU compared to other specialists (note that the MGMA total compensation per wRVU is calculated with a complicated equation and that not all practices report both total compensation plus wRVUs so the reported values will be different than if you simply divide the mean total compensation by the median wRVUs for any given specialty).

There is a pretty sharp demarcation between the lowest paid specialties (geriatrics, endocrinology, rheumatology, infectious disease, and nephrology) and the highest paid specialties. The lowest paid specialties had the lowest percentage of fellowship positions filled with U.S. allopathic medical graduates. The one exception to this is palliative medicine but palliative is unique in that palliative physicians make about twice as much income per wRVU at $110.57 than the other specialties.

Another reason why newly trained residents may choose one specialty over another is how happy the physicians are in a given specialty. The 2018 Medscape Physician Compensation Report also has information on job satisfaction of physicians, including a survey question of “Would I choose the same specialty again?”. Once again, there is a relationship between job satisfaction and the percentage of residents who choose that specialty. The least happy internal medicine specialists are nephrologists (in fact, per the Medscape survey, they are the least happy specialty of all physicians, not just internal medicine specialties). Nephrology is also the specialty that filled the least of available positions with U.S. allopathic medical school graduates. So, if you choose to do 2 additional years of training compared to a general internist, you are rewarded by being 30% less happy with your job than a general internist. On the other hand, the three specialties with the most satisfied physicians are also the 3 specialties with the highest percentage of fellowship positions filled with U.S. allopathic medical school graduates (hematology/oncology, cardiology, and gastroenterology).

As we project the results of this year’s fellowship match into the future, we should anticipate future physician shortages in nephrology, geriatrics, infectious disease, and palliative medicine. Additionally, we should anticipate that there will be more nephrologists who are foreign medical school graduates than U.S. allopathic medical school graduates in the future based on the total number of U.S. versus foreign medical school graduates filling fellowship positions this year. Endocrinology and geriatric medicine are not far behind and in the future, your endocrinologist and geriatrician will probably be more likely to be an immigrant from another country than a graduate of a U.S. medical school.

If American medicine was an entirely free market economy, then as the supply of nephrologists, geriatric physicians, infectious disease physicians, and palliative medicine physicians goes down, then their salaries should go up. But physician supply and demand is complex and since it can take many years to train enough physicians to fill a specialty with a physician shortage, we may be looking at a medical economy with too few nephrologists, geriatric physicians, infectious disease physicians, and palliative medicine physicians for years to come.

An alternative explanation is that we have too many fellowship positions open in these specialties and that academic medicine is out of alignment with the needs of American medicine. Academic medical centers tend to create fellowship positions based on the needs of the individual divisions and departments within a given medical center and not necessarily based on the needs of the medical economy as a whole. This can also contribute to a misalignment of specialist supply and demand.

One thing seems certain, however. American nephrologists are less happy with their job, work harder in terms of wRVUs, and are compensated less per wRVU than other physicians. I believe that these are the reasons why so few physicians are entering the nephrology fellowship match and why so few nephrology fellowship positions filled in this year’s fellowship match. To insure that our patients’ medical needs are met in the future, hospitals will need to be sure that their nephrologists are happy and are compensated appropriately for the work that they do. On the other hand, gastroenterologists, hematology/oncologists, and cardiologists are the happiest and most highly compensated and this may explain why these three specialties filled so well in this year’s match.

February 26, 2019

Inpatient Practice Medical Economics

The 2019 Medicare Readmission Penalty

In September, CMS released the financial penalties that hospitals will pay for excessively high percentages of readmissions within 30 days of discharge. This is an annual event when hospitals get to find out how much their reimbursement from Medicare will be cut the next year. CMS focuses on 6 diagnoses when calculating the readmission penalty:

  1. COPD
  2. Coronary artery bypass surgery
  3. Myocardial infarction
  4. Heart failure
  5. Knee and hip replacement surgery
  6. Pneumonia

CMS looks at readmission data from July 2014 through June 2017. The penalties go into effect October 2018 and continue through September 2019. This year, 3,173 hospitals were evaluated and 2,599 (82%) were penalized. Certain classes of hospitals were exempt from evaluation including children’s hospitals, Veterans hospitals, hospitals in the State of Maryland, psychiatric hospitals, and critical access hospitals.

In the past, hospitals that take care of low income patients were penalized more than hospitals that take care of high income patients. For that reason, safety net hospitals and academic hospitals tended to get penalized more highly than other hospitals, in other words, hospitals got penalized for taking care of the poor. CMS overcame some of the limitations of previous years’ penalties by comparing hospitals to other hospitals that have similar patient demographics, rather than comparing all hospitals in the U.S. together. They calculated the number of dual eligible patients (those who have both Medicare and Medicaid) divided by the total number of Medicare patients. Because dual eligible patients are generally lower income than patients with Medicare only, this permitted CMS to compare hospitals that care for similar percentages of low income patients This is an improvement over previous calculations since lower income patients have higher 30-day readmission rates regardless of how good or bad their care was during their initial hospitalization. CMS stratified hospitals into 5 groups based on this calculation. Group 1 had 0-15% dual eligible patients whereas group 5 had 30-100% dual eligible.

The total amount of the penalties for next year is $566,000,000. Hospitals can be penalized a maximum of 3% of their entire Medicare revenues for that fiscal year but nationwide, the average penalty was 0.70%. There are 47 hospitals that incurred the maximum 3% penalty: Texas having the most at 8 hospitals, followed by Louisiana, Missouri, and Kentucky with 4 hospitals each. Here in Central Ohio, our hospitals all did quite well with only minimal penalties:

  1. Dublin Methodist – 0.03%
  2. Ohio State University – 0.06%
  3. Riverside Methodist – 0.17%
  4. Mt. Carmel West – 0.17%
  5. Grant – 0.23%
  6. St. Ann’s – 0.23%
  7. Doctor’s – 0.44%

States that expanded Medicaid have more hospital closures than states that did not expand Medicaid so one might hypothesize that hospitals in Medicaid expansion states would have more financial resources to put into reducing readmissions. So, I spent a few hours with an Excel spreadsheet of the 2019 Medicare penalties for all hospitals in the U.S. and it turns out that there was no difference in the average penalty incurred by hospitals in Medicaid expansion states versus hospitals in non-Medicaid expansion states.

So, overall, next year’s readmission penalties will be more fair than last year’s. But hospitals cannot control everything that a patient does or does not do once they leave the hospital and so the responsibility for fully reducing 30-day readmissions cannot lie solely on the hospitals.

November 24, 2018

Medical Economics

Pre-Existing Conditions And Individual Mandates

The Affordable Care Act is the piece of legislation that Americans simultaneously both love and hate. Americans love not being excluded from health insurance if they have pre-existing conditions but hate being forced to buy health insurance if they don’t want to. And so, there has been an effort to eliminate the individual mandate to buy insurance while preserving the the inability of insurance companies to deny coverage for pre-existing conditions. I think this is a great idea – every body wins… right?

Just think, now you can wait until you get sick before you have to buy health insurance! You can go for years without paying insurance premiums and then when your knee finally starts to give out, you can buy health insurance that year and get a knee replacement and then charge it to Blue Cross. While you’re at it, you can go ahead and get your screening colonoscopy, your mammogram, your cholesterol check, and your shingles vaccine all in that year while you are covered under your insurance policy. And then at the end of the year, you can opt out of health insurance and then not buy it again until 6 year later when you need the other knee replaced!

This is such a great idea, just think of how well it would work if we applied it to other forms of insurance:

Life Insurance. Americans will be delighted to not have to buy life insurance until they are already dead. Gone will be all of those years of having to pay life insurance premiums when you don’t die. Death is the ultimate pre-existing condition. This will save the average person tens of thousands of dollars in premiums over their lifetime.

Automobile Insurance. The average American pays $1,400 per year for car insurance. By not having to buy automobile insurance until you actually have a car wreck, we can now spend that $1,400 every year buying consumable goods and we will put our nation’s economy into hyperdrive. The average person is involved in 3 – 4 car accidents during their lifetime so that means that we would only have to buy automobile insurance for three or four years. And if you couple that with our new plan for life insurance, you can save twice as much by not having to buy either one until you are in a fatal car wreck!

Malpractice Insurance. Here in Ohio, internists pay an average of $14,00 per year for medical malpractice insurance. In New York City, it costs the same internist $36,000. And if you are an obstetrician in New York City, it will cost you $215,000 per year for malpractice insurance. By applying the same principle of pre-existing condition and individual mandates to doctors, we would no longer have to buy malpractice insurance until we are actually sued.

Umbrella Insurance. Every physician should have an umbrella insurance policy. Doctors have a big red bull’s eye painted on their backs that every personal injury attorney in the country can see. And so most physicians have a $1 million umbrella insurance in case one of those personal injury attorneys’ clients slips and falls on the physician’s driveway or is involved in a car accident with the physician. With this new insurance principle, doctors all across America can stop shoveling the snow from their sidewalks and if some pedestrian slips and breaks their hip, the the doctor can call his insurance agent and buy a quick $200,000 umbrella policy. If that pedestrian falls and breaks both hips, then you can buy a $400,000 umbrella policy.

Disability Insurance. Just about all physicians younger than age 55 need to buy disability insurance. But by eliminating the pre-existing condition & individual mandate requirements, we would no longer need to buy disability insurance until we are actually disabled.

Travel Insurance. No one really thinks that they are going to be in a plane crash. So I have a great idea – an app for your phone for travel insurance. Let’s face it, you don’t really need travel insurance if you actually get to your vacation destination. But with the new travel insurance app, if you hear your pilot say over the intercom, “Brace for impact”, in just seconds, you can open your app and buy travel insurance, before your plane actually crashes!

The way you keep health insurance premiums down is by selling it to as many healthy people as possible – it is the fundamental basis of actuarial science. If the only people who have to buy health insurance are those who are already sick or likely to become sick, then premiums will skyrocket. You can’t have the pre-existing condition clause without some element of an individual mandate.

November 13, 2018

Medical Economics Physician Finances

The Final 2019 Medicare Physician Fee Schedule: Some Winners, No Losers

Last week, CMS released the final rule for the 2019 Medicare Physician fee schedule. The initial proposed fee schedule was released last summer and would have radically changed the way that physicians are paid for outpatient clinical practice. There was a lot of criticism of the proposed fee schedule with most professional medical societies opposing it. To give CMS credit, they listened to the critics and modified the fee schedule accordingly. The end result is that not much will change in how physicians are paid next year.

At the core, the proposed fee schedule was going to establish a single CPT code for all new patient visits with a physician and a single code for all return patient visits with a physician. Thus, the current CPT codes 99202 – 99205 (new patient visits level 2 – 5) would be collapsed into a single CPT code. Similarly, the current CPT codes 99212 – 99215 (return patient visits level 2 – 5) would be collapsed into a single CPT code. The advantage of this is that it would have reduced documentation requirements, therefore reducing physician work. The disadvantage is that physicians would be paid the same amount for seeing and caring for a new patient with a cold as they would for seeing a patient with newly diagnosed breast cancer. Therefore, physicians who mainly take care of relatively simple medical problems would be winners whereas physicians who take care of a lot of complex medical problems would be losers. Since my outpatient practice is primarily limited to interstitial lung disease (a complex medial problem), I estimated that my total Medicare income would drop by 12%. In the outpatient world, about half of total income goes toward overhead expense and half goes toward paying the doctor – since overhead expenses would not change and would still have to be paid, the net effect of a 12% reduction in total Medicare revenue is that my personal income from seeing Medicare patients would drop by 24%.

After realizing this unintended consequence of the proposed 2019 Medicare Physician Fee Schedule, CMS decided to leave the current level 2 – 5 new and outpatient CPT codes in place and not consolidate them into single codes… at least for now. Instead, CMS plans to institute a revised version of this plan in 2021. The revised plan will consolidate level 2 – 4 outpatient visits into a single CPT code and leave the level 5 outpatient visit CPT code. Thus, instead of being 4 outpatient billing levels for physicians, there would only be 2 outpatient billing levels. The advantage is that there would less documentation requirements for all of the the lower level visits, thus freeing physicians from what is seen as a lot of unnecessary documentation in progress notes that requires a lot of physician time but adds nothing to the care of the patient.

The proposed 2019 Medicare Physician Fee Schedule would have also significantly reduced payment to podiatrists. However, the final schedule did not change podiatry reimbursement.

The proposed physician fee schedule was also going to cut by 50% the reimbursement for doing a procedure on the same day as an office visit. Therefore, a physician who saw a new patient and then did an EKG would only get paid 50% of the normal reimbursement for the EKG. This would have greatly impacted my practice since many (or most) of my patients get pulmonary function tests immediately before seeing me so that I can determine their response to treatment. In order to continue to be paid full reimbursement for these procedures, they would need to be done on a different day, thus requiring the patients to come in on 2 different days rather than getting their test and their physician visit on the same day. This would be a minor annoyance for patients who live in town but a significant burden on those patients who live 2-3 hours away. Fortunately, CMS decided to not institute this proposal in 2019.

So, in the end, not will change when it comes to physician reimbursement. However, there will be 2 important new reimbursable CPT codes that will allow physicians to now be paid for some of the services that they have been providing patients for free up to now. These are two new codes that pay physicians for telemedicine services. Physicians provide a lot of care over the phone and through patient portals of the electronic medical record systems. Sometimes, patients call or use the patient portal because it is more convenient than coming into the office. Sometimes it is because the physician’s regular office schedule is booked up and the patients can’t get in to see the physician. Sometimes, it is because a medical problem arises at night or on the weekend when the office is closed. And sometimes it is because the patient doesn’t want to pay a co-pay to be seen in person with an office visit. Here are the 2 new codes:

  1. G2012 – Brief communication technology-based service (virtual check-in). This will be used when a patient contacts the physician by phone or via an electronic medical record patient portal to decide if an office visit is needed. If the patient does end up coming into the office to be seen, you can’t bill the code but if the physician manages the patient’s condition by phone or via the patient portal without the patient coming into the office, you can bill the code. The patient cannot have seen the patient for a regular billable encounter for 7 days prior to the phone/portal encounter or for 24 hours after the phone/portal encounter. The medical discussion should be between 5 – 10 minutes. The patient will have to give verbal consent acknowledging that the telephone/portal visit will be billed. The patient must have been seen by the physician or a physician in the physician’s group within the past 3 years. This CPT code will be compensated at 0.25 work RVUs ($9.00).
  2. G2010 – Remote evaluation of recorded video and/or images submitted by an established patient. This will allow a patient to send the physician a photo or video for that physician to decide if an office visit is necessary. As an example, if a patient sends their physician a photo of a rash and the physician makes a diagnosis and directs treatment for the rash without the patient actually coming in to be seen. Similar to the “virtual check-in” code, patients cannot have been seen within the 7 previous days or within 24 hours after the video/image review. The patient must be an established patient of the physician. The patient must provide verbal or written consent acknowledging that the service will be billed. This CPT code will be compensated at 0.18 work RVUs ($6.50)

Lastly, CMS is going to give physicians a raise from $35.99 per RVU to $36.04 per RVU. That is a 1/10th of 1 percent raise in case you wondered.

November 9, 2018

Medical Economics

What Medicaid Expansion Tells Us About America’s Health

Medicaid expansion has been a rallying cry for both the politically far left and the politically far right. This post will examine some of the healthcare differences between those states that participated in Medicaid expansion and those states that did not expand Medicare.

At its most simplistic, Medicaid expansion has increased federal and state government costs but decreased the cost of providing uncompensated care to the uninsured by hospitals and doctors. Medicaid expansion was a product of the 2010 Affordable Care Act but it did not roll out until 2014. In the five years since initial roll out, 33 states either have already implemented or are in the process of implementing Medicaid expansion and 17 states chose to not participate in Medicaid expansion.  The differences in health metrics between these states are summarized in the following table:


  1. Life expectancy. Based on data from the CDC, people living in states that adopted Medicaid expansion live an average of 1 year longer than those living in states that did not adopt Medicaid expansion (79 years versus 78 years). Because life expectancy reflects the cumulative effects of decades of healthcare and lifestyle habits, it is unlikely that Medicaid expansion per se resulted in a longer life expectancy but the difference in life expectancy instead reflects differences in long-standing approaches to health in different states.
  2. Smoking prevalence. Smoking rates are not much different in states that expanded Medicaid (17.3% of adults) compared to those states that did not expand Medicaid (18.0% of adults) as illustrated in a report by the United Health Foundation. In a previous post, I noted that cigarette smoking cuts the life expectancy of men by an average of 12 years and women by an average of 11 years. Given that the smoking prevalence is reasonably similar in both groups of states, the longer life expectancy in Medicaid expansion states cannot be fully attributed to smoking, alone. It should be noted, however, that the smoking prevalence in the non-Medicaid expansion states is heavily skewed by Utah, where the smoking prevalence is only 8.8% (dramatically lower than any other state and likely due to the Mormon influence in Utah). When Utah is excluded from the analysis, the smoking prevalence in non-Medicaid expansion states is 18.5%, which is significantly higher than in Medicaid expansion states.
  3. Hospital closure. Since 2010, there have been 85 rural hospitals that have closed in the United States. Most of these closures were in those states that opted out of Medicaid expansion. Overall, 59 rural hospitals have closed in states that did not expand Medicaid (3.5 per state) compared to 26 rural hospitals closing in Medicaid expansion states (0.8 per state). Unlike life expectancy and smoking prevalence, it is likely that hospital closures are directly related to Medicaid expansion. Rural and safety net hospitals usually serve a lower income population of patients and are more dependent on revenue from Medicaid than urban and suburban hospitals. In the past, those rural hospitals were supported by federal Disproportionate Share Hospital (DSH) funds that helped to cover the cost of caring for the uninsured and underinsured. The DSH funds are distributed based on a formula that includes the number of Medicaid patients seen at the hospital. However, the Affordable Care Act reduces DSH funds in order to offset the cost of expansion of Medicaid. Therefore, rural hospitals in states that did not expand Medicaid had less money coming in from DSH funds plus no increase in Medicaid revenues and so they found themselves going out of business. The states with the most hospital closures were Texas (14) and Tennessee (8), both states that did not expand Medicaid.
  4. Obesity. States that adopted Medicaid expansion have a lower prevalence of obesity (29.2%) than those states that did not adopt Medicaid (31.3%). Medicaid expansion has been in place for 5 years and it takes a lot longer than 5 years for the population of a state to become obese. Therefore, it is unlikely that there is a cause and effect relationship between Medicaid expansion and obesity. This is supported by the fact that if you examine maps of the trends of obesity in the United States over the past 20 years, the states that did not expand Medicaid had a higher prevalence of obesity before Medicaid expansion.
  5. Infant mortality. The United States has a disturbingly high overall infant mortality rate compared to the rest of the industrialized world. In fact, the only economically developed countries that have a higher infant mortality rate than the U.S. are Mexico and Turkey. Infant mortality is a good measure of access of care to economically and demographically disadvantaged populations since mothers are young and often uninsured or underinsured. Data from the CDC indicates that in 2016, states that participate in Medicaid expansion have an infant mortality rate of 5.8/1,000 live births as compared to states that did not participate in Medicaid expansion that have an average of 6.6/1,000.
  6. Teen birth rate. This measure tends to track with the infant mortality rate. Based on 2016 data from the CDC, the teen birth rate is higher in states that did not expand Medicaid (21.2%) compared to those states that did expand Medicaid (19.0%). There are many risk factors for teenage pregnancy including dropping out of high school, alcohol or drug use, lack of access to birth control, and poverty.
  7. Education. Based on a report by the U.S. Census, it turns out that there is less than one percentage point difference in the percent of adults completing high school in Medicaid expansion states (88.5%) compared to states that did not expand Medicaid (87.6%). However, there is a much larger difference in the percent of adults with a college degree in states that expanded Medicaid (30.1%) compared to states that did not expand Medicaid (26.8%). This suggests that the overall educational level in Medicaid expansion states is higher than in non-Medicaid expansion states.
  8. Firearm-related deaths. States that expanded Medicaid have a lower rate of deaths due to firearms (11.9 per 100,000 population) compared to states that did not expand Medicaid (15.3 per 100,000 population). There are many risk factors for death involving firearm including the availability of firearms and the prevalence of mental health disorders. The map of firearm mortality shown here is fairly similar to a map of gun ownership per capita so it does appear that guns are more abundant in those states that did not expand Medicaid. However, it is also possible that Medicaid expansion resulted in patients with depression and suicidal ideation having more access to mental health services thus making those people less likely to kill themselves or others with guns.

With the exception of rural hospital closures and possibly infant mortality, I do not believe there is a direct cause-and-effect relationship between Medicaid expansion and most of the indices of health discussed in this post. Instead, I believe that whether or not a state chose to expand Medicaid is a reflection of that state’s cultural attitude toward health. The residents of Medicaid expansion states probably had a higher valuation of health, had adopted more healthy lifestyle habits, and were more educated about health than those residents of states that elected to not expand Medicaid. There are also states that are clearly exceptions. For example, Utah (a non-Medicaid expansion state) has the lowest smoking prevalence, one of the longest life expectancies, no hospital closures, a high percentage of collage graduates, and very low rates of obesity, infant mortality, and teen pregnancy. However, because of availability of Medicaid for the disadvantaged population, for the majority of the country, differences in the adoption of Medicaid expansion has the potential to result in the healthiest states becoming more healthy, and the least healthy states become less healthy.

July 5, 2018

Medical Economics

Gaming Hospital Metrics

Hospital and physician performance is measured by a lot of different quality metrics and everyone wants to look good. Sometimes, there are ways to make yourself look good without actually being good – by gaming the metrics. Nobody actually talks about these gaming strategies but a lot of people think about them. I’m not recommending that anyone actually do these strategies but you should be aware of them so that when a physician or hospital reports fantastic quality metrics, you can be sure that the quality is real. Here are some of the ways to game the metrics:

  1. Never write an admission order in the emergency department between 9:00 PM and midnight. Hospitals calculate length of hospital stay based on the number of days that a patient is an inpatient. But… most hospitals count those numbers of days by the midnight census. So, a 4-day length of stay means the the patient was an inpatient for 4 midnights in a row. An easy way to cut a day off of the length of stay is to wait until after midnight to admit patients who show up in the ER in the evening. With electronic medical records, the determination of when a patient gets admitted is usually based on when the admission order is placed. Therefore, the sly hospitalist who wants to keep his/her length of stay low will do all of the work of admitting patients in the evening but not actually sign the admission order until 12:01 AM.
  2. Transfer long-length of stay patients to another service just before discharge. Outliers (those patients who are in the hospital for many more days than would be normally expected for their medical condition) are like hot potatoes – no one wants to be holding them when the get discharged. This is because the hospital length of stay is generally credited to the discharging physician and not necessarily the physician who cared for the patient during the rest of the hospital stay. So, the surgeon with a patient with catastrophic post-operative complications who has been in the hospital for 3 months would want to transfer that patient to the endocrinologist for “diabetic care optimization” a day or two before discharge – that way the endocrinologist gets credited with the 3 month length of stay.
  3. Keep unhappy patients in observation status. Hospitals have to report their patient satisfaction scores and internally, hospitals track patient satisfaction by physician – these scores are often used as part of bonus incentives for hospital-employed physicians. Patient satisfaction is measured by the HCAHPS surveys that are sent to patients after discharge. However, Medicare only requires patients who are in inpatient status to get the HCAHPS surveys and so patients in observation status do not get sent HCAHPS surveys. So, if a physician has an angry patient who they know is going to give them a terrible HCAHPS score, by keeping that patient in observation status, even if it is for 4 or 5 days, then the physician can keep the HCAHPS survey out of their hands.
  4. Give all of the hospital staff buttons that say “I’m a 10”. The HCAHPS survey boils down the results to those responses that are “top box” which means that they are rated either a 9 or a 10 on the 10-point rating scale. Medicare doesn’t allow hospitals to try to directly influence how a patient answers on the surveys (for example, a nurse discharging a patient cannot tell the patient “We’ll be sending you a survey about your hospital stay and I hope that you will rate us a 10”). However, every advertising executive knows the importance of subliminal messages. So by putting the number “10” at the top of the patient’s menu, on the inpatient walls, and on signs in the hospital lobby, subliminal messaging can work its wonders.
  5. Keep patients who are admitted near death in observation status. Inpatient mortality rate is also a metric that physicians (and hospitals) are measured by. However, if a patient dies in observation status, then that patient does not count toward the hospital’s reported mortality rate. Observation status is used for patients who are anticipated to be in the hospital for less than “2 midnights” and means that the patient remained an outpatient without being formally admitted to the hospital. So, no matter how sick a patient is, if the physician anticipates that the patient is going to die before two midnights pass, then by keeping that patient in observation status, that patient never counts toward the physician’s (or the hospital’s) reported mortality.

“Gaming the system” is defined as using the rules and procedures meant to protect a system in order, instead, to manipulate the system for a desired outcome. It is a law of nature to attempt to gain advantage within one’s environment in order to succeed and there will always be those who interpret the rules to their advantage. Awareness of how the rules can be manipulated can prevent us from being the ones who are taken advantage of.

June 23, 2018

Medical Economics

The 2017 Medicare Hospital Star Ratings: Better But Still Not Perfect

Two years ago, I wrote a (rather scathing) post about the 2016 Medicare star rating system for hospitals. Last January, I posted an update about the 2017 star ratings. The 2017 star ratings were supposed to be published last summer but Medicare delayed the release in order to revise the methodology and address some of the criticisms in the original formula. The 2017 star ratings were released this past winter and we anticipate the new ratings to be released in the next 2 months. Since the 2017 star rating was released, there has been more of an opportunity to evaluate who the star winners and losers were.

In 2016, there were 62 measures across 7 groups of quality markers and hospitals reported only those measures that pertained to their patients. Smaller and specialty hospitals that take care of more selective groups of patients reported smaller numbers of measures. For example, an orthopedic specialty hospital would not report on obstetric care measures since no obstetric care is delivered in that hospital. In an analysis of the 2016 star ratings, an article in JAMA determined that there was in inverse relationship between the number of measures that a hospital reports on and the number of stars that a hospital was awarded. In other words, the more measures reported, the lower the number of stars Medicare awarded those hospitals. There were two ways to interpret this data: either smaller and specialty hospitals provide better care to patients or the equation that Medicare used to determine the star rating was flawed.

The 2017 star rating system improved on the 2016 system with the result that there were more hospitals with 1 or 5 stars and fewer with 3 stars, thus decompressing the center of the rankings. However, once again, specialty hospitals reported only about half the average number of measures as major teaching hospitals (27.2 versus 51.4) but had significantly higher star ratings than major teaching hospitals. In all, 83% of specialty hospitals received a 4 or 5 star rating whereas 24% of major teaching hospitals received 4 or 5 stars.

It appears that how the different measures are weighted affects results significantly so that some hospitals can appear to be better or worse performers than they really are. Teaching hospitals care for a completely different population of patients than specialty hospitals – the patients are sicker, have different medical/surgical conditions, and have longer average lengths of stay. Patients at a tertiary care academic medical center have a higher mortality rate and readmission rate than patients at an orthopedic specialty hospital.

The 2018 Medicare star ratings will be released next month. We’ll have to wait to see if the formula improves over last year’s.

June 10, 2018

Inpatient Practice Medical Economics

The High Cost Of Observation Status

Medicare and commercial insurance companies love observation status. When a patient has to be hospitalized but only for “less than 2 midnights hospital stay”, then that patient is classified as being in observation status rather than admitted to the hospital. This classification means that the patient is technically an outpatient and not an inpatient and therefore the cost to Medicare is considerably less. Medicare uses this as a way to control the high cost of health care. But is it really less expensive when you look at the big picture?

To understand observation status, you have to understand the difference between Medicare Part A and Part B:

  1. Part A – covers inpatient care, skilled nursing care, home health care, and hospice care. Part A has no co-pay.
  2. Part B – covers outpatient medical and surgical care, emergency department care, durable medical equipment, and physician charges. Part B has a 20% co-pay.

So, if a person is in observation status, then Medicare Part A does not cover the bills, instead, Part B does. From the patient’s standpoint, this is a critical distinction because if that patient is considered an inpatient, then Part A covers the inpatient charges, including medication costs, with no co-pay. On the other hand, if that patient is considered to be in observation status (i.e., an outpatient), then Part B covers it but the patient is billed a co-pay and importantly, the patient is also billed the medication costs.

If you read the Medicare website, it sounds like the decision about whether a patient is considered as observation or inpatient is the hospital’s decision but this really could not be farther from the truth. The Medicare website states: Your hospital status (whether the hospital considers you an “inpatient” or “outpatient”) affects how much you pay for hospital services. But the hospital is not really the one making the decision about inpatient or outpatient. Medicare has very strict rules about what they consider to be criteria for inpatient status versus observation status. If the hospital believes that the patient should be inpatient status, and bills Medicare Part A, then Medicare can audit that patient’s chart and if they determine that the patient should have really been in observation status by their definition, then Medicare will ask for the Part A money back and in some cases can even fine the hospital for “fraudulent” billing. Moreover, if the patient has already been discharged then the hospital cannot bill the patient for their Part B co-pay or for the cost of the medications that the patient received due to the “MOON” regulations. In other words, if the hospital bills the patient as an inpatient but Medicare does not agree, then the hospital loses a boatload of money on that patient. Therefore, the hospital really, really wants that patient to be in inpatient status and not in observation status but risks not getting paid at all if Medicare disagrees with the inpatient status decision.

Another key difference between observation status and inpatient status is that there is a weird rule in Medicare that a patient has to be an inpatient for 3 days before being discharged to a skilled nursing facility for Medicare to pay for that nursing home charge. If the patient is in observation status, then the hospital cannot discharge the patient to a nursing home and instead has to discharge the patient to their regular home.

The reason Medicare and commercial insurance companies like observation status so much is because they don’t have to pay the hospital as much. But the costs do not go away, they are just transferred to the patient, instead. Many patients are shocked when they were sick enough that they needed to be hospitalized but then get an enormous bill for a 20% co-pay for all of their costs plus a bill for all of the medications that they received when they were in the hospital. These additional co-pays and medication bills require additional layers of administrative costs on the part of the hospital in order to bill and collect and can also be both costly and confusing to the patient.

So, from the patient’s perspective and from the hospital’s perspective. it is better to be an inpatient than to be in observation status. But from Medicare and insurance companies’ perspective, it is better to be in observation status than inpatient status. This has resulted in the hospitals becoming a battle ground for deciding who is an inpatient and who is in observation status.

The most recent victim of this battle is the knee replacement. in the past, knee replacement surgery was considered an inpatient procedure but recently, Medicare has classified knee replacement as an outpatient surgery. But almost no one goes home the same day that they have a knee replacement – those patients need physical therapy, need to recover from general anesthesia, and need to have their pain managed. The result is that almost all patients spend at least one night in the hospital after a knee replacement.

But Medicare will allow for a knee replacement surgery to be billed as an inpatient if the physicians and the hospital can document extenuating circumstances why that particular patient needs to be an inpatient (in other words, why that patient is expected to spend more than 2 midnights in the hospital). These extenuating circumstances are usually co-morbid medical diseases, like sleep apnea, heart failure, insulin-dependent diabetes, etc. But the catch is that these conditions need to be documented in the patient’s chart before surgery because Medicare rules require the decision about whether an order for inpatient admission to be made before or at the time of surgery. This generally means that the patient has to go to a “pre-admission testing evaluation” by a physician, nurse practitioner, or physician assistant where these medical illnesses can be laid out in a way that justifies the surgery taking place as an inpatient procedure.

Furthermore, the hospital has to employ a “physician advisor” who can then review the chart and confirm that the patient actually meets the requirements for inpatient status. The physician advisor must then document his/her opinion about whether the patient really needed to be an inpatient or not and also document their reasoning why so that the hospital has documentation to submit to Medicare auditors in the event that the patient’s admission is denied by Medicare. Many smaller hospitals cannot afford to have a group of their own physicians who are trained in the nuances of inpatient versus observation status and be on-hand for 24 hour decision-making so they will contract with an external physician advisor company such as EHR (Executive Health Resources) who they will pay to have physicians who can review the patient’s chart and offer a determination about whether or not the patient should be inpatient or observation status.

In the final analysis, Medicare and insurance companies pay less for observation status, the hospital has to add an expensive layer of administrative costs, and the patient is personally responsible for more of the costs. The net effect is more total societal costs to deliver health care but lower costs directly paid by Medicare.

So, does observation status reduce healthcare costs? The sad answer is no. It actually adds administrative costs and transfers those costs back to the patient or to the hospital.

June 3, 2018

Medical Economics Physician Retirement Planning

Age Of Physicians By Specialty

At this month’s American Thoracic Society meeting, it was reported that 1/3 of practicing pediatric pulmonologists in the United States are over age 60, a scary number since that indicates we are soon facing shortages of pediatric pulmonologists. It turns out that it is not the only specialty with disproportionately older physicians and these statistics have implications for our future physician workforce. In the U.S., air traffic controllers have a mandatory retirement age of 56 years-old, national park rangers are 57, military officers are 64, and pilots have a mandatory retirement age of 65. In the Roman Catholic Church, priests have a mandatory retirement age of 70. There is no mandatory retirement age for physicians and consequently, some specialties have become very top-heavy with older doctors.

The Association of American Medical Colleges tracks physicians in different specialties by the percent who are under age 55 versus those who are over 55 and the data is summarized in the graph below:

For all physicians combined, 56.8% are under age 55 and 43.2% are over age 55. However, some specialties are disproportionately older or younger. My own specialties of pulmonary (15% under age 55) and critical care (84.3% under age 55) is probably more a reflection that most of these physicians are dual certified and tend to do more critical care earlier in their career and migrate to more pulmonary later in their career. Similarly, emergency medicine with 65.4% under age 55 is a relatively new discipline that did not become recognized as a specialty until 1979 and did not offer a board examination until 1980.  There are some specialties that are more concerning. For example, pathologists, psychiatrists, cardiologists, and thoracic surgeons tend to be older whereas interventional radiologists, nephrologists, and pediatricians tend to be younger. Those specialties that have more than 50% of the physicians over age 55 are likely to be in high demand in the next 10 years as these older physicians retire.

A study in the Journal of Medical Regulation from 2017 analyzed the U.S. physician workforce by a number of parameters, including age. Taking all physicians together, 29.3% of practicing physicians are over the age of 60. The reasons why there are so many older physicians in the workforce are complex and include (1) a later age of entry into the workforce due to lengthy training requirements, (2) a high amount of debt from the cost of education, and (3) less physical demands than many other professions.

The median age of retirement from clinical activities by physicians is age 65 years as shown in this graph from a study published in the Annals of Family Medicine. The retirement age varies by specialty, for example, the median age of retirement from clinical activities is about 64.5 years for OB-GYNs and 66.5 years for cardiologists. Women tend to retire 1 year earlier than men. Because many physicians continue to be active in other professional activities after retirement from clinical activities (such as administration or education), the median age of retirement from any professional activity tends to be about 1 year later than the retirement from clinical activity. Therefore, the median age of retirement from any professional activity is at age 66 years when examining all physicians. But remember that these data are for the median age of retirement and that means that half of all physicians retire from clinical activity after age 65 years.

As a medical director, one of the most uncomfortable tasks I have to do is to tell an older physician with a long history of dedication to the medical profession and the community that it is time that he or she needs to stop seeing patients. It is not because of age per se but because of quality concerns. It turns out that this is a valid issue. A study in the BMJ found that for Medicare patients, the 30-day survival after hospital discharge depends on the age of the physician. 30-year old physicians had a 10.5% 30-day patient mortality rate whereas 70-year old physicians had a 13.5% 30-day patient mortality rate. Although part of these results could be because older physicians tend to have combined inpatient and outpatient practices with an older (and sicker) panel of patients whereas younger physicians tend to be hospitalists that care for a wider age range of Medicare patients, it is also quite possible that older physicians do not practice as high of quality of medical care as younger physicians. This has unfortunately been my experience with some older physicians.

There can be a lot of reasons why physicians retire and last year I wrote a blog post about “When Physicians Reach Their Use-By Date” to reminisce about how some of the more memorable physicians I have known retired. The keys are to have enough self-awareness to know when your clinical skills are lagging behind your peers and to be willing to pick up on subtle hints from those peers that you are not the clinician that you used to be.

So, what does all of this mean? First, doctors retire later than people in many other professions. Second, doctors who chose to work beyond age 65 need to be attentive to their quality of practice. Third, and perhaps most important from a national health care standpoint, certain specialties are dominated by older physicians who will be retiring soon, thus creating demands for those specialties that will be difficult to meet.

The hockey legend, Wayne Gretzky, famously said: “Skate to where the puck is going, not to where it has been”. I think that this has implications for our medical students who are selecting specialties – knowing what specialties are going to be in demand rather than what are currently in demand should affect their career choices.

May 26, 2018

Hospital Finances Medical Economics

How Many Days Cash On Hand Should A Hospital Have?

When a hospital runs a positive margin and makes money at the end of the year, everyone wants some of it – hire more doctors, hire more nurses, buy a new MRI machine, build a new hospital wing. It can be tempting to spend it all but should you? Liquidity is survival insurance for a hospital and it is essential to hold some money back. Every individual person should have an “emergency fund” with 2-6 months of expenses held in a checking/savings/money market account, and so should hospitals. These cash reserves are called “days cash on hand” and represent the amount of money it takes to pay all of the hospital’s expenses for that number of days.

A few year ago, Dr. Charles V. (Bo) Sanders gave a presentation at the annual Association of Professors of Medicine meeting that I was attending. He is the Chairman of the Department of Medicine at LSU School of Medicine in New Orleans and was describing the effect of Hurricane Katrina on the hospital and his department. Katrina flooded Charity Hospital which then closed, displacing most of the doctors in his department. With limited cash reserves, the hospital could not pay physician salaries and many of the doctors and nurses moved on. The hospital essentially died and was never able to reopen.

Charity Hospital is just one example of why a hospital needs to have sufficient days cash on hand but there are many things other than hurricanes that can temporarily close a hospital and require it to draw from cash reserves to cover payroll so that all of the employees don’t leave. Fire, flood, lapse of malpractice insurance coverage, prolonged power outage, unpredictable admission rates, you can think of a dozen other reasons that a hospital might have to reduce or close operations for a day, a week, a month, or longer.

In 2011, Moody’s Investors Service reported on the financial statements of 400 hospitals in their database. The overall median number of days cash on hand was 165 with a mean of 183. The range was from 11 days to 521 days. A 2013 analysis of critical access hospitals reported a median of 68 days. More recently, in 2015, Moody’s reported that the average for 350 hospitals and health systems had increased to 212 days. A 2014  Fitch Ratings report of nonprofit hospital and healthcare systems found that the credit rating of the hospital correlated with the number of days cash on hand with the “AA” hospitals having 289 days cash on hand and hospitals having a “BBB” rating only having 161 days cash on hand. A 2016 report by S&P Global Ratings indicated that “AA+” hospitals had 420 days cash on hand whereas “BBB+” hospitals had 149 days cash on hand; speculative grade hospitals (those that finance with “junk bonds”) had only 74 days cash on hand. The implication is that if your days cash on hand is high, the hospital’s credit rating is high and consequently, the hospital can get a better interest rate on bonds to do expansion, etc. In other words, more cash on hand equates to lower interest rates for loans.

From this analysis, it appears that the number of days cash on hand that is held by hospitals appears to be increasing over the past decade. I believe that there are at least three reasons. (1) The economic recovery since the great recession has led to overall better financial positions of U.S. hospitals. (2) Medicaid expansion (by those states that elected to participate) has led to a reduction in uninsured patients and this translates to improved margins. (3) Better analytics that are attendant to electronic medical records, better inventory management programs, and improved staffing programs has resulted in better hospital operational efficiency as well as better hospital billing efficiency.

There are a number of factors that can influence the ideal number of days cash on hand for any given hospital:

  1. Geographic location. Hospitals in areas vulnerable to natural disasters are themselves vulnerable to unexpected closings. For example, hospitals in low-lying coastal cities are vulnerable to flooding – Charity Hospital is an example of this. Other geographic vulnerabilities include susceptibility to regional wildfires, tornados, and earth quakes.
  2. Centralized versus decentralized. A hospital system built around a single large hospital (standalone hospital) is more vulnerable to closing operations than a hospital system with multiple buildings in different locations in the region. For example, if a water line breaks and floods the operating rooms taking them out of commission for 3 months to renovate, the centralized hospital will have no place to perform surgeries whereas the decentralized hospital system can redirect surgeries to alternative locations. The S&P bond rating analysis confirms this and indicates that standalone hospitals that had a “AA” bond rating had 100 more days cash on hand than decentralized health systems.
  3. Need for borrowed money. A hospital that is planning a $750 million expansion is going to need to borrow money by selling bonds. The current interest rate on a 20-year AAA rated municipal bond is 3.00% whereas an A rated bond is 3.50%. That is a $3.75 million dollar per year difference and over the course of the bond, a total of $75 million dollars additional cost for just that slight increase in the interest rate on the bond. For a “B” rated hospital, the difference in the interest rate that they can get on a bond can be even more, up to a full 1% higher. A hospital with a higher number of days cash on hand will be able to get a better bond rating.
  4. Need to pay off borrowed money. Most hospitals will have both cash and loans. If those loans were taken out at relatively high interest rates, then it may be more desirable to pay them off using the hospital’s cash. This can reduce the number of days cash on hand but can strengthen the hospital’s long-term financial position.
  5. Competitiveness of the regional health insurance market. When hospitals negotiate rates with commercial health insurance companies, one of their greatest leverage points is the ability to walk away from the table. By that I mean, the ability to tell the insurance company that if they won’t give the hospital the reimbursement rates that the hospital wants, the hospital will stop taking patients covered by that insurance company. If that hospital is the only hospital in a 50-mile radius, then they have a  pretty good bargaining position because the insurance company can’t easily send their patients to another hospital. On the other hand, if there are a number of other hospitals in the community, then the hospital has less leverage because the insurance company can simply redirect its insured patients to another hospital in town. If such a hospital has relatively few days cash on hand, then the commercial insurance company will know that the hospital really has no bargaining position since they don’t have the resources to survive a sudden drop in admissions if the insurance company sends them all to a competitor hospital. On the other hand, if such a hospital has a lot of days cash on hand, the threat of walking away from the table is much more real and that stronger negotiating position is more likely to translate into higher reimbursement rates from the insurance company. Thus, a hospital in a region with other competing hospitals needs to have more days cash on hand in order to effectively compete for the best insurance reimbursement rates.
  6. Anticipated large capital purchases. Hospitals will not generally sell bonds for purchases such as a new electronic medical record but these can be quite costly and are better paid for out cash. If the hospital plans on buying a new EMR or some other large-priced purchase in the next few years, then it it best to increase the days cash on hand in anticipation of that purchase.
  7. Admission fluctuation. Our hospital in Central Ohio has a fairly consistent number of admissions per month; it tends to go up during the influenza season but otherwise is fairly constant. On the other hand, a hospital in a ski resort community in Colorado may see a significant rise in admissions in the winter whereas a hospital in the Outer Banks of North Carolina may see a significant rise in admissions in the summer. Hospitals with greater fluctuation in admissions and ambulatory visits will need to have more days cash on hand than those with very predictable admissions and visits.
  8. Medicaid expansion. Between 2010 and 2018, 83 U.S. hospitals went out of business. The overwhelming majority of these were in states that did not expand Medicaid under the Affordable Care Act. In all, 19 states did not expand Medicaid and 63 of the hospitals that closed were in these states. That means that 76% of all hospitals that closed were in these states. Six of these states had more than 5 hospitals close: Alabama (5), Mississippi (5), North Carolina (5), Georgia (6), Tennessee (8), and Texas (14). Hospitals in states that did not participate in medicaid expansion have been faced with higher numbers of uninsured patients and are at a competitive disadvantage to hospitals that did expand Medicaid. Having a larger number of days cash on hand is desirable for hospitals in states that did not expand Medicaid.
  9. Donor attractiveness. Wealthy donors are wealthy because they have a lot of financial sophistication. Donors will often examine the financial viability of a hospital before committing large endowments: why donate to a hospital that is on the verge of going out of business? More days cash on hand is one way of demonstrating the hospital’s financial solvency and stewardship. More days cash on hand can translate to larger endowments from wealthy donors.

Although having a large number of days cash on hand sounds good, too high of a number can be bad. For example, it may be better to invest that money in a better-paying investment, for example expanding the hospital’s primary care base by hiring additional primary care physicians. Or, if the hospital’s quality metrics are below average, it is better to spend additional money to improve patient satisfaction, decrease hospital readmissions, or improve the infection control efforts. For public hospitals that are owned by the city, county, or state, having too high a number of days cash on hand can create a perception to lawmakers and the public that the hospital is hoarding the public’s money.

So, what is the best number of days cash on hand? From this post, it should be clear that there is not a single best number for all hospitals. I’d start with a number of about 300 for standalone hospitals and about 250 for hospital systems. Then move that number up or down depending on all of the variables mentioned above. For a decentralized health system in a Medicaid expansion state that does not have excessive natural disaster risk and does not anticipate purchasing large amounts of bonds, 130 days may be plenty. On the other hand, a centralized standalone hospital in a competitive market in a state that did not participate in Medicaid expansion and is at risk of natural disasters and also plans on a major building expansion requiring bond sales, 400 days may be more desirable.

The number of days cash on hand is something that physicians rarely think about and almost never incorporate in their decision-making about whether to take a job at a hospital. But I think that physicians should take notice of this number and if it is too low, ask the hospital administrators why it is low. If you can’t get a good answer, think long and hard about whether that hospital represents a risky career choice.

May 5, 2018