Of all topics pertaining to medicine, perhaps none is more consequential than our workforce. It affects our ability to meet patient needs in a competent manner, speaks to how we interact and compete with one another, shows the adequacy of our training systems and how they change over time, and handles what may be required of us in the future. The otolaryngology workforce is also a factor in our ability to sustain a rewarding practice and provide for our families. Because its health affects all of us in serious ways, it requires a careful and routine analysis.
Explore This Issue
May 2022This three-part series is my attempt to provide an updated analysis of where the health of our workforce is headed. In this first part, we’ll examine some evidence from published studies and updated supply models about what’s really happening when it comes to supply and demand in our workforce. Our specialty was initially worried about whether we would have enough otolaryngologists when we perhaps should have been worrying about whether we might have too many.
Historical Supply Figures
To understand our current situation, we need a historical understanding of our workforce supply. While other studies preceded this, the Academy of Otolaryngology–Head and Neck Surgery (AAO-HNS) commissioned a workforce study in the late 1990s. Among many important findings in this study was that in 1997 we had 9,017 otolaryngologists, or 3.36 otolaryngologists per 100,000 population. It was thought that this overage would correct itself over the ensuing 20 years to the managed care demand range estimate of 1.8 to 3 otolaryngologists per 100,000 due to the aging of the workforce and population growth (Otolaryngol Head Neck Surg. 2000;123:341-356).
A follow-up study was commissioned and published in 2004. The authors made some interesting observations—in particular, that the workforce supply trends were the opposite of those that had previously been projected. The total number of otolaryngologists and number per 100,000 population were increasing in all areas of the country. We had risen from 3 otolaryngologists per 100,000 in 1995 to 3.2 per 100,000 in 2002 (Otolaryngol Head Neck Surg. 2004;131:1-15). An American College of Surgeons Bulletin article in 2012 also supported the idea that our supply per 100,000 was increasing, with 2.72 otolaryngologists per 100,000 population in 1981 rising to 3.32 per 100,000 in 2006 (ACS Bulletin. March 1, 2012). A slight decline to 3.26 was noted around 2009. And it was then that the shortage narrative surfaced.
Several studies peering into the horizon sounded alarm bells about the otolaryngology supply. While the American Association of Medical Colleges (AAMC) issued attention-grabbing headlines, starting in the 2010s, about looming shortages of tens of thousands of physicians, we had specific studies highlighting our apparently dire situation.
Predictions of severe specialty shortages over the next 20 years began as early as 2002 (Acad Med. 2002;77:761-766). New modeling forecast a shortage of between 1,600 (HRSA. 2016;1-14) and 2,300 (Otolaryngol Head Neck Surg. 2012;146:196-202) otolaryngologists per 100,000 by 2025, and 2,500 by 2030 (Ann Surg. 2009;250;590-597). New demand estimates from the 2012 paper also showed that we now needed 3.11 otolaryngologists per 100,000 population, which accounted for part of our expected shortfall.
One of the more recent analyses of our workforce supply, published in 2016, showed that we had 10,522 otolaryngologists in 2011 and 10,800 in 2014 (Laryngoscope. 2016;126:S5-S11). Despite these supply increases, we were still projected to have a shortfall. What struck me about this last study is that we were edging closer to the time periods of predicted supply shortfalls, yet our supply (and supply per 100,000 population) was increasing. I began to question the supply numbers we’d been relying on. Every study, save this most recent study, had used American Medical Association (AMA) supply data.
This begged the question as to whether AMA data are even accurate. I believe there’s reason to be skeptical.
To try to understand the accuracy of the AMA supply data, I looked at the American Urological Association (AUA) census data, which has been gathered since 2014. The AUA uses two internal databases implying some level of active practice (AUA roles and board certification data) as well as the NPI file data to construct a list of possible practicing urologists and then further ensures ongoing practice by confirming names on at least two of these databases. Physicians not meeting this criteria are then systematically checked to ensure each individual is still practicing—if physicians aren’t found, they aren’t included. The AAMC publishes a specialty data report every other year using the AMA physician master file that most of the previous studies used. Data were available for recent odd years for comparison.
The total urology workforce was approximately 28% greater than the AMA numbers would suggest. The AMA supply data are based on an “actively practicing” definition of 20+ hours of work per week. The AUA also calculates “actively practicing” urologists, although their criteria for active practice is more stringent at 25+ hours per week. And, under these more stringent definitions, the AUA workforce supply analysis was still between 1% and 11% greater than the AMA supply numbers over the three comparison years (Table 1).
Table 1. © Andrew Tompkins, MD, MBA
The 2016 Laryngoscope study used our internal AAO-HNS and American Board of Otolaryngology Head and Neck Surgery (ABOHNS) data, not AMA data. Maybe this study’s numbers were more spot on.
Current and Future Supply Modeling
To understand our current situation, I looked at all of our residency programs and counted each resident by graduation year, accounting for research years. In 2021, we graduated 330 residents. Had you looked at the 2016 National Resident Matching Program (NRMP) match positions, you would have predicted 302 graduates, but this number wouldn’t account for DO or military programs, which are not included, or those who take research years or drop out. The difference between NRMP position prediction and reality was 8.6%.
I then applied this 9.3% difference to the predicted graduate number, based on NRMP positions five years prior, back to the 2011 graduation year, in order to derive a predicted actual graduate number for these preceding years (National Resident Matching Program, Results and Data: 2006-2016 Main Residency Match. National Resident Matching Program, Washington, DC. 2006-2016.). Using the 2016 Laryngoscope study supply inputs, yearly U.S. population numbers from the United Nations, and the AMA attrition rate of 1.7% (from the 2016 study), we can see that our numbers appear to be increasing both in absolute terms and also on a per 100,000 population basis. We now appear to stand at 3.5 otolaryngologists per 100,000 population—well over previous demand predictions (Table 2, below).
Table 2
| Year | Predicted* | Actual** | U.S. Population | Otolaryngologists | Attrition | Oto/100K |
|---|---|---|---|---|---|---|
| 2011 | 259 | 283 | 311,584,047 | 10,522 | 179 | 3.38 |
| 2012 | 268 | 293 | 314,043,885 | 10,626 | 181 | 3.38 |
| 2013 | 269 | 294 | 316,400,538 | 10,738 | 183 | 3.39 |
| 2014 | 273 | 298 | 318,673,411 | 10,800 | 184 | 3.39 |
| 2015 | 279 | 305 | 320,878,310 | 10,915 | 186 | 3.40 |
| 2016 | 280 | 306 | 323,015,995 | 11,034 | 188 | 3.42 |
| 2017 | 283 | 309 | 325,084,756 | 11,152 | 190 | 3.43 |
| 2018 | 290 | 317 | 327,096,265 | 11,272 | 192 | 3.45 |
| 2019 | 295 | 322 | 329,064,917 | 11,397 | 194 | 3.46 |
| 2020 | 298 | 326 | 331,002,651 | 11,526 | 196 | 3.48 |
| 2021 | 302 | 330 | 332,915,073 | 11,656 | 198 | 3.50 |
*Predicted residency graduates based on NRMP filled positions from five years prior
** Actual based on 9.3% higher rate on direct measure for 2021 class
Note: There was no difference the workforce projection when based on positions offered or positions filled–both yielded 3.50 otolaryngologists per 100,000 population.
© Andrew Tompkins, MD, MBA
While the AMA attrition rate may be inaccurate, the 1.7% rate would have to be 41% greater, at 2.4% annually, in order to have a stable per capita supply that’s still above previous demand estimates. A 2020 workforce paper (Otolaryngol Head Neck Surg. 2020;162:649-657) used 2019 ABOHNS supply data to show the geographic dispersion of practicing otolaryngologists, which yielded a total of 11,124 based on their calculated per capita ratio and census data. Remarkably, inserting this supply number into the above table yields an attrition rate of 2%, still not enough to stem an oversupply of otolaryngologists on a total number and per capita basis.
One of the reasons for our increasing supply ratio is that U.S. population growth has been on a steady decline for decades, with 2021 being the lowest year on record. The U.S. Census Bureau projects our population growth rate to continue at a steady decline over the next four decades, to 0.4% annually by 2060. Most previous modeling for otolaryngology demand with a projected supply need assumed an annualized growth rate of 0.7%. But we may not see that level of growth again in our lifetimes.
What struck me about [a recent] study is that we were edging closer to the time periods of predicted supply shortfalls, yet our supply (and supply per 100,000 population) was increasing. —Andrew J. Tompkins, MD, MBA
Our supply situation may become more heated in the coming years. Based on my trainee analysis, graduates will increase from 330 in 2021 to 367 in 2025. We have added nine new otolaryngology training programs in the last six years, three of which will start accepting new residents in 2022. Other programs have increased their complements. This rate of growth is more than double that of the previous two decades, at a time when we appear to be in oversupply. This growth belies the notion that our graduate numbers are stable or that we need more GME funding to grow residency slots—our programs are happily doing so anyway.
Projecting our supply per 100,000 population out to 2025 is also instructive. Using the directly measured resident graduate numbers, the AMA attrition rate of 1.7%, and a stable population growth of 0.5%, we should have an expected supply of 3.61 otolaryngologists per 100,000 in 2025. Remember: We were to have significant supply shortages by 2025, below the demand estimate of 3.11 otolaryngologists per 100,000 population.
We aren’t suffering from a deficit—we’re growing in number, and far too quickly.
Fellowship Growth
We’ve also witnessed a steady increase in fellowship training. Studies show that the pursuit of fellowship has increased from an already historically elevated 46% in 2011 to 62% in 2019 (Otolaryngol Head Neck Surg. 2021;165:655-661). Without question we expanded our reach into the head and neck, but have we gone too far? Evidence shows that we almost assuredly have.
Over the subsequent seven years, however, while the U.S. population saw a 4.6% increase, the number of neurotology fellowship positions increased by 37%, fueling a 49% increase in the neurology workforce by 2020. —Andrew J. Tompkins, MD
In 2013, a neurotology workforce study was conducted, estimating that, due to supply numbers and shifting treatment paradigms, we had a 10% to 15% oversupply of neurotologists (Otol Neurotol. 2013;34:755-761). Over the subsequent seven years, however, while the U.S. population saw a 4.6% increase, the number of neurotology fellowship positions increased by 37%, fueling a 49% increase in the neurotology workforce by 2020 (Otol Neurotol Open. 2021;1:e007).
This trend isn’t unique to neurotology. Head and neck oncology appears to be on a similarly compromising path. In 1997, we graduated seven accredited head and neck fellows per year, a number that grew to 43 fellows per year by 2017, with 50 positions offered (Head Neck. 2020;42:1024-1030). Some of this increase was due to assimilation of non-accredited programs, and our ability to expand into skull base and reconstructive surgery has justified some growth. But our incidence of head and neck cancer per 100,000 population has been on a steady decline over this same time period. We may be overestimating the career demand for current graduating fellows.
Rhinology fellowship spots have grown by 14%, compounded annually, from 2006 to 2017 (Laryngoscope. 2020;130:1116-1121). Modeling in this paper predicts that we’ll surpass demand by 2024, with a 40% excess supply of rhinologists in 15 years if nothing changes, assuming that our current rhinologist-to-population ratio is appropriate. But is it? A 2017 survey of rhinology fellowship directors showed that most believe that we’ve been training too many rhinologists—a strong majority (59%) thought this was true with respect to the private practice supply, 85% with respect to academia, and 62% for general healthcare delivery (Am J Rhinol Allergy. 2019;33:8-16).
Rural patients ideally need well-trained general otolaryngologists and convenient access to specialty care. We seem to be trending away from the well-trained generalist, however, toward an overabundance of subspecialty care concentrated in urban settings. —Andrew J. Tompkins, MD
And these thoughts aren’t unique to rhinology. Academic pediatric otolaryngologists had similar thoughts in 2014. Again, a strong majority (70%) said that job prospects nationwide were going to look worse over the coming three years. Over 85% believed this to be true for their local community, where they had more intimate knowledge, with 68% saying job prospects over the next three years would be limited or extremely limited (JAMA Otolaryngol Head Neck Surg. 2016;142:823-827). This study looked out only to 2017. Since then, we’ve increased the number of pediatric otolaryngology fellowship spots by 30%.
Two things happen with fellowship training—a new skillset is produced that needs to be maintained/sharpened, and unused, previously acquired skillsets wane. This trade-off affects more than just the supply of generalists. It can present a safety issue for patients when oversupply leads to underutilization of a newly acquired skillset. On the flip side, a waning skillset decreases the ability to fully meet actual demand. In short, what appears to be happening is that we’re creating excess supply while making that supply less generally capable.
Advanced Practice Provider Growth
Advanced practice providers (APPs)—physician assistants and nurse practitioners—increase our productivity, which increases our effective supply. The 2012 workforce paper modeled out APP growth to 2025, estimating we would have between 2,944 and 3,351 APPs in our specialty by 2025 (Otolaryngol Head Neck Surg. 2012;146:196-202). Two recent studies showed APP growth rates in our field between 2012 and 2017, which demonstrated between 8.7% and 16.6% annualized growth over that time period (Otolaryngol Head Neck Surg. 2021;165:69-75; Otolaryngol Head Neck Surg. 2021;165:809-815). These growth rates are well in excess of the 2012 paper’s supply modeling. It’s important to note, though, that modeling APP growth has proved difficult because few direct supply measurements exist, and the two recent APP studies noted above counted only independent Medicare billing. That means APP supply may be below the 2012 estimates, or well above
And these productivity increases allow us to meet the same patient demand with fewer otolaryngologists. According to a 2016 survey, the average physician assistant had 5,000 office visits and performed 400 procedures a year (Laryngoscope. 2018;128:2490-2499). A recent pediatric otolaryngology APP assessment showed that 90% of academic practices used APPs, who, on average, handled 16% of total visits (Int J Pediatric Otorhinolaryngol. 2020;129:1-4). Unfortunately, none of our prior otolaryngology supply modeling accounted for APP use and the resultant productivity gains in their shortfall predictions.
The Challenges of Rural Care
The challenge of rural healthcare—lack of delivery of talent when and where it’s needed—could be its own article. Our studies are quite clear: Otolaryngologists tend to cluster in urban centers (Otolaryngol Head Neck Surg. 2020;162:649-657). Urban employment is a growing trend by age bracket, with younger surgeons pursuing more urban employment (2019 Urology Census, Page 24). Fewer medical students now hail from rural areas, where they would be more likely to practice (Health Aff. 2019;38:2011-2018). We aren’t uniformly dispersed by either county or hospital referral region (Otolaryngol Head Neck Surg. 2020;162:649-657), and some of our supply gap areas are massive. The question is, can we do better?
Access to care, more specifically convenience and timeliness, affects outcomes. Almost 98% of head and neck surgeons practice in urban settings (Head Neck. 2020;42:1024-1030). Given this number, is it any wonder that we see disparate Kaplan-Meier curves for rural patients, most notably for minority rural patients? A recent 10-year analysis that included tens of thousands of head and neck cancer patients (excluding oropharyngeal cancer) demonstrated this result (Cancer Epidemiol Biomarkers Prev. 2020;29:1955-1961). Differences persisted after controlling for socioeconomic status, demographics, and clinical factors. Stop and think about this for a moment. These aren’t curves—they represent thousands of people who might live with better access and care coordination.
Maximizing the impact of our workforce is perhaps one of the biggest quality improvement initiatives we could undertake. Rural patients ideally need well-trained general otolaryngologists and convenient access to specialty care. We seem to be trending away from the well-trained generalist, however, toward an overabundance of subspecialty care concentrated in urban settings. How much of this migration is driven by generational preferences, or by fellowship training? And are our patients paying the price in both locations due to market saturation of fellows in urban centers and nonavailability of generalists with maximized skillsets in rural settings?
A Cautionary Note
Some may think that my analysis is too bleak. After all, haven’t we always had good jobs available? Hasn’t everyone predicted a shortfall in supply? Didn’t I just read that we’ll need over 100,000 more physicians in just 12 more years? My answer would be to examine the situation with emergency medicine (EM).
Like otolaryngology, EM was in demand and shortages were predicted. But after years of skyrocketing training program growth and APP involvement, researchers now think that in just eight years there will be over 10,000 more EM physicians than jobs available if nothing changes (Ann Emerg Med. 2021;78:726-737). Adaptive changes produce only a modest reduction in this number. (Coincidentally, this amounts to a 13% excess supply in their workforce, which matches our supply excess between the above supply and demand estimates of 3.61 and 3.11 for 2025). If you started medical school right now and wanted to go into EM, you might not find a job.
Correcting this oversupply of EM physicians relies on transparent, widely disseminated knowledge and an adaptable system. As you’ll see in the next article, I don’t believe we have enough transparency in our field, and our system, as presently designed, isn’t accountable.
We have serious issues to investigate if we want to optimize our workforce and care delivery. We should question historical narratives, take ownership of our own data, and seek to understand the truth about what’s happening in the otolaryngology market. To that point, we should also rethink the basic supply metric itself—the number of otolaryngologists per 100,000 population. The truth of our supply adequacy lies in different metrics.
(Next Month: Rethinking Supply)
Dr. Tompkins is a private practice otolaryngologist in Columbus, Ohio.