Chapter 3.  Estimating Pediatric Rheumatology Workforce Requirements

In 2002, the chair of the American Academy of Pediatrics Section on Pediatric Rheumatology called for a doubling of the number of United States pediatric rheumatologists to 400. ²³ The appropriateness of this goal has yet to be evaluated and depends, in part, on the geographic distribution of pediatric rheumatologists and patient demand for pediatric rheumatology care.  Moreover, academic medical centers continue to be the primary employers of pediatric rheumatologists; as such, the educational and research needs of these institutions heavily influence the demand for these providers.

Patient Demand for Pediatric Rheumatology Care by State

Estimating demand for pediatric rheumatology care is challenging.  Because of low incidence rates, national sample surveys of the general population generally fail to identify sufficient patients with juvenile rheumatic diseases to generate reliable estimates.  Using an exhaustive list of 48 possible International Classification of Disease (ICD-9) codes, the National Ambulatory Medical Care Survey (NAMCS) contains only 16 records for rheumatic diseases visits among children under 18.  National Hospital Ambulatory Care Survey (NHAMCS) similarly contains 57 visit records for rheumatic conditions among children under 18.  As a result, reliable estimates of patient demand for care are not possible using such data.

In the absence of a measure of need for pediatric rheumatology services or even utilization of these services, prevalence rates ^[iii] allow estimates of patient to provider ratios as a proxy for patient demand.  Using state-level population data from the Bureau of the Census ²⁴ and physician data from the American College of Rheumatology (ACR) Membership File and American Board of Pediatrics (ABP) Diplomate File, ratios of pediatric population to pediatric rheumatologists were generated for each State.  State level ratios were used because many States have either no pediatric rheumatologist or only one pediatric rheumatologist.  As such, a state-level analysis helps identify relatively underserved States and provides an estimate of the number of pediatric rheumatologists that are needed nationwide.  As in analyses of the MSA-level ratios, these estimates rely on “head counts” and do not adjust for the percentage of time that a pediatric rheumatologist is involved in patient care as these data are not available for each pediatric rheumatologist in the United States.   

Table 9:  Ratio of Pediatric Population to Board-Certified Pediatric Rheumatologists, 2003 American Board of Pediatrics Diplomate File

Table 9:  Ratio of Pediatric Population to Board-Certified Pediatric Rheumatologists, 2003 American Board of Pediatrics Diplomate File, cont.

Ratios were calculated separately using the ACR and ABP files; States without pediatric rheumatology providers are highlighted in yellow in Tables 9 and 10.  Using the ABP file, Washington D.C. ^[iv] has the lowest ratio of children to pediatric rheumatologists at 47,645:1; Texas has the highest with a ratio of 953,206:1 (Table 9).  Assuming a prevalence of pediatric rheumatic conditions of 390 per 100,000 children, ratios range from 186 children with rheumatic disease per provider in Washington D.C. to 3,718:1 in Texas.  Among States that lack Board-certified pediatric rheumatologists, the population size ranges from 126,000 in Wyoming to over 1.3 million in Arizona.        

Table 10:  Ratio of Pediatric Population to Self-identified Pediatric Rheumatologists, 2003 American College of Rheumatology

Table 10:  Ratio of Pediatric Population to Self-identified Pediatric Rheumatologists¹, 2003 American College of Rheumatology (continued)

¹ Excludes trainees and physicians who are not involved in patient care

The ratios change slightly using the ACR data on Board-certified pediatric rheumatologists. The ratio of the pediatric population to pediatric rheumatology providers ranges from a low of 91,225:1 in Delaware to a high of 953,206:1 in Texas (Table 9).  In terms of the number of children with rheumatic diseases per provider, these ratios translate into approximately 356 children with rheumatic diseases per pediatric rheumatologist in Delaware to 3,718 children with rheumatic diseases per pediatric rheumatologist in Texas.  Three States that lack Board-certified pediatric rheumatologists have pediatric populations in excess of 950,000; the remaining 10 States have pediatric populations of fewer than 500,000.

Projected Need for Pediatric Rheumatologists

Previous studies of physician market entry have assumed that the population size needed to support a physician entrant increases with the level of specialization. ²⁵  One model estimated, for example, that the population needed to attract the first family practice physician to an area is 3,300; in contrast, a population of 69,000 was needed to attract the first cardiologist to a market area.  This study also showed that the population increments needed to attract additional providers were smaller than the population needed to attract the first provider.

Table 11:  Estimated Number of Pediatric Rheumatologists Needed by State

¹ Excludes trainees and physicians not currently involved in patient care

One previous report has suggested that a total population base of 1 million is needed to provide sufficient patient demand for a pediatric rheumatologist; ¹⁹   therefore, estimates presented here use this population size as a starting point. Given that children represent approximately 25 percent of the United States population, a pediatric population of 250,000 was used as a threshold for identifying States that could support their first pediatric rheumatologist.  Since a previous study has shown that the population increment needed to attract the second provider is smaller than the population size needed to attract the initial provider, ²⁵ the model assumed that each additional pediatric population increment of 200,000 could support an additional pediatric rheumatologist.  The results that presented here used State level data from the ACR; however, the calculations using ABP data are comparable.

Assuming that a pediatric population of 250,000 is needed to support a pediatric rheumatologist, 9 of the 13 States that currently lack a Board certified pediatric rheumatologist could generate enough demand to support a pediatric rheumatologist.  As Table 11 shows, the population under age 18 in Alaska, North Dakota, South Dakota, and Wyoming are below the 250,000 threshold and, therefore, may not generate sufficient patient demand to support a pediatric rheumatologist.  However, combined regions, such as North and South Dakota may be able to support a pediatric rheumatologist.  Assuming that entry continues with an additional rheumatologist for each additional 200,000 children, a minimum of 331 rheumatologists would be needed in the United States.  Table 10 shows that several States, especially those with training programs, have more pediatric rheumatologists than are “needed” based on population size.  Many of these providers may be primarily involved in research and teaching, creating a situation in which “head counts” lead to an overestimate of actual supply.  If one allows States with training programs to have more rheumatologists than are clinically needed based on population size and considers the current supply per State as a minimum, the number of rheumatologists needed nationwide is 337.

The 250,000 threshold, however, may be unreasonably high.  Given a prevalence rate of 390 per 100,000, this threshold translates into 975 children with rheumatic diseases per rheumatologist.  Given the multitude of needs that characterize this patient population, it is unlikely that one rheumatologist could care for nearly 1,000 patients.  A previously unpublished survey of pediatric rheumatology programs found that the average number of children seen annually by each pediatric rheumatology unit was 443. ^[v]   Only 12 percent of pediatric rheumatology units, including those with multiple providers, saw more than 1,000 children a year.  Therefore, the actual number of patients that a pediatric rheumatologist is able to treat may be far lower than 975.  These estimates easily allow the use of different population thresholds or prevalence rates to estimate the pediatric rheumatology workforce requirements.

Estimates of pediatric rheumatology workforce requirements should also consider the roles of internist rheumatologists and primary care providers (PCPs) in caring for this population.  The role of PCPs appears to be very limited, especially for diagnosis, initial management, and refractory cases.  While PCPs may be able to extend pediatric rheumatologists by managing or co-managing mild cases, their role is likely to be limited by the small number of cases in which they are involved.  Past research has shown, conversely, that internist rheumatologists figure prominently in the care of children with rheumatic diseases ^{4, 6, 21, 22} and may provide nearly one-half of the care to children under age 18 with rheumatic diseases. ²²  There is also evidence that the involvement of internist rheumatologists is largely due to the lack of available pediatric rheumatology providers.  The quality of care provided by internist rheumatologists to children with rheumatic diseases and the extent to which they adequately substitute for pediatric rheumatologists remains unknown and has important implications for the supply of pediatric rheumatologists.  The role of pediatric and internist rheumatologists is discussed in greater detail in Chapter IV.

Open Positions and Salary Concerns

In September 2004, the “Job Openings” page of the American Academy of Pediatrics Rheumatology Section ²⁶ listed 25 advertised positions at 21 institutions or practices.  Of the 23 positions in academic medical centers, one was at the level of division chair; 9 at the level of full, associate, or assistant professor; 7 positions of unspecified rank; and 6 research positions, some of which were also open to non-physician researchers; 2 positions were in private practice.  A few of the listings suggest a level of desperation at the recruiting institutions. 

A listing from Tennessee reads: 

“… We would prefer someone who does research, who would be willing to do some clinical but could have protected time. We are open, however, to considering any Pediatric Rheumatologist who might be interested. They would become the second pediatric rheumatologist.” 

Another from Plano, TX describes an area with particularly constrained access: 

“The nearest, and only, pediatric rheumatologist is in Dallas and serves both the Dallas and Fort-Worth cities. The next nearest pediatric rheumatologists are in Oklahoma City and Houston, both are more than two hours away. As a result, this one rheumatologist in Dallas has over a six-month waiting list. The practice is in a well-established hospital that specializes in high quality medical care, with the largest private practice neonatal intensive care unit in the area. The hospital has a significant pediatric staff (over 40), and all pediatric sub-specialists. A significant demand for pediatric rheumatology exists in this area and at the hospital. The hospital and community are making a very good offer for the qualified applicant.”

While not all advertised positions were at the entry level, it is safe to assume that positions vacated by senior faculty would need to be filled by either another senior pediatric rheumatologist or a newly graduated fellow.  As noted in the previous chapter, only 10 pediatric rheumatology fellows entered their final year of training in 2003.  One expects, therefore, that these 10 fellows were available on the job market in July 2004.  Given that 23 positions were open in September 2004, there appears to be either excess demand for, or a shortage of, pediatric rheumatologists. 

Geographic Distribution of Recently Graduated Pediatric Rheumatologists

The practice location decisions of recent pediatric rheumatology fellowship graduates provide important insights into the persistent tendency for these physicians to locate in certain areas.  Using 2003 diplomate data from the ABP file, physicians Board certified in pediatric rheumatology who graduated from medical school after January 1, 1987 were classified as “recent” graduates (n=55); allowing for 16 years for physicians to complete residency and fellowship and to certify; all others were classified as non-recent graduates.  Graduation date was used to classify diplomates rather than certification date because the first certifying exam in pediatric rheumatology was not offered until 1992.  These data were used to study the practice locations of recent graduates.^[vi] 

Over 80 percent of recent diplomates practice in a county that also has at least one non-recent pediatric rheumatology diplomate (Table 12).  Only eight United States counties currently have a pediatric rheumatologist who recently graduated but no pediatric rheumatologists who graduated prior to 1987:  Johnson, IA (city:  Iowa City); Hampden, MA (city:  Chicopee); Norfolk, MA (City:  Norwood); Livingston, MI (City:  Brighton); Jackson, MO (city:  Kansas City) ^{[vi] i}; Multnomah, OR (City:  Portland); Providence, RI (city:  Providence) and Dane, WI (City: Madison).  These data suggest that the geographic distribution of pediatric rheumatologists may remain unchanged without incentives to practice in underserved areas.

Table 12:  Board-Certified Pediatric Rheumatologists by County and Graduation Cohort, American Board of Pediatrics, 2003

Table 12:  Board-Certified Pediatric Rheumatologists by County and Graduation Cohort American Board of Pediatrics, 2003, cont.

Summary

Given the size of the pediatric population and the number of providers in each State, it appears that a number of States have rheumatologist to patient ratios that exceed typical pediatric rheumatology practice capacity.  It is estimated that 60 percent of States have more than 1,000 children with rheumatic diseases per pediatric rheumatologist.  Assuming that a pediatric population of 250,000 is needed to attract an initial provider and increments of 200,000 are needed to attract additional providers, it is estimated that a minimum of 337 pediatric rheumatologists is needed nationwide.  Thus, there is a national deficit of approximately 135 to 145 providers.  Furthermore, the number of trainees completing fellowship is less than the number of advertised positions.  Thus, there is considerable evidence that the current supply of rheumatologists is not adequate to meet employer demand and results in a distribution of providers that limits access for a substantial segment of the pediatric population.

Practice location analyses of certified pediatric rheumatologists demonstrate, however, that 80 percent of recently trained pediatric rheumatologists (i.e., those who completed medical school in or after 1987) practice in a county that also has an older pediatric rheumatologist.  These results suggest that newly trained rheumatologists are not necessarily entering underserved areas.