The influence of payer mix and visit volume on NRMP match rates for community-based family medicine residency programs.

Background: With declining U.S. medical student interest in family medicine, it is important to study factors that may predict for individual residency program fill rates. Predictors of family medicine residency program match rates are few, with location by region of the country being the best predictor. Using a limited database, this study evaluated whether community-based residency program match rates are associated with patient visit volume and payer status.

Methods: An existing database of characteristics of 42 community-based family medicine residency programs was analyzed for associations of match rate with patient visit volume and with insurance payer status.

Results: Nonsignificant correlations were calculated for match rate with the following factors: patient visit volume per provider, percentage of patient visits with no insurance or with Medicaid payment, percentage of visits with Medicare payment, and percentage of visits with private third party payers. A marginal negative association of match rate with percentage of patient visits with Medicare payment was not significant when adjusted for region of the country.

Conclusions: In this study, neither patient visit volume nor payer status were predictive of match rate for community-based family medicine residency programs. Limitations to this study are acknowledged.

Key Words: family practice, medical residency, health insurance, medically uninsured, office visits

**********

Family medicine is perhaps the specialty with the greatest impact on healthcare provision. Family physicians provide a substantial amount of service to the Medicare population, provide a disproportionate amount of care to the uninsured and those with Medicaid, and provide the most common location for outpatient visits. (1-6) Family physicians "distribute like people," more closely reflecting population distribution than other specialties. (7,8) Without family physicians, a majority of counties become primary care physician shortage areas. (9) Family physicians fill all of these roles while costing less than other providers. (7,10,11)

This importance of family medicine exists in the face of declining interest in the specialty by U.S. medical students. Entrance of senior U.S. medical graduates has fallen from 2340 in 1997 to 1132 in 2006. (12) This pattern has been termed a "death spiral" in the past. (13) Scholars struggle to define the reasons for the difficulty in attracting students to family medicine. (14) Medical specialties, which have simultaneously witnessed a rise in the entrance of U.S. medical graduates, produce physicians who will generate larger personal incomes yet provide a smaller number of patient visits. Given the nature of family medicine to care for those with lower reimbursement rates by payers and provide a high number of patient visits, we asked whether there has been an association between visit volume or payer mix of family medicine residency programs and program attractiveness to students.

The best predictors of program fill rate through the National Residency Matching Program (the "match") have been location, location, and location. Programs near Fortune Magazine's "Best Places to Live," or located in the western and northeastern U.S., fare better than those in the southern and central U.S. (15-17) Other positive predictors include being a program of smaller size, having a relatively low percentage of international medical graduates, a positive institutional climate for the specialty of family medicine, less emphasis on research, presence of state financial support for university programs, and the program director's perception of high resident morale, high quality of residents, and exemplary program reputation. (15-19) Positive predictors to a marginal degree have included the resident call schedule, a suburban location, and a lack of use of marketing materials. (15,17) Some of these predictive factors have shown greater associations with success in periods of lower match rates for the specialty. (15)

Potential predictive factors that have shown no association with match success include the number of faculty, provision of obstetric care by faculty, number of hospital beds, teaching of first year medical students by family medicine faculty, faculty satisfaction, and resident salary and benefits. (15,16,19)

To this point, there have been no analyses of associations of patient visit volume or payer type with match rate success. Using a limited database, we investigated whether variation in match rate among community-based family medicine residency programs was associated with patient visit volume per provider, the percentage of patient visits with no insurance or with Medicaid, the percentage of patient visits with Medicare, and the percentage of patient visits with private third party payment.

Methods

Budget and payer data for community-based family medicine residency programs were used from a study described previously. (20) As reported, a small national sample of community-based family medicine residency programs contributed to a comparative database of financing and budgeting that included patient visit volume delineated by payer type for the year 2004. A simultaneous effort was undertaken by the Association of Departments of Family Medicine for medical school-based programs. Programs were categorized by region of the country and by size using the 1984 National Study of Financing of Family Practice Residency Training Programs design. (21) Regions were defined as West, Plains, Southern, and Northeast. Program size categories were small (four or less residents prescribed per training year), medium (five to eight per training year), and large (more than eight prescribed residents per training year). These community-based programs were classified as medical school-administered, medical school-affiliated, or having no medical school affiliation. Medical school-based programs were not included in this effort.

For the present study, an average percentage match rate of U.S. medical graduates for the years 2003, 2004 and 2005 was calculated for each program that participated in the concurrent financing study. Match rate here was defined as the percentage of offered positions filled by U.S. allopathic and osteopathic senior graduates during the period of the National Resident Matching Program. JMP version 5.1 statistical software (SAS Institute, Inc., Cary, NC) was used to analyze for associations of the average match rates with five items: number of patient visits per resident physician full-time equivalent (FTE), number of patient visits per the sum of resident plus faculty FTE, percentage of patient visits with Medicaid payment or no insurance, percentage of patient visits with Medicare payment, and percentage of visits with private third party payers. Pearson correlation coefficients were calculated as measures of associations. For any item that had a statistically significant association with match rate, that item was entered into a multivariate linear regression model with a second independent variable, "region of the country" in prediction of match rate, to adjust for the known effect of region.

Results

For patient visit volume and payer type, 42 programs provided data. Only 33 programs had faculty FTE data available for calculations (Table).

The chi square statistic test for homogeneity was not statistically significant by region of the country, program size, or type of medical school affiliation. Programs that participated in the financing study had a higher match rate compared with programs that did not participate (63% versus 51%, P < 0.001).

For the present study, region of country and program size predicted 18.5% of the variance in match rate ([r.sup.2] = 0.185) for the years tested. No associations of variation in match rates of community-based family medicine residency programs were found with patient visit volume, percentage of visits with no insurance or with Medicaid payment, or percentage of visits with private third party payment. There was a marginal negative association between match rate and the percentage of visits with Medicare payment ([r.sup.2] = 0.11, P = 0.03) (Table). When adjusted for region of the country, correlation of Medicare visit percentage and match rate was not statistically significant (P = 0.16).

Discussion

The influences of region of country and program size were consistent with those seen by Skinner and Newton for the years 1984 to 1998 in predicting variance in match rate accounted for by those two variables. Those researchers used a larger sample size and longer longitudinal analysis. (15) This study limited analysis of match rates to the years 2003, 2004, and 2005, and compared those to the fiscal year 2004. This was a period of relatively low match rates of U.S. graduates into family medicine programs. We attempted to minimize individual program match rate variability around the fiscal year studied by using the cumulative match rates for the years 2003 to 2005.

Percentage of visits with private third party payers and percentage of visits with no insurance or with Medicaid payment showed no associations with match rate variance. There was, however, a slight association between match rate variance and percentage of patient visits with Medicare payment. When adjusting for region of country, no association was observed between match rate and percentage of visits with Medicare payment. It appears that U.S. medical graduates are not avoiding individual family medicine programs with high visit volumes or with payer mixes marked by indigent populations. The tendency toward a negative association of match rate and percentage of visits with Medicare may hint at student disinclination toward programs with higher rates of geriatric patient care. More likely, this reflects the lower match rates and higher volume of Medicare patients present in the South and Plains.

There are limitations to this study and its generalizations. This study is a small one, and its insights may be limited by the small sample size drawn from community-based programs. Also, participating programs had a significantly higher match rate than did nonparticipants. This leaves open the questions of differences in organization and financing patterns between participating and nonparticipating programs and the generalization of study findings. Finally, medical-school based programs were not studied.

Skinner and Newton have described the annual match as the "public report card for residency directors." (15) Evidence would support such a statement. Out-of-match program entrants have higher rates of leaving programs, lower board scores, and more likelihood of requirements for remediation. (22) Identifying factors that do and those that do not influence match rates is part of an intelligent strategy to ensure the viability of training programs.

It appears that geography remains the best predictor of match rate for family medicine residency programs, with the West consistently faring better. Differences in payer mix of patient populations and differences in patient visit volume do not appear to be large driving forces for program choice by U.S. senior medical graduates.

References

1. Stoever J. Third annual increase: Match brings slight uptick in applicants entering specialty. AAFP News Now 2006 3/17/06.

2. Centers for Medicare and Medicaid Services. Medicare Part B physician/supplier national data--calendar year 2004. Expenditures and services by specialty. http://www.cms.hhs.gov/MedicareFeeForSvcPartsAB/Downloads/Specialty04.pdf. Accessed March 2, 2007.

3. Cohen D, Guirguis-Blake J, Jack B, et al. Family physicians are an important source of newborn care: the case of the State of Maine. Am Fam Physician 2003;68:593.

4. Cohen D, Guirguis-Blake J, Jack B, et al. Family physicians make a substantial contribution to maternity care: the case of the State of Maine. Am Fam Physician 2003;68:405.

5. Fryer GE, Dovey SM, Green LA. The importance of having a usual source of health care. Am Fam Physician 2000;62:477.

6. Biola H, Green LA, Phillips RL, et al. The US primary care physician workforce: undervalued service. Am Fam Physician 2003;68:1483.

7. South East Family Practice Forum. The value of family physicians. The health and wealth of communities. August 15, 2003.

8. Brooks RG, Walsh M, Mardon RE. The roles of nature and nurture in the recruitment and retention of primary care physicians in rural areas: a review of the literature. Acad Med 2002;77:790-798.

9. Fryer GE, Green LA, Dovey SM. The United States relies on family physicians, unlike any other specialty. Am Fam Physician 2001;63:1669.

10. Ciracy EW, Liang FZ, Godes JR. The cost and funding of family practice graduate education in the United States. J Fam Pract 1985;20:285-295.

11. Colwill JM. Financing graduate medical education. Acad Med 1989;64:154-158.

12. The National Resident Matching Program. Programs, positions ranked and filled in 2006by U.S. seniors and all applicants. http://www.nrmp.org/res_match/data_tables.html. Accessed March 2, 2007.

13. Goldsmith GA. Interest in family medicine: 1982 Revisited? Fam Med 2004;36:447-448.

14. Dorsey ER, Jarjoura D, Rutecki G. The influence of controllable lifestyle and sex on the specialty choices of graduating US medical students, 1996-2003. Acad Med 2005;80:791-796.

15. Skinner BD, Newton WP. A long-term perspective on family practice residency match success: 1984-1998. Fam Med 1999;31:559-565.

16. Malaty W, Pathman DE. Factors affecting the match rate of rural training tracks in family practice. Fam Med 2002;34:258-261.

17. Kikano GE, Galazka SS, Flocke SA, et al. Markers of successful recruitment of students of family practice residency programs. Fam Med 1994;26:492-496.

18. Whitcomb RE, Miller RS. Comparison of IMG-dependent and non-IMG-dependent residencies in the National Resident Matching Program. JAMA 1996;276:700-703.

19. Kutob RM, Senf JH, Campos-Outcalt D. Declining interest in family medicine: perspectives of department heads and faculty. Fam Med 2003;35:504-509.

20. Carlisle RT. Financing and budgeting of community-based family medicine residency programs. South Med J 2006;99:576-578.

21. Ramsey CN, Durrett JP, Grant WD, et al. A national study of financing of family practice residency training programs. Oklahoma City, University of Oklahoma Press, 1984.

22. Blonski J, Rahm S. The relationship of residency performance to match status and US versus international graduate status. Fam Med 2003;35:100-104.

Robert Carlisle, MD, MPH, and C. Ken Shannon, MD, PhD

From the Department of Family Medicine, West Virginia University, Morgantown, West Virginia.

Reprint requests to Dr. Robert Carlisle, WVU Department of Family Medicine, PO Box 9152, Morgantown, WV 26506-9152. Email: [e-mail]

Accepted October 17, 2006.

The authors have no proprietary interest in the topics discussed.

Institutional Review Board approval was received from the University of Illinois-Chicago.

RELATED ARTICLE: Key Points

* Family physicians provide a substantial amount of interpersonal, nonprocedural care to those with Medicare, Medicaid, and no insurance.

* U.S. medical graduates have drifted toward procedural-driven, higher paying specialties with less patient-visit volume.

* In this study, region of the country predicted 18.5% of the variation in match rate, a figure consistent with prior research.

* In this study, match rate was not associated with propensity of programs to care for those with no insurance or Medicaid payment, or with propensity to provide a higher volume of patient visits.

* A small negative association of match rate with propensity to care for those with Medicare payment was not observed when controlling for region of country.

* U.S. graduates do not appear to be preferentially selecting family medicine residency programs with higher compensating payer mixes or lower patient visit volumes, based on this small study.

Table. Associations of community-based family medicine program match
rates with patient visit volume and payer characteristics

                                       Correlation
                        Participating  coefficient
                        programs       (r)          [r.sup.2]  p-value

Percentage of patient   42              0.065        0.004     0.68
  visits with Medicaid
  or no insurance
Percentage of visits    42             -0.331        0.110     0.03
  with Medicare
Percentage of visits    42             -0.159        0.025     0.31
  with private third
  party payment
Annual patient visits   42              0.002       <0.01      0.76
  per resident FTE
Annual patient visits   33              0.000       <0.01      0.91
  per sum of resident
  and faculty FTE

FTE = full-time equivalent.
Not adjusted for region of country.