Effects of physician supply on melanoma incidence and mortality in Florida. (Original Article).

Background: Increasing supplies of dermatologists and family physicians have been associated with earlier detection of malignant melanoma. We investigated whether physician supply was similarly related to incidence and mortality rates of malignant melanoma.

Methods: Using the state tumor registry, we determined melanoma incidence and mortality rates for the years 1993 to 1995 for each Florida county. We measured physician supply for each Florida county using data from the 1994 American Medical Association Physician Masterfile. Multiple linear regression analysis was used to determine relationships between physician supply and melanoma incidence and mortality rates, controlling for other county-level characteristics.

Results: Among male patients, an increasing supply of family physicians was associated with higher melanoma incidence and lower melanoma mortality. Increasing supplies of dermatologists were associated with lower overall melanoma mortality rates, and increasing supplies of general internists were associated with higher overall melanoma mortality.

Conclusion: We found that melanoma incidence and mortality rates varied substantially among Florida's 67 counties, and that differences in physician supply explained some of this variability. Further study is needed to confirm these findings and to elucidate possible mechanisms that would account for these associations.

Key Words: health care personnel, health policy, melanoma, physicians

**********

During the past 3 decades, the incidence of malignant melanoma increased by 85%, more than any other major cancer. (1, 2) There were more than 47,700 new cases of melanoma in 2000, with 7,700 deaths. (3) The lifetime risk of acquiring melanoma has reached 1 in 74. (4)

The prognosis of malignant melanoma depends to a great extent on the stage at which it is diagnosed. (5) Early thin lesions are almost entirely curable. (6) Five-year survival for melanomas that have spread to regional lymph nodes is 54%; for metastatic disease, it is only 13%. (1)

Physicians have many opportunities to detect melanoma. Eighty-seven percent of patients diagnosed with melanoma report having a regular physician, and 63% report having seen their physician in the previous year. (7) Patients who see physicians frequently are more likely to have their melanoma discovered by their physician, and physician-detected melanomas are often thinner lesions. (8,9)

Given their important role in early detection, the availability of physicians may be an important factor affecting the public health impact of malignant melanoma. One measure of this availability is the number of physicians practicing near a patient's residence. We discovered in a previous study, for example, that an increased supply of dermatologists and family physicians in a person's county of residence was associated with earlier detection of malignant melanoma. (10) In contrast, increased proportions of general internists led to decreased odds of early detection of melanoma. It is not known, however, whether physician supply similarly impacts incidence and mortality rates for malignant melanoma.

We therefore investigated whether physician supply was related to incidence and mortality rates of malignant melanoma. Incidence and mortality were examined for the years 1993 to 1995, the most current years for which complete data were available, in the 67 counties of Florida, a state with a high incidence of melanoma. (11) We hypothesized that counties having greater supplies of family physicians and dermatologist specialists would have higher incidence rates of melanoma (because of more frequent detection) but decreased mortality (because of earlier detection).

Methods

Data Collection

We obtained data on malignant melanoma incidence and mortality rates from the Florida Cancer Data System (FCDS), Florida's population-based tumor registry. The FCDS was created in 1978 and is a member of the North American Association of Central Cancer Registries. The FCDS has well-established methods to ensure complete case finding including cooperative arrangements with other state tumor registries. North American Association of Central Cancer Registries audits have estimated the completeness of overall cancer case ascertainment for the time period 1990 to 1994 to be 99%. By state law, all cases of malignant melanoma are reportable to the FCDS. For each county, we recorded melanoma incidence and mortality rates for all persons, and for men and women separately. To account for year-to-year variations in rates, we averaged incidence and mortality rates for each county over the 3-year period 1993 to 1995. All incidence and mortality rates are age adjusted.

Data on physician supply was obtained from the 1994 American Medical Association (AMA) Physician Masterfile, which includes allopathic and osteopathic physicians regardless of AMA membership. (12) Population estimates were obtained from the 1990 U.S. Census. Physician supply was measured at the county level with variables created for the supply of family physicians, general practitioners, general internists, dermatologists, and all remaining physicians.

The AMA Physician Masterfile allows physicians to designate both a primary and secondary specialty. Primary specialty is self-designated by physicians as the specialty in which they spend the majority of their clinical time. Physicians were classified as family practice, general practice, or general internal medicine if this was indicated as their primary specialty, regardless of their secondary specialty designation. (13) Primary care practice content has been verified for physicians meeting this definition. (14) In contrast, physicians who indicate a primary care field only as their secondary specialty have been found to have markedly less primary care practice content. (14)

Physician supply for this study was based on their providing direct patient care and measured as full-time equivalents. Physicians who indicated they were engaged in fulltime direct patient care were counted as one full-time equivalent (FTE); those who indicated in the Masterfile that they were either "semi-retired," in residency training, or also engaged in teaching or research were counted as 0.5 FTE. Physicians who indicated they were no longer involved in direct patient care were excluded. Previous studies have validated data contained in the 1994 AMA Physician Masterfile. (12,15,16)

We obtained data from the 1990 U.S. Census to account for other characteristics of counties that might affect melanoma incidence and mortality. Previous studies have suggested that melanoma incidence and prognosis are related to age, (17,18) sex (19-21) marital status, (22) and socioeconomic status. (19,23,24) Using census data, we determined for each Florida county the following characteristics: the percentage of the population that was white versus nonwhite, the percentage of the population having less than a high school education, the percentage of men who were married versus unmarried, the percentage of women who were married versus unmarried, the median household income, and the percentage of persons living in an urban versus nonurban setting. We also ascertained whether each county was coastal or inland.

Data Analysis

Counties were the unit of analysis for this study. We used multiple linear regression analysis to determine relationships between physician supply and melanoma incidence and mortality rates, controlling for other county level characteristics. The following variables were included in all initial models: the percentage of persons who were white versus nonwhite, the percentage living in urban versus nonurban settings, the percentages of both men and women who were married, the percentage of persons having less than a high school education, the median household income for the county, and whether the county was coastal versus inland. We also added measures of physician supply (full-time equivalents) to all models including family/general practitioners, general internists, obstetrician/gynecologists, dermatologists, and all other nondermatology specialists. Variables remained in final regression models if they maintained statistical significance at the 0.05 levels using a backward-variable selection algorithm. Sepa rate regression models were performed for age-adjusted incidence rates and mortality rates. We also stratified analyses by gender. We confirmed normal distributions for outcomes using graphical methods.

Results

Table 1 reports the melanoma incidence and mortality rates among Florida's 67 counties. Incidence rates for melanoma varied widely by county, with some counties having no cases of melanoma in the 3-year period and others having rates as high as 16.7 cases per 100,000 population. Melanoma mortality rates similarly varied widely among counties, with some counties having no deaths from melanoma during the study period and others having melanoma death rates as high as 6.5 deaths per 100,000. The average numbers of physicians for Florida's 67 counties are reported in Table 2. There was again substantial variation among counties in the numbers of physicians.

Linear regressions were performed to determine which county level characteristics, as described above, were associated with melanoma incidence and mortality rates. Only one variable was associated with melanoma incidence rates by county: the percentage of the county population that was white (P = 0.0003). The melanoma incidence rate increased by 1.5 cases per 100,000 for each 10% increase in the proportion of the population that was white. This variable accounted for 17% of the variation in melanoma incidence rates among Florida counties.

Two variables were predictive of the melanoma incidence rate among males: the percentage of the population that was white and the supply of family/general practitioners. For each 10% increase in the proportion of the population that was white, the incidence rate increased by 2.0 cases per 100,000 (P = 0.001). For every 10 additional family physicians per 100,000 population, there was a corresponding increase in melanoma incidence of 1 case per 100,000 (P = 0.02). These two variables explained 18% of the variation in male incident rates among counties.

There was only one significant predictor of melanoma incidence rates among women: the percentage of the population that was white. For each 10% increase in the proportion of the population that was white, the incidence rate of melanoma increased by 1.3 cases per 100,000 (P = 0.004). This variable explained 12% of the variation in female incidence rates among counties.

Overall mortality rates attributable to melanoma were associated with two variables: the supply of general internists and the supply of dermatologists. Each additional dermatologist per 100,000 population was associated with a drop in melanoma mortality of 0.19 case per 100,000 (P = 0.02). An increasing supply of general internists, however, was associated with higher melanoma mortality, with a 0.04 case per 100,000 increase in the mortality rate for each additional general internist per 100,000 population (P = 0.02). These two variables explained 11% of the variation in mortality rates among counties.

Among men, three variables were found to be associated with melanoma mortality rates. Each 10% increase in the proportion of women within the county who were married was associated with a reduction of male melanoma mortality of 1.3 cases per 100,000 (P = 0.01). Each 10% increase in the proportion of men within the county who were married was associated with an increase in melanoma mortality of 0.80 case per 100,000 (P = 0.02). Increasing supplies of family/general practitioners were associated with decreased melanoma mortality rates. Each additional family physician per 100,000 population was associated with a reduction in melanoma mortality of 0.04 case per 100,000 (P = 0.01). These three variables explained 17% of the variation in male melanoma mortality rates among counties.

Finally, three variables were associated with melanoma mortality rates among women. Increasing supplies of family/general practitioners were associated with higher female melanoma mortality rates (0.05 case per 100,000 for each additional physician per 100,000 population, P = .0001) as were increasing supplies of general internists (0.06 case per 100,000 for each additional physician per 100,000 population; P = 0.03). Increasing supplies of obstetrician/gynecologists, however, were associated with lower melanoma mortality (0.14 case per 100,000 for each additional physician per 100,000 population; P = 0.006). These three variables explained 23% of the variation in female melanoma mortality rates among counties.

Discussion

Melanoma incidence and mortality rates varied substantially among Florida's 67 counties during the study period. We found that the supply of physicians within each county accounted for some of the variation in melanoma incidence and mortality. More specifically, counties having a greater supply of family physicians had greater melanoma incidence rates among men. Counties having a greater supply of dermatologists were found to have lower melanoma mortality rates. Among men, lower mortality rates were observed among counties with greater supplies of family physicians, whereas among women, lower mortality rates were observed in counties having greater supplies of obstetrician/gynecologists.

There are several mechanisms by which physicians could influence melanoma incidence rates. Some physician services might lead to a reduction in incidence rates, such as providing advice on sun avoidance, use of protective sunscreens, and self-examinations. Patients practicing self-examination, for example, have been found to have one-third lower odds of developing melanoma. (25) Physicians could also reduce the incidence of melanoma by the detection and excision of premalignant dysplastic lesions (clinically atypical moles). In contrast, some physician services (such as greater detection of occult melanomas) could lead to higher incidence rates. Greater physician detection of occult lesions has been one explanation put forth to explain the rapid increase in melanoma incidence rates that has occurred in the past several decades. (26) It is not known how often physicians provide these services to patients, nor is it known how the provision of melanoma-related health services varies by physician specialty.

In addition to lowering incidence, physician-provided services could also measurably impact melanoma mortality rates by the mechanisms discussed above. Furthermore, timely detection of early thin lesions by physicians could significantly improve mortality rates for persons diagnosed with melanoma. Physician-detected melanomas are often thinner lesions than those discovered by patients themselves. (9) In support of this premise, we have previously reported that patients residing in areas having greater supplies of dermatologists and family physicians were more likely diagnosed with melanoma at an early stage. (10)

In some cases, we found that increasing measures of physician supply were associated with higher melanoma mortality rates. Although it is easy to envision how an adequate supply of physicians might contribute to better melanoma outcomes, it is less obvious why an increasing supply of physicians would be detrimental. One possibility is that physician specialties differ in their provision of melanoma-related health services, and that they at times compete to provide for patients' health care needs. An abundant supply of physicians who do not address skin problems, for example, may result in decreased patient exposure to early detection services that would otherwise be supplied by physicians that do incorporate these services in their care. We have found supporting evidence for this in earlier studies showing that earlier detection of colorectal cancer was found in counties having higher supplies of primary care doctors but lower supplies of non-primary care specialists. (27)

We found higher melanoma mortality rates among counties having greater supplies of general internists. This finding is consistent with a previous study, which found that patients residing in areas with greater supplies of general internists were diagnosed with melanoma at later stages. This raises the question of whether general internists are less likely than other medical specialties to provide melanoma-related health services to their patients. Training programs in general internal medicine may provide less extensive training in dermatology and dermatologic procedures than the other primary care specialties. (28-30) Among women, we found that increasing supplies of obstetrician/gynecologists were associated with lower melanoma mortality rates, whereas the reverse was seen for the supplies of family physicians and general internists. This again suggests there may be differences among the primary care specialties in their provision of melanoma-related health services or in their willingness to refer patients more promptly if an abnormal skin lesion is noted.

The study has a number of important limitations. A central tenet of our hypothesis is that the availability of physicians influences patients' use of their services. As an example, we hypothesize that persons would be more likely to use the services of a dermatologist if they are in abundant supply and would be less likely to receive their services if their supply is limited. Furthermore, it seems reasonable that there would only be an abundance of dermatologists if the patient population sufficiently used their services. We did not have specific data, however, on patients' actual use of health services, and can only make inferences on the basis of group level data. Although physician supply is an important variable that is relevant to policy in its own right, it can be considered only an aggregate measure of patients' use of physician services. In future research, it would be desirable to measure actual use of physician services at the individual patient level to confirm these relationships. In addition, we were limited in the number of factors that could be controlled in the analysis, so it is possible that the observed relationships were the result of some unmeasured confounding factor. Finally, our study was restricted to cases of malignant melanoma in Florida, which may not be representative of other parts of the country.

In conclusion, we found that melanoma incidence and mortality rates varied substantially among Florida's 67 counties, and that differences in physician supply explained some of this variability. Further study is needed to confirm these findings and to elucidate possible mechanisms that would account for these associations. Our results suggest that the composition of the physician work force may affect important health outcomes and is worthy of further study.

Table 1

Average melanoma incidence and mortality rates for Florida counties,
1993-1995 (n = 67)

           Average
            rate (a)  Range    SD

Incidence
  Female    6.72      0-25    6.72
  Male     11.2       0-20.8  4.99
  Total     8.76      0-16.7  3.41
Mortality
  Female    1.51      0-8.6   1.74
  Male      3.50      0-9.63  1.98
  Total     2.42      0-6.47  1.28

(a) Rates are expressed as the number of cases or deaths per 100,000
population.

Table 2

Average physician supply for Florida counties, 1994 (n = 67)

                                Physicians
                                per 100,000
                                population   Range     SD

Primary care
  Family/general practitioners     28.1      0-110.7  14.6
  General internists               13.8      0-47.4   10.7
  Obstetrician/gynecologists        7.1      0-25.1    5.8
Nonprimary care
  Dermatologists                    1.9      0-9.4     2.3
  Other specialists                83.3      0-123.9  69.7

Accepted August 28, 2002.

References

(1.) Miller B, Rics L, Hankey B, Kosary C, Edwards B. Cancer Statistics Review: 1973-1989 (NIH Pub. No. 92-278). Bethesda, MD, U.S. Department of Health and Human Services, 1992.

(2.) Friedman RJ, Rigel DS, Silverman MK, Kopf AW, Vossaert KA. Malignant melanoma in the 1990s: The continued importance of early detection anti the role of physician examination and self-examination of the skin. CA Cancer J Clin 1991;41:201-226.

(3.) Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer statistics, 2000. CA Cancer J Clin 2000;50:7-33.

(4.) Rigel DS, Carucci JA. Malignant melanoma: Prevention, early detection, and treatment in the 21st century. CA Cancer J Clin 2000;50:21 5-236.

(5.) Ahmed I. Malignant melanoma: Prognostic indicators. Ma o Clin Proc 1997;72:356-361.

(6.) Girgis A, Campbell EM, Redman 5, Sanson-Fisher RW. Screening for melanoma: A community survey of prevalence and predictors. Med J Aust 1991;154:338-343.

(7.) Geller AC, Koh HK, Miller DR, Clapp RW, Mercer MB, Lew RA. Use of health services before the diagnosis of melanoma: implications for early detection and screening. J Gen Intern Med 1992;7:154-157.

(8.) Koh HK, Miller DR, Gelter AC, Clapp RW, Mercer MB, Lew RA. Who discovers melanoma? Pattems from a population-based survey. J Am Acad Dermatol 1992;26:914-919.

(9.) Epstein DS, Lange JR, Gruber SB, Mofid M, Koch SE. Is physician detection associated with thinner melanomas? JAMA 1999;281:640-643.

(10.) Roetzheim RG, Pal N, Van Durme DJ, Wathington D, Ferrante JM, Gonzalez EC, et al. Increasing supplies of dermatologists and family physicians are associated with earlier stage of melanoma detection. J Am Acad Dermatol 2000;43:211 1-218.

(11.) Kamath D, Cantor AB, Glass F, Fenske N, Cruse CW, Wells K, et al. Florida's undeclared epidemic: Malignant melanoma. JFIa Med Assoc 1997;84:161-165.

(12.) Kenward K. The scope of the data available in the AMA's Physician Masterfile, Am J Public Health 1996;86: 1481-1482 (letter).

(13.) Kindig D. Counting generalist physicians. JAMA 1994;271:1505-1507.

(14.) Shea JA, Kletke PR, Wozniak GD, Poisky D, Escarce JJ. Self-reported physician specialties and the primary care content of medical practice: A study of the AMA Physician Masterfile-American Medical Association. Med Care 1999;37:333-338.

(15.) Grumbach K, Becker SH, Osbom EH, Bindman AB. The challenge of defining and counting generalist physicians: An analysis of Physician Masterfile data. Am J Public Health 1995;85: 1402-1 407.

(16.) Williams P, Whitcomb M, Kessler J. Quality of the family physician component of the AMA Masterfile. J Am Board Fam Pract 1996;9:94-99.

(17.) Austin PF, Cruse CW, Lyman G, Schroer K, Glass F, Reintgen DS. Age as a prognostic factor in the malignant melanoma population. Ann Surg Oncol 1994;1:487-494.

(18.) Goodwin JS, Samet JM, Key CR, Humble C, Kutvirt D, Hunt C. Stage at diagnosis of cancer varies with the age of the patient. J Am Geriatr Soc 1986;34:20-26.

(19.) Chang AE, Kamell LH, Menck HR. The National Cancer Data Base report on cutaneous and noncutaneous melanoma: A summary of 84,836 cases from the past decade--The American College of Surgeons Commission on Cancer and the American Cancer Society. Cancer 1998;83:1664-1678.

(20.) Cress RD, Holly EA. Incidence of cutaneous melanoma among nonHispanic whites, Hispanics, Asians, and blacks: An analysis of California cancer registry data, 1988-1993. Cancer Causes Control 1997;8:246-252.

(21.) Hersey P, Sillar RW, Howe CG, Burton RC, Darbar SV, Foster HM, et al. Factors related to the presentation of patients with thick primary melanomas. Med J Aust 1991;154:583-587.

(22.) Roetzheim RG, Pal N, Tennant C, Voti L, Ayanian JZ, Schwabe A, et al. Effects of health insurance and race on early detection of cancer. JNatl Cancer Inst 1999;91:1409-1415.

(23.) Vagero D, Persson G. Risks, survival and trends of malignant melanoma among white and blue collar workers in Sweden. Soc Sci Med 1984;19:475-478.

(24.) Geller AC, Miller DR, Lew RA, Clapp RW, Wenneker MB, Koh HK. Cutaneous melanoma mortality among the sociocconomically disadvantaged in Massachusetts. Am JPublic Health 1996;86:538-544.

(25.) Berwick M, Begg CB, Fine JA, Roush GC, Barnhill RL. Screening for cutaneous melanoma by skin self-examination. JNatl Cancer Inst 1996;88:17-23.

(26.) Swerlick RA, Chen S. The melanoma epidemic: Is increased surveillance the solution or the problem? Arch Dermatol 1996;132:881-884.

(27.) Roetzheim RG, Pal N, Gonzalez EC, Ferrante JM, Van Durme DJ, Ayanian JZ, et al. The effects of physician supply on the early detection of colorectal cancer. JFam Pract 1999;48:850-858.

(28.) Jones TP, Boiko PE, Piepkorn MW. Skin biopsy indications in primary care practice: A population-based study. JAm Board Fain Pract 1996; 9:397-404.

(29.) Boiko PE, Koepsell TD, Larson EB, Wagner EH. Skin cancer diagnosis in a pnmary care setting. J Am AcadDermatol 1996;34:608-611.

(30.) Rivo ML, Saultz JW, Wartman SA, DeWitt TG. Defining the generalist physician's training. JAMA 1994;271:1499-1504.

RELATED ARTICLE: Key Points

* The incidence and mortality rate of malignant melanoma varies considerably in different Florida counties.

* The specialty and supply of physicians providing direct patient care may explain some of the variability in malignant melanoma incidence and mortality.

* Further study is needed in the area of physician supply and disease incidence and mortality.

From the Department of Family Medicine, University of South Florida, and the Division of Cancer Control, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL.

Dr. Roetzheim was supported by a Robert Wood Johnson Foundation Generalist Physician Faculty Scholars Award.

Reprint requests to Daniel J. Van Durme, MD, Department of Family Medicine, University of South Florida, 12901 Bruce B. Downs Blvd., MDC 13, Tampa, FL 33612. Email: [e-mail]

Copyright [C]2003 by The Southern Medical Association

0038-4348/03/9607-0656