Objective To determine whether there is an association between dermatologist density and melanoma mortality.
Design A regression model was developed to test the association between melanoma mortality and dermatologist density, controlling for county demographics, health care infrastructure, and socioeconomic factors. Data were collected from the Area Resource File, US Centers for Disease Control and Prevention, and National Cancer Institute's Surveillance, Epidemiology, and End Results program and National Program for Cancer Registries.
Setting US counties.
Patients Melanoma mortality and incidence data were reported as age-adjusted mean rates per 100 000 people from January 2002, through December 2006.
Main Outcome Measure The primary outcome measure was melanoma mortality rate per 100 000 people at the county level.
Results Geographic variation exists in the distribution of dermatologists across the United States. Multivariate analysis demonstrated that the presence of 0.001 to 1 dermatologist per 100 000 people was associated with a 35.0% reduction in the melanoma mortality rate (95% CI, 13.4%-56.6%) when compared with counties with no dermatologist. The presence of 1.001 to 2 dermatologists per 100 000 people was associated with a 53.0% reduction in the melanoma mortality rate (95% CI, 30.6%-75.4%). Having more than 2 dermatologists per 100 000 people did not further lower melanoma mortality rates.
Conclusion Within a given county, a greater dermatologist density is associated with lower melanoma mortality rates compared with counties that lacked a dermatologist.
Physician density influences health care outcomes.1-3 Some investigators have suggested that having more specialists may not be as valuable as increasing the numbers of primary care physicians.4,5 Yet, specialists are associated with improved outcomes in dermatology, urology, neonatology, colorectal surgery, and orthopedic surgery.1,6-10
Each year younger dermatologists replace aging practitioners, which decreases the availability of medical dermatologic care because younger physicians are more prone to engage in education, research, surgery, or cosmetics.11 An unmet demand for dermatologic care in the United States has led to greater use of physician assistants, nurse practitioners, and nonphysician medical professionals in some practices.11-13 Dermatologist and subspecialist (eg, pediatric dermatology, dermatopathology, and Mohs surgery) density varies from city to city, and such physicians are absent in certain areas.14 An understanding of the regional distribution of dermatologists is necessary to develop health care policies that will positively affect patient outcomes.
Melanoma mortality and/or extent of disease may be predicted by factors such as physician specialty density, neighborhood racial heterogeneity, and median household income.1,15-18 We sought to determine whether there is an association between melanoma mortality and dermatologist density at the county level in the United States.
This study received an exemption for approval from the Case Western Reserve University Institutional Review Board.
Melanoma mortality and incidence data were obtained from a merged data set from the National Program for Cancer Registries; National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER); and the US Centers for Disease Control and Prevention's National Vital Statistics System.19 Using this merged data set, mortality and incidence data were reported as age-adjusted mean rates per 100 000 people from January 2002, through December 2006. Incidences and mortality rates were assigned to counties based on each patient's residence at the time of diagnosis and death. A total of 669 rural counties were excluded from the analysis, given insufficient data, because mortality data were reported for less than 2.5% of these counties and less than 5.0% of these counties had any dermatologists.
Dermatologist density, geographic classification, and population data were provided by the 2008 Area Resource File, published by the Health Resource and Services Administration of the US Department of Health and Human Services.20 Dermatologist densities were calculated as 5-year means for the years 2002 to 2006. Dermatologist density was categorized as 0, 0.001 to 1, 1.001 to 2, 2.001 to 4, or more than 4 dermatologists per 100 000 people.
County distribution was determined by the 2000 US Census, which divided the United States into 3141 counties.20 Primary care physicians were defined as those trained in the fields of pediatrics, family medicine, and general internal medicine. The percentage of the population with health insurance was obtained from the US Census Small Area Health Insurance Estimates 2000.21 Access to nondermatologic oncology care was estimated by the number of hospitals with oncology services in 2006. Metropolitan, nonmetropolitan, or rural status of each county was defined using the Department of Agriculture Rural/Urban Continuum Codes, which consider population size, degree of urbanization, and adjacency to a metropolitan area.22
Univariate associations between the predictor variables and melanoma mortality were tested using the t test for categorical predictors and simple linear regressions for continuous predictors. The predictor variables were dermatologist density, melanoma incidence, metropolitan county status, health professional shortage area classification, primary care physician density, percentage of the population older than 65 years, percentage of the population older than 25 years with a high school diploma or equivalent, number of hospitals with oncology departments, percentage of non-Hispanic white population, percentage of health-insured population, median household income, and unemployment rate. The outcome variables were normally distributed.
Multivariate regression models were built using backward stepwise selection with a univariate P < .15 for inclusion into the models. Variance inflation factors were used to assess for interactions among variables. Statistical significance for the final model was determined at P < .05. The reference group in all analyses was derived from the age-adjusted means of counties with 0 dermatologists and nonmetropolitian county status but not classified as an area of health professional shortage. Statistical analysis was performed using Stata statistical software, version 9.2 (StataCorp). Dermatologist density categories were mapped using the mapping software ArcGIS, version 9.2 (Environmental Systems Research Institute Inc).
We included 2472 counties in our analysis because rural counties were excluded. Within the counties in our analysis, the total population was 225 981 679. The percentage of males was 48.9% and the percentage of females was 51.1%. A total of 72.8% were white, 13.2% were black or African American, 4.3% were Asian, and 0.6% were American Indian and Alaskan Native. The median age was 34.9 years, and the percentage of the population older than 65 years was 11.88%.
Predictors of melanoma mortality are listed in the Table. Greater dermatologist density at the c ounty level is associated with lower melanoma mortality rates (Table). The presence of 0.001 to 1 dermatologist per 100 000 people is associated with a 35.0% reduction in mortality (95% CI, 13.4%-56.6%). More than 1 to 2 dermatologists per 100 000 people is associated with a 53.0% reduction in melanoma mortality (95% CI, 30.6%-75.4%). The addition of more than 2 dermatologists per 100 000 people does not further decrease mortality rates.
In the reference group, the absolute melanoma mortality rate was 1.45 (95% CI, 0.71-2.18; P < .001) based on a prevalence of 19 cases per 100 000 people. The presence of 0.001 to 1 dermatologist per 100 000 people is associated with a −0.51 absolute reduction in mortality (P = .002), 1.001 to 2 dermatologists with a −0.77 absolute reduction in melanoma mortality (P < .001), 2.001 to 4.0 dermatologists with a −0.56 absolute reduction in melanoma mortality (P < .001), and more than 4 dermatologists with a −0.66 absolute reduction in melanoma mortality (P < .001).
A slightly lower melanoma mortality rate is associated with the presence of hospitals with oncology departments (1.9%; 95% CI, 0.6%-3.1%). Metropolitan status of a county is associated with a 30.3% (95% CI, 17.3%-43.3%) reduction in the mortality rate.
Melanoma mortality rates are greater in counties with a higher incidence of melanoma (2.3%; 95% CI, 1.6%-3.1%), greater non-Hispanic white population (1.5%; 95% CI, 1.1%-1.9%), and greater health-insured populations (1.5%; 95% CI, 0.2%-2.8%).
Primary care physician density, percentage of population older than 65 years, health professional shortage area classification, percentage of population older than 25 years with a high school diploma or equivalent, median household income, and unemployment rate were parameters not associated with melanoma mortality rate.
There are 0 dermatologists per 100 000 people in a large portion of the country (Figure). Greater densities of dermatologists were noted along the Pacific Coast, throughout the state of Hawaii, and in clusters along the East Coast and the Midwest. Significantly fewer dermatologists are in the central United States relative to the rest of the country.
In 2000, Roetzheim et al23 found that people with limited access to care were more likely to receive a late-stage melanoma diagnosis. More recent studies24,25 have illustrated that diagnosis by a dermatologist is associated with an earlier stage of melanoma compared with detection by a nondermatologist. Our analysis demonstrated that the presence of 0.001 to 1 dermatologist per 100 000 people is associated with a lower melanoma mortality rate by more than one-third compared with counties that lack dermatologists. We believe that this reflects that access to a dermatologist allows for an accurate and earlier diagnosis of melanoma, more appropriate therapy, or a combination of these factors. Our analysis makes the assumption that patients receive relevant health services, such as screening skin examinations, melanoma preventive education, and treatment in the county where the diagnosis was made.
The lowest melanoma mortality rate was observed in counties with 1.001 to 2 dermatologists per 100 000 people. However, increasing dermatologist density beyond this level did not further improve mortality rates, suggesting a plateau effect with increasing physician supply. Areas with a dermatologist density higher than 2 may represent academic centers, where dermatologists have educational and research responsibilities in addition to clinical practice. Alternatively, this plateau may not reflect access to physicians but instead may represent the limitations of our current therapeutics. Similar plateau effects have been described when relating physician density and health outcomes in the fields of urology and neonatology.6,7
No statistically significant differences in melanoma mortality rates were observed when comparing a dermatologist density of 1.001 to 2 dermatologists to counties with higher densities. Our analysis suggests that increasing the number of dermatologists in counties with existing physicians is unlikely to significantly alter melanoma mortality because having 2 dermatologists is not statistically significant from having 4 dermatologists when considering mortality rates. The most notable finding in our analysis demonstrates that having a single dermatologist compared with a county with 0 dermatologists reduces the mortality rate by nearly 35%. Thus, the placement of dermatologists in counties currently lacking physicians may be the most optimal approach to combating the increasing incidence of melanoma in the United States.26 Developing incentives to attract dermatologists to underserved counties may allow us to meet this need. For example, dermatology residency positions or student loan repayments could be tied to a contract of service in an area that lacked physicians.
Other factors associated with reduction in melanoma mortality rates are the presence of hospitals with oncology departments and metropolitan county status. Metropolitan county status is associated with a 30.27% reduction in melanoma mortality that cannot be accounted for by socioeconomic factors, including difference in educational level, age, income, or employment. In other reports,27,28 socioeconomic status has been associated with stage of melanoma at diagnosis; our analysis demonstrates that these factors are not predictive of mortality.
Several factors were associated with higher melanoma mortality rates. Counties with a greater incidence of melanoma and a greater percentage of whites had increased melanoma mortality rates. It is unclear why melanoma mortality rates were higher in counties with a greater percentage of health-insured individuals because other reports2,29,30 have demonstrated better outcomes among insured patients. Perhaps the melanoma mortality rate is higher among the health-insured population because patients with health insurance are more likely to receive long-term disease management and are less likely to die of comorbidities, such as hypertension, diabetes mellitus, stroke, or heart disease.
Physician supply may be correlated with cancer outcomes by serving as a proxy for overall health care resources. Previous studies5 have demonstrated associations between higher numbers of primary care physicians and lower all-cause mortality, cancer mortality, and cardiovascular mortality rates. Our analysis did not find a relationship between primary care physician density and melanoma mortality. Our findings suggest that dermatologist density, which may serve as a proxy for specialists, may be a more suitable proxy for melanoma mortality than primary care physicians.
The limitations of this analysis include the exclusion of 669 counties due to their rural county status. These counties were excluded because limited mortality data were reported from these areas and there were few rural counties ( <5%) with any dermatologists. We hypothesize that inclusion of rural counties, where the incidence of melanoma is low, would skew the analysis because the findings may imply that 0 dermatologists are associated with no melanoma mortality. However, counties where the incidence of melanoma is greater than 0 would have heightened the association we observed between dermatologist density and improved melanoma mortality.
The incidence data used in our analysis did not account for severity of disease at the time of diagnosis. We hypothesized that greater dermatologist density allows for disease detection at an earlier stage. Other reports have made use of the SEER database, which accounts for severity of disease at presentation; however, the merged data set used in our analysis does not reflect extent of disease at presentation. Lastly, in our analysis, no distinction was made between dermatologists engaged in clinical activities full time vs part time or in the way to account for dermatologic services provided by physician assistants, nurse practitioners, and other health care professionals.
In conclusion, our analysis found that within a given county, the presence of a dermatologist is associated with a lower melanoma mortality rate compared with counties that lacked a dermatologist. We speculate that efforts to recruit dermatologists to counties currently lacking such specialists could result in a population-level reduction in melanoma mortality.
Our analysis uses melanoma mortality to highlight the consequence of a maldistributed dermatologist workforce. We have demonstrated that the geographic variation in melanoma mortality is associated with the densities of dermatologists. Given the nature of this field, it is unclear whether dermatologist density affects prevention, diagnosis, treatment, or some combination of the aforementioned. Further studies are needed using staging information to highlight whether dermatologist density is associated with earlier diagnosis of melanoma or improved treatment.
Correspondence: Jeremy S. Bordeaux, MD, MPH, Department of Dermatology, University Hospitals Case Medical Center, 11100 Euclid Ave, Lakeside 3500, Cleveland, OH 44124 (Jeremy.Bordeaux@uhhospitals.org).
Accepted for Publication: August 4, 2011.
Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Savina Aneja, Sanjay Aneja, and Bordeaux. Acquisition of the data: Savina Aneja, Sanjay Aneja, and Bordeaux. Analysis and interpretation of the data: Savina Aneja, Sanjay Aneja, and Bordeaux. Drafting of the manuscript: Savina Aneja. Critical revision of the manuscript for important intellectual content: Sanjay Aneja and Bordeaux. Statistical analysis: Savina Aneja and Sanjay Aneja. Administrative, technical, or material support: Savina Aneja. Study supervision: Bordeaux.
Financial Disclosure: None reported.
Funding/Support: Dr Bordeaux is supported by the Dermatology Foundation Clinical Career Development Award in Dermatologic Surgery.
Online-Only Material: Listen to an author podcast about this article.
1.Eide MJ, Weinstock MA, Clark MA. The association of physician-specialty density and melanoma prognosis in the United States, 1988 to 1993.
J Am Acad Dermatol. 2009;60(1):51-5818937998
PubMedGoogle Scholar 2.Gorey KM. Breast cancer survival in Canada and the USA: meta-analytic evidence of a Canadian advantage in low-income areas.
Int J Epidemiol. 2009;38(6):1543-155119386825
PubMedGoogle Scholar 3.Ricketts TC, Holmes GM. Mortality and physician supply: does region hold the key to the paradox?
Health Serv Res. 2007;42(6, pt 1):2233-225117995563
PubMedGoogle Scholar 4.Philips RL Jr, Dodoo MS, Green LA. Adding more specialists is not likely to improve population health: is anybody listening?
Health Aff (Millwood). Jan-Jun 2005;(suppl Web exclusives):W5-111-W5-11415769798
PubMedGoogle Scholar 5.Lee J, Park S, Choi K, Kwon SM. The association between the supply of primary care physicians and population health outcomes in Korea.
Fam Med. 2010;42(9):628-63520927671
PubMedGoogle Scholar 6.Odisho AY, Cooperberg MR, Fradet V, Ahmad AE, Carroll PR. Urologist density and county-level urologic cancer mortality.
J Clin Oncol. 2010;28(15):2499-250420406931
PubMedGoogle Scholar 7.Goodman DC, Fisher ES, Little GA, Stukel TA, Chang CH, Schoendorf KS. The relation between the availability of neonatal intensive care and neonatal mortality.
N Engl J Med. 2002;346(20):1538-154412015393
PubMedGoogle Scholar 8.DeFazio JL, Marghoob AA, Pan Y, Dusza SW, Khokhar A, Halpern A. Variation in the depth of excision of melanoma: a survey of US physicians.
Arch Dermatol. 2010;146(9):995-99920644028
PubMedGoogle Scholar 9.Hagen TP, Vaughan-Sarrazin MS, Cram P. Relation between hospital orthopaedic specialisation and outcomes in patients aged 65 and older: retrospective analysis of US Medicare data.
BMJ. 2010;340:c16520150193
PubMedGoogle Scholar 10.Biondo S, Kreisler E, Millan M,
et al. Impact of surgical specialization on emergency colorectal surgery outcomes.
Arch Surg. 2010;145(1):79-8620083758
PubMedGoogle Scholar 11.Jacobson CC, Resneck JS Jr, Kimball AB. Generational differences in practice patterns of dermatologists in the United States: implications for workforce planning.
Arch Dermatol. 2004;140(12):1477-148215611425
PubMedGoogle Scholar 12.Kimball AB, Resneck JS Jr. The US dermatology workforce: a specialty remains in shortage.
J Am Acad Dermatol. 2008;59(5):741-74518723242
PubMedGoogle Scholar 13.Resneck JS Jr, Kimball AB. Who else is providing care in dermatology practices? trends in the use of nonphysician clinicians.
J Am Acad Dermatol. 2008;58(2):211-21618222319
PubMedGoogle Scholar 14.Craiglow BG, Resneck JS Jr, Lucky AW,
et al. Pediatric dermatology workforce shortage: perspectives from academia.
J Am Acad Dermatol. 2008;59(6):986-98918835061
PubMedGoogle Scholar 15.MacKie RM, Hole DJ. Incidence and thickness of primary tumours and survival of patients with cutaneous malignant melanoma in relation to socioeconomic status.
BMJ. 1996;312(7039):1125-11288620127
PubMedGoogle Scholar 16.Reyes-Ortiz CA, Goodwin JS, Freeman JL, Kuo YF. Socioeconomic status and survival in older patients with melanoma.
J Am Geriatr Soc. 2006;54(11):1758-176417087705
PubMedGoogle Scholar 17.Van Durme DJ, Ullman R, Campbell RJ, Roetzheim RG. Effects of physician supply on melanoma incidence and mortality in Florida.
South Med J. 2003;96(7):656-66012940314
PubMedGoogle Scholar 18.Greenlee RT, Howe HL. County-level poverty and distant stage cancer in the United States.
Cancer Causes Control. 2009;20(6):989-100019199061
PubMedGoogle Scholar 20.US Department of Health and Human Services. Health Resources and Services Administration: Area Resource File (ARF): national county-level health resource information database.
http://www.arfsys.com. Accessed October 1, 2010 21.US Census Bureau. Small area health insurance estimates: model-based small area health insurance estimates (SAHIE) for counties and states.
http://www.census.gov/did/www/sahie/. Accessed October 1, 2010 23.Roetzheim RG, Pal N, van Durme DJ,
et al. Increasing supplies of dermatologists and family physicians are associated with earlier stage of melanoma detection.
J Am Acad Dermatol. 2000;43(2, pt 1):211-21810906640
PubMedGoogle Scholar 24.Pennie ML, Soon SL, Risser JB, Veledar E, Culler SD, Chen SC. Melanoma outcomes for Medicare patients: association of stage and survival with detection by a dermatologist vs a nondermatologist.
Arch Dermatol. 2007;143(4):488-49417438181
PubMedGoogle Scholar 25.Stitzenberg KB, Thomas NE, Dalton K,
et al. Distance to diagnosing provider as a measure of access for patients with melanoma.
Arch Dermatol. 2007;143(8):991-99817709657
PubMedGoogle Scholar 26.Linos E, Swetter SM, Cockburn MG, Colditz GA, Clarke CA. Increasing burden of melanoma in the United States.
J Invest Dermatol. 2009;129(7):1666-167419131946
PubMedGoogle Scholar 27.Pollitt RA, Clarke CA, Shema SJ, Swetter SM. California Medicaid enrollment and melanoma stage at diagnosis: a population-based study.
Am J Prev Med. 2008;35(1):7-1318482824
PubMedGoogle Scholar 28.Ortiz CA, Goodwin JS, Freeman JL. The effect of socioeconomic factors on incidence, stage at diagnosis and survival of cutaneous melanoma.
Med Sci Monit. 2005;11(5):RA163-RA17215874907
PubMedGoogle Scholar 29.Starfield B. Commentary: how does ‘insurance ’ improve equity in health?
Int J Epidemiol. 2009;38(6):1551-155319553232
PubMedGoogle Scholar 30.Robbins AS, Pavluck AL, Fedewa SA, Chen AY, Ward EM. Insurance status, comorbidity level, and survival among colorectal cancer patients age 18 to 64 years in the National Cancer Data Base from 2003 to 2005.
J Clin Oncol. 2009;27(22):3627-363319470927
PubMedGoogle Scholar