Effect of Health Care Delivery Models on Melanoma Thickness and Stage in a University-Based Referral Center: An Observational Pilot Study | Dermatology | JAMA Dermatology | JAMA Network
[Skip to Navigation]
Sign In
Table 1. 
Characteristics of 234 Patients*
Characteristics of 234 Patients*
Table 2. 
Differences in Tumor Characteristics, Patient Delay, and Physician Delay According to Health Care Access Route*
Differences in Tumor Characteristics, Patient Delay, and Physician Delay According to Health Care Access Route*
Table 3. 
Sensitivity Analysis of Patient Delay* Data With Inclusion of Missing Data Points†
Sensitivity Analysis of Patient Delay* Data With Inclusion of Missing Data Points†
January 2007

Effect of Health Care Delivery Models on Melanoma Thickness and Stage in a University-Based Referral Center: An Observational Pilot Study

Author Affiliations

Author Affiliations: Dermatology Services, VA Health Care Systems, Palo Alto, Calif (Dr Swetter), and Atlanta, Ga (Dr Chen); Health Services Research and Development, Veterans Affairs Health Care System, Atlanta (Dr Chen); and Departments of Dermatology, Stanford University Medical Center, Stanford, Calif (Drs Swetter and Harrington), and Emory University School of Medicine, Atlanta (Drs Soon and Chen).

Arch Dermatol. 2007;143(1):30-36. doi:10.1001/archderm.143.1.30

Objective  To compare the effect of differing health care delivery models, specifically, gatekeeper (GK) vs direct access (DA) routes, on melanoma outcome as measured by tumor thickness and cancer stage at diagnosis.

Design  Retrospective medical record review of patients previously diagnosed as having cutaneous melanoma who were referred to a university-based clinic from January 1, 1996, through December 31, 2000.

Setting  Stanford Pigmented Lesion and Cutaneous Melanoma Clinic, Stanford, Calif.

Patients  Two hundred thirty-four patients with primary melanoma stratified according to health care access route (GK or DA).

Main Outcome Measures  Differences in Breslow thickness, American Joint Committee on Cancer stage, histologic features, patient delay in seeking medical attention, and physician delay in diagnosis (time between initial physician visit and diagnostic biopsy procedure).

Results  Of 234 patients, 168 (72%) were referred through the DA route and 66 (28%) through the GK route. A significant association was found between physician delay and access route; patients in the DA group underwent biopsy sooner (≤3 months vs >3 months) than those in the GK group (P<.001). No significant difference was observed in stage at diagnosis (predominantly stage IA), proportion of nodular melanoma (DA 4% vs GK 2%), patient delay, or median tumor thickness between DA and GK routes (0.42 mm vs 0.50 mm, respectively). A trend toward a greater proportion of histologically ulcerated melanoma was observed in the DA group compared with the GK group (12% vs 5%, respectively; P = .06).

Conclusions  This pilot study demonstrated no difference in outcome between GK and DA routes as measured by melanoma thickness and stage, although patients in the DA group underwent diagnostic biopsy sooner than those in the GK group. The potential effect of health care models on melanoma outcomes merits further study.

The most important prognostic factors in cutaneous melanoma are Breslow thickness and cancer stage at diagnosis.1-4 Whether differing health care delivery models influence these melanoma outcome measures is largely unknown. Recent studies suggest that cancers for which effective screening services are available, including breast, cervical, colorectal cancer, and melanoma, may be diagnosed at earlier stages in health maintenance organization (HMO) enrollees than in fee-for-service (FFS) plan participants.5-8 Improved survival rates for melanoma have also been demonstrated in Medicare patients enrolled in HMOs compared with matched patients enrolled in FFS plans.9 These findings have been attributed to diagnosis at earlier stages rather than to differences in health care delivery, but have largely been derived from analyses of Medicare data and are, thus, limited to populations 65 years or older. Increased scrutiny of the effect of differing health care models on melanoma outcome in all age groups is lacking, as is a broader analysis of tumor thickness and other clinicopathologic factors in combination with American Joint Committee on Cancer (AJCC) stage at diagnosis.

Heightened public and professional awareness of the correlation between early diagnosis and treatment of cutaneous melanoma and improved survival has intensified the effort to identify sources of delay in diagnosis.10-13 However, published data on the effect of patient delay in seeking care and physician delay in obtaining a diagnostic biopsy are inconclusive, and most studies fail to demonstrate a correlation between delay in diagnosis and increased tumor thickness.12-18

While various studies have suggested that diagnostic accuracy of melanoma improves with specialty training in dermatology,19-23 few have specifically addressed the role of the health care delivery model on melanoma outcome or have been performed outside of the older Medicare population.5,9 We conducted a retrospective study to assess the role of direct dermatology access compared with primary care/gatekeeper–regulated specialty access on melanoma outcome as measured by tumor thickness and cancer stage at diagnosis.


Institutional review board approval for the study was obtained from Stanford University Medical Center and Emory University. Retrospective medical record review was conducted for all patients with an International Classification of Diseases, Ninth Revision, diagnosis of cutaneous melanoma who were referred to the Stanford Pigmented Lesion and Cutaneous Melanoma Clinic for evaluation or treatment between January 1, 1996, and December 31, 2000. This time frame was selected as a period of high managed-care activity at Stanford Hospital and Clinics, with patients in managed-care programs constituting up to 38% of the Stanford Dermatology Clinic payer mix (range, 32%-38% per fiscal year assessed). Since January 1, 2001, Stanford Hospital and Clinics has no longer accepted managed-care contracts.

Patients belonging to an exclusive provider organization, capitated HMO, or referred HMO were stratified to the gatekeeper (GK) group, whereas patients who self-paid or belonged to MediCal (California Medicaid program), Medicare, or a preferred provider organization were stratified to the direct access (DA) group. Patients belonging to the Kaiser Permanente HMO were not included in this analysis because DA to dermatology occurs within this health plan in northern California. All patients in the GK group were referred to the Stanford Pigmented Lesion and Cutaneous Melanoma Clinic by a primary care provider (family practice vs obstetrics-gynecology) who either obtained a biopsy specimen of the suspect skin lesion on initial examination or subsequently referred the patient to a health-plan affiliated dermatologist or surgeon for evaluation and diagnostic biopsy. All patients in the DA group were initially seen by a dermatologist for evaluation of a skin lesion that subsequently proved to be melanoma.

To construct variables reflecting diagnostic delay, information was retrieved from each medical record about patient delay and physician delay (ie, professional delay). Patient delay was defined as the interval between the patient first noticing the suspect skin lesion and medical presentation to a dermatologist, primary care physician, or other health professional. This measurement was based on the patient's self report of how long it took to get medical attention for the skin lesion. Physician delay was defined as the interval between patient's initial medical presentation and a diagnostic skin biopsy, as objectively measured by the date of referral from the GK to a dermatologist (for patients in the GK group), the date the patient was first seen by a dermatologist, and the date when a biopsy of the melanoma was initially obtained. Referral time from melanoma diagnosis to initial Stanford Pigmented Lesion and Cutaneous Melanoma Clinic visit was not included in physician delay because a diagnosis of primary melanoma was required for referral to this clinic. The time frame for physician delay was coded as no delay if the patient underwent diagnostic skin biopsy at the first health care provider visit. For statistically meaningful analysis, delay categories were divided into 3 months or less and more than 3 months for patient delay, and no delay, 3 months or less, and more than 3 months for physician delay.

Primary tumor characteristics were analyzed according to health care model to identify statistically and clinically important differences. Histopathologic characteristics included in situ vs invasive melanoma, Breslow thickness, presence or absence of ulceration, histogenetic subtype, anatomical site, and pathologic stage at diagnosis, according to the AJCC 2002 melanoma staging system.24

Additional information included patient age, sex, and race/ethnicity, and personal or family history of melanoma; anatomical site and clinical appearance of the lesion (pigmented vs amelanotic); and specialty of the physician who performed the biopsy (dermatologist or primary care physician). Exclusion criteria included lack of histopathologic confirmation of cutaneous melanoma on initial Stanford visit, or GK referral to a dermatologist because of an unrelated skin disorder that led to an incidental diagnosis of cutaneous melanoma.

We considered that the predominance of thin lesions in our sample may skew the median Breslow depth and, thus, obscure a clinically important difference between the access routes for this factor. Therefore, we analyzed tumor thickness by access group using 2 samples: all tumors and tumors thicker than 1.0 mm. Given the prognostic importance of histologic ulceration, we similarly analyzed our sample for this histologic feature for both all tumors and tumors thicker than 1.0 mm.

Given the large number of missing data points for the patient delay data set as a result of incomplete patient recall, we performed a sensitivity analysis to gauge the robustness of our findings. In the case of patient delay, we alternately analyzed all missing values in the context of 2 scenarios: all missing data actually represented patient delay of 3 months or less, or all missing data actually represented patient delay of more than 3 months. Analyzing patient delay under these 2 extreme scenarios provided information on the extent of the validity of our study results; that is, a change in our results in the face of these hypothetical worst case scenarios would call into question the validity of our results, whereas no change would suggest that our results were valid.

Descriptive and hypothesis-driven statistics were performed using SAS statistical software (SAS Institute Inc, Cary, NC). Categorical variables were analyzed using the χ2 test when fewer than 20% of table cells contained expected counts of 5 or fewer or the Fisher exact test when 20% or more of the table cells contained expected counts of 5 or fewer. Nonparametric continuous variables (eg, Breslow depth) were analyzed using the Wilcoxon rank sum test. Parametric continuous variables (eg, age at diagnosis) were analyzed using the 2-sample t test. Statistical and marginal significance were defined as P ≤.05 and P ≤.10, respectively.


Two hundred thirty-four eligible patients were identified during the study period, 168 (72%) from the DA group and 66 (28%) from the GK group. Patient characteristics according to health care delivery model and system/tumor characteristics for the entire study cohort are given in Table 1. Data were unavailable for 52 patients (35 in the DA group and 17 in the GK group) for the patient delay analysis; 2 patients (GK group) for the physician delay analysis; 4 patients (3 in the DA group and 1 in the GK group) for the Breslow thickness analysis; and 1 patient (GK group) for the histogenetic subtype analysis. The missing information for patient delay was the result of the subjective nature of this data point. Data extracted from the medical records revealed that these patients were unable to accurately recall when they first became aware of the lesion that later proved to be melanoma or that this information was not recorded; patient delay in these cases was coded as unknown.

No difference was observed between the DA and GK groups in terms of sex, race/ethnicity, family history of melanoma, or presence of atypical moles. However, a greater proportion of patients in the DA group were older compared with the GK group (mean age, 55 ± 16 years vs 49 ± 16 years, respectively; P = .01) and had a personal history of melanoma (11% vs 2%, respectively; P = .02; Table 1).

Our study sample reflected the prevalence of thin melanoma in the general population (65% of melanomas were ≤1 mm thick) and demonstrated a highly left-skewed AJCC stage distribution in both access groups (75% stage 0 or I). No statistically significant difference was observed between access routes for median Breslow thickness or in the proportion of in situ vs invasive melanoma, AJCC stage, histologic subtype, melanoma pigmentation (amelanotic vs visibly pigmented), and anatomical site (Table 2). Although tumors were relatively thicker in the GK group for both all tumors (DA 0.42 mm vs GK 0.50 mm) and the subgroup analysis consisting of tumors thicker than 1.0 mm (DA 2.0 mm vs GK 2.40 mm), these differences were neither statistically nor clinically significant based on estimated 5-year survival according to the 2002 AJCC melanoma thickness classification. A significant difference, however, was observed in the proportion of patients with histologically ulcerated melanomas, with a higher proportion of ulcerated tumors occurring in the DA group for both all tumors (DA 12% vs GK 5%; P = .06) and for tumors thicker than 1.0 mm (DA 28% vs GK 11%; P = .02).

For patient delay, no significant difference was observed between the DA and GK groups (32% vs 26% for the ≤3-month time frame and 47% vs 49% for the >3-month time frame, respectively, P = .64) (Table 2). The sensitivity analysis (assuming all missing data fell in the ≤3-month or >3-month time frame) demonstrated no change in study results (P = .76 and P = .37, respectively), which suggests that our findings were valid (Table 3).

Analysis of physician delay revealed a significant difference between comparison groups, with 147 patients (88%) in the DA group undergoing immediate biopsy (no delay) at the initial dermatologist visit, compared with 29 patients (44%) in the GK group in whom a biopsy specimen was obtained at the initial primary care physician visit. Likewise, there were more patients with delay of both 3 months or less and more than 3 months in the GK group compared with the DA group (38% vs 8% and 15% vs 4%, respectively, P = .001) (Table 2).


Limited data are available on the effect of health care systems on melanoma stage at diagnosis, and most studies examining HMO effects on cancer stage or mortality have been confined to the Medicare population. Riley et al5 showed earlier stage at diagnosis for melanoma and cancers of the breast, cervix, and colon, malignant neoplasms for which effective screening services were available for patients enrolled in Medicare HMOs compared with age-matched patients in an FFS plan. In this 5-year analysis, stomach cancer was diagnosed at later stages in HMO enrollees; however, no differences were observed between HMO and FFS enrollees for cancers of the prostate, rectum, buccal cavity and pharynx, bladder, uterus, kidney, and ovary. The HMO effects were strongest in areas with large, well-established HMOs, and promotion and increased coverage of screening services by HMOs was proposed as an explanation for the findings.5 Roetzheim et al25 assessed the effects of Medicare HMO and FFS plans, Medicaid coverage, commercial HMO, preferred provider organization, and indemnity providers, as well as the uninsured, on early detection of melanoma and colorectal, breast, and prostate cancer in Florida and, as with other health care system analyses,26-30 demonstrated that only Medicaid coverage or lack of health insurance was associated with cancer diagnosis at later stages.

Gatekeeper HMOs typically require primary care referral to specialists, as opposed to direct specialist access through preferred provider organizations, FFS plans, or Medicaid or Medicare, assuming that specialists accept patients enrolled in these federal programs. Physician supply alone has been correlated with melanoma mortality, with increasing numbers of both dermatologists and family physicians associated with earlier detection of melanoma in a recent Florida analysis.31 The promotion of and accessibility to cancer screening may be better established in the large HMO setting than in FFS plans, and lack of screening and medical access most certainly has a role in increased mortality for various cancers in the uninsured.32,33

A recent study by Kirsner et al9 demonstrated diagnosis of earlier stage melanomas in a population of Medicare patients enrolled in HMO health care delivery systems compared with matched patients enrolled in FFS plans. In addition, patients enrolled in HMOs had improved survival after melanoma diagnosis compared with patients in FFS plans; median overall survival was more than 26 months longer. While the HMO-associated survival advantage did not persist when controlled for stage at diagnosis, the authors suggest that this was the result of diagnosis at earlier stages compared with that in the FFS group. Increased coverage of preventive services, routine access to primary care providers, and direct access to dermatologists through HMO plans were proposed as possible reasons for the study findings.9

Our 5-year analysis in a university-based melanoma clinic similarly showed no differences in tumor thickness or stage between GK and DA routes, although patients in the DA group underwent biopsy sooner and seemed to have more ulcerated primary melanomas. While our pilot study is small in comparison with the analysis of Kirsner et al,9 it has the advantages of including patients of all ages with melanoma; expanded and more accurate determination of health care model per subject; inclusion of tumor characteristics including ulceration, visible pigmentation, and histogenetic subtype; and more precise staging of cutaneous melanoma, adhering to the 2002 AJCC melanoma staging system, as opposed to the broader categories of in situ, local, regional, and distant disease used for staging in the Medicare analysis.9 Furthermore, it addresses the question of how GK-associated managed care affects melanoma thickness, an issue not completely captured in the large-scale Medicare HMO and FFS comparative analyses, since traditional HMO-directed care may include health plans with direct access to dermatology, as in the Kaiser Permanente system.

Patient delay in seeking medical care has been proposed as the principal factor impeding early diagnosis of melanoma,12,15,17 although few studies15,16 have demonstrated a positive correlation between duration of patient delay and prognostic indicators such as tumor thickness or stage at diagnosis. Most studies have shown a weak positive correlation18 or no association (including ours) between patient delay and tumor thickness.10,12,13,17,23,34,35 However, increased patient knowledge and awareness of cutaneous melanoma has been associated with reduction in delay time to seek medical attention36 and with thinner lesions at diagnosis.10,12,13,37

The effect of medical delay on melanoma stage or prognosis has been less controversial, with most studies (including ours) suggesting that physician delay has little or no effect on melanoma thickness or outcome.11,13,17,18,23,37 In a prospective multicenter study in France of 590 patients with melanoma,11 physician responsibility accounted for only a small part of the total delay before melanoma removal. However, shorter medical delays were apparent in patients who visited a dermatologist rather than a general practitioner. That our data showed that reduced time to biopsy by dermatologists compared with primary care physicians is not surprising, given the lower threshold dermatologists have for performing biopsy of suspect pigmented lesions by virtue of training. As with previous studies, however, shorter physician delay in the DA group did not affect tumor thickness or stage at diagnosis.

While we found no difference in median tumor thickness between the DA and GK groups, ulceration of the primary tumor was significantly more common in the DA group when considering tumors more than 1.0 mm thick and was marginally significant when considering all tumors. This difference suggests a potential advantage in earlier biopsy of primary melanomas with this adverse prognostic factor,4,24,38-40 although no difference in thickness or stage was evident in our analysis. This result could also be explained by the older age of patients in the DA group compared with the GK group (mean age, 55 vs 49 years, respectively; P = .01), because older age correlates with both increased tumor thickness and higher incidence of histologic ulceration in primary melanoma.41

Study limitations include the retrospective, nonrandomized nature of the study; referral bias to a tertiary care facility; small sample size; and potential geographic variation in health care delivery and predominance of one type of model over another in northern California. All of these factors may make our results less generalizable to other populations or regions of the United States. The imbalance of patients in the GK and DA groups in our study seems to reflect the proportion of managed-care patients in the Stanford University dermatology clinics during the period assessed. A larger sample size is required to adequately differentiate between the reported distributions of median thicknesses; 1114 cases in each sample (DA and GK) would be necessary to maintain an α value of .05 and power of 80%. In addition, the difference in personal history of melanoma between the DA and GK groups (11% vs 2%, respectively; P = .02) may have resulted in a self-selection bias whereby patients at increased risk for second primary tumors opted for DA to dermatology rather than GK referral. Finally, although our data capture rate was high, we had a 25% missing data rate for the patient delay analysis. Recognizing this, we performed a sensitivity analysis, which showed our results to be robust insofar as lack of significant difference in patient delay according to health care model.

While preliminary, our data are consistent with those of other Medicare-based HMO vs FFS analyses and demonstrate that DA vs GK health care delivery models may have no appreciable effect on median thickness or stage of melanoma at diagnosis and, thus, no likely effect on melanoma-related outcomes. Rather, the variable services and rules within health care delivery models, including potential factors such as availability and use of routine cancer screening and DA to dermatology, are most likely to affect melanoma outcome. However, evidence suggests that routine skin cancer screening, regardless of health care model, may be less prevalent by primary care physicians or even dermatologists than screening for other cancers, such as prostate, breast, and colorectal cancers.42-44

Our data are consistent with previous reports of the weak correlation between medical delay and increased tumor thickness or melanoma stage. Although we found a difference between the DA and GK groups for histologic ulceration, this represents only 1 factor in melanoma staging. Confirmatory studies and further analysis of the potential effect of health care delivery models on prompt diagnosis of more aggressive melanomas (eg, those with ulceration, lymphovascular invasion, high mitotic rate, or nodular subtype) are warranted.

Correspondence: Susan M. Swetter, MD, Department of Dermatology, Stanford University Medical Center, 900 Blake Wilbur Dr, W0069, Stanford, CA 94305 (sswetter@stanford.edu).

Financial Disclosure: None reported.

Previous Presentation: This study was presented in part at the 65th Annual Meeting of the Society for Investigative Dermatology, International Dermato-Epidemiology Association Americas Chapter; April 29, 2004; Providence, RI.

Back to top
Article Information

Accepted for Publication: May 25, 2006.

Author Contributions: All authors had full access to all of 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: Swetter, Soon, Harrington, and Chen. Acquisition of data: Swetter and Harrington. Analysis and interpretation of data: Swetter, Soon, and Chen. Drafting of the manuscript: Swetter, Soon, Harrington, and Chen. Critical revision of the manuscript for important intellectual content: Swetter, Soon, Harrington, and Chen. Statistical analysis: Soon and Chen. Administrative, technical, and material support: Swetter and Soon. Study supervision: Swetter and Chen.

Funding/Support: This study was supported in part by Mentored Patient Oriented Career Development Award K23AR02185-01A1 from the National Institute of Arthritis, Musculoskeletal, and Skin Diseases (Dr Chen) and by a David Carter Martin Career Development Award from the American Skin Association (Dr Chen).

Acknowledgment: We thank Duke Khuu, MD, for assistance with initial data collection while a medical student at Stanford University School of Medicine; and Emir Veledar, PhD, Emory University School of Medicine, for statistical support.

Breslow  A Thickness, cross-sectional areas and depth of invasion in the prognosis of cutaneous melanoma.  Ann Surg 1970;172902- 908Google ScholarCrossref
Clark  WH  JrElder  DEGuerry  D  IV  et al.  Model predicting survival in stage I melanoma based on tumor progression.  J Natl Cancer Inst 1989;811893- 1904PubMedGoogle ScholarCrossref
Garbe  CBüttner  PBertz  J  et al.  Primary cutaneous melanoma: identification of prognostic groups and estimation of individual prognosis for 5093 patients.  Cancer 1995;752484- 2491PubMedGoogle ScholarCrossref
Balch  CMSoong  SJGershenwald  JE  et al.  Prognostic factors analysis of 17,600 melanoma patients: validation of the American Joint Committee on Cancer staging system.  J Clin Oncol 2001;193622- 3634PubMedGoogle Scholar
Riley  GFPotosky  ALLubitz  JDBrown  ML Stage of cancer at diagnosis for Medicare HMO and fee-for-service enrollees.  Am J Public Health 1994;841598- 1604PubMedGoogle ScholarCrossref
Potosky  ALMerrill  RMRiley  GF  et al.  Breast cancer survival and treatment in health maintenance organization and fee-for-service settings.  J Natl Cancer Inst 1997;891683- 1691PubMedGoogle ScholarCrossref
Potosky  ALBreen  NGraubard  BIParsons  PE The association between health care coverage and the use of cancer screening tests: results from the 1992 National Health Interview Survey.  Med Care 1998;36257- 270PubMedGoogle ScholarCrossref
Kirsner  RSMa  FFleming  L  et al.  The effect of Medicare health care delivery systems on survival for patients with breast and colorectal cancer.  Cancer Epidemiol Biomarkers Prev 2006;15769- 773PubMedGoogle ScholarCrossref
Kirsner  RSWilkinson  JDMa  FPacheco  HFederman  DG The association of Medicare health care delivery systems with stage at diagnosis and survival for patients with melanoma.  Arch Dermatol 2005;141753- 757PubMedGoogle ScholarCrossref
Richard  MAGrob  JJAvril  MF  et al.  Delays in diagnosis and melanoma prognosis: the role of patients.  Int J Cancer 2000;89271- 279PubMedGoogle ScholarCrossref
Richard  MAGrob  JJAvril  MF  et al.  Delays in diagnosis and melanoma prognosis: the role of doctors.  Int J Cancer 2000;89280- 285PubMedGoogle ScholarCrossref
Schmid-Wendtner  MHBaumert  JStange  J  et al.  Delay in the diagnosis of cutaneous melanoma: an analysis of 233 patients.  Melanoma Res 2002;12389- 394Google ScholarCrossref
Carli  PDe Giorgi  VPalli  D  et al.  Dermatologist detection and skin self-examination are associated with thinner melanomas: results from a survey of the Italian Multidisciplinary Group on Melanoma.  Arch Dermatol 2003;139607- 612PubMedGoogle ScholarCrossref
Cassilith  BRClark  WH  JrHeiberger  RMMarch  VTenaglia  A Relationship between patient's early recognition of melanoma and depth of invasion.  Cancer 1982;49198- 200PubMedGoogle ScholarCrossref
Temoshok  LDiClemente  RJSweet  DMBlois  MSSagebiel  RW Factors related to patient delay in seeking medical attention for cutaneous malignant melanoma.  Cancer 1984;543048- 3053PubMedGoogle ScholarCrossref
Temoshok  LHeller  BWSagebiel  RW  et al.  The relationship of psychosocial factors to prognostic indicators in cutaneous malignant melanoma: a prospective study in 250 patients.  J Psychosom Res 1985;29139- 153PubMedGoogle ScholarCrossref
Krige  JEJIsaacs  SHudson  DAKing  HSStrover  RMJohnson  CA Delay in the diagnosis of cutaneous malignant melanoma: a prospective study in 250 patients.  Cancer 1991;682064- 2068PubMedGoogle ScholarCrossref
Richard  MAGrob  JJAvril  MF  et al.  Melanoma and tumor thickness.  Arch Dermatol 1999;135269- 274PubMedGoogle ScholarCrossref
Fisher  NMSchaffer  VJBerwick  MBolognia  JL Breslow depth of cutaneous melanoma: impact of factors related to surveillance of the skin, including prior skin biopsies and family history of melanoma.  J Am Acad Dermatol 2005;53393- 406PubMedGoogle ScholarCrossref
Kopf  AWMintzis  MBart  RS Diagnostic accuracy in malignant melanoma.  Arch Dermatol 1975;1111291- 1292PubMedGoogle ScholarCrossref
Ramsay  DLFox  AB The ability of primary care physicians to recognize the common dermatoses.  Arch Dermatol 1981;117620- 622PubMedGoogle ScholarCrossref
Cassilith  BRClark  WH  JrLusk  EJFrederick  BEThompson  CJWalsh  WP How well do physicians recognize melanoma and other problem lesions?  J Am Acad Dermatol 1986;14555- 560PubMedGoogle ScholarCrossref
Cassilith  BRTemoshok  LFrederick  BE  et al.  Patient and physician delay in melanoma diagnosis.  J Am Acad Dermatol 1988;18591- 598PubMedGoogle ScholarCrossref
Balch  CMBuzaid  ACSoong  SJ  et al.  Final version of the American Joint Committee on Cancer staging system for cutaneous melanoma.  J Clin Oncol 2001;193635- 3648PubMedGoogle Scholar
Roetzheim  RGPal  NTennant  C  et al.  Effects of health insurance and race on early detection of cancer.  J Natl Cancer Inst 1999;911409- 1415PubMedGoogle ScholarCrossref
Ayanian  JZKohler  BAAbe  TEpstein  AM The relation between health insurance coverage and clinical outcomes among women with breast cancer.  N Engl J Med 1993;329326- 331PubMedGoogle ScholarCrossref
Osteen  RTWinchester  DPHussey  DH  et al.  Insurance coverage of patients with breast cancer in the 1991 Commission on Cancer Patient Care Evaluation Study.  Ann Surg Oncol 1994;1462- 467PubMedGoogle ScholarCrossref
Hunter  CPRedmond  CKChen  VW  et al. Black/White Cancer Survival Study Group, Breast cancer: factors associated with stage at diagnosis in black and white women.  J Natl Cancer Inst 1993;851129- 1137PubMedGoogle ScholarCrossref
Roetzheim  RGPal  NGonzalez  ECFerrante  JMVan Durme  DJKrischer  JP Effects of health insurance and race on colorectal cancer treatments and outcomes.  Am J Public Health 2000;901746- 1754PubMedGoogle ScholarCrossref
Roetzheim  RGGonzales  ECFerrante  JMPal  NVan Durme  DJKrischer  JP Effects of health insurance and race on breast carcinoma treatments and outcomes.  Cancer 2000;892202- 2213PubMedGoogle ScholarCrossref
Van Durme  DJUllman  RCampbell  RJRoetzheim  R Effects of physician supply on melanoma incidence and mortality in Florida.  South Med J 2003;96656- 660PubMedGoogle ScholarCrossref
National Cancer Institute Cancer Screening Consortium for Underserved Women, Breast and cervical cancer screening among underserved women: baseline survey results from six states.  Arch Fam Med 1995;4617- 624PubMedGoogle ScholarCrossref
Hayward  RAShapiro  MFFreeman  HECorey  CR Who gets screened for cervical and breast cancer? results from a new national survey.  Arch Intern Med 1988;1481177- 1181PubMedGoogle ScholarCrossref
Blum  ABrand  CUEllwanger  U  et al.  Awareness and early detection of cutaneous melanoma: an analysis of factors related to delay in treatment.  Br J Dermatol 1999;141783- 787PubMedGoogle ScholarCrossref
Baccard  MChevret  SChemaly  PMorel  P Delay in diagnosing melanoma: a prospective study in 102 patients [in French].  Ann Dermatol Venereol 1997;124601- 606PubMedGoogle Scholar
Oliveria  SAChristos  PJHalpern  ACFine  JABarnhill  RLBerwick  M Patient knowledge, awareness, and delay in seeking medical attention for malignant melanoma.  J Clin Epidemiol 1999;521111- 1116PubMedGoogle ScholarCrossref
Rampen  FHRumke  PHart  AA Patients' and doctors' delay in the diagnosis and treatment of cutaneous melanoma.  Eur J Surg Oncol 1989;15143- 148PubMedGoogle Scholar
Vollmer  RT Malignant melanoma: a multivariate analysis of prognostic factors.  Pathol Annu 1989;24383- 407PubMedGoogle Scholar
Balch  CM Cutaneous melanoma: prognosis and treatment results worldwide.  Semin Surg Oncol 1992;8400- 414PubMedGoogle ScholarCrossref
Buzaid  ACRoss  MIBalch  CM  et al.  Critical analysis of the current American Joint Committee on Cancer staging system for cutaneous melanoma and proposal of a new staging system.  J Clin Oncol 1997;151039- 1051PubMedGoogle Scholar
Chao  CMartin  RC  IIRoss  MI  et al.  Correlation between prognostic factors and increasing age in melanoma.  Ann Surg Oncol 2004;11259- 264PubMedGoogle ScholarCrossref
Federman  DGKravetz  JDHaskell  SGMa  FKirsner  RS Full-body skin examinations and the female veteran: prevalence and perspective.  Arch Dermatol 2006;142312- 316PubMedGoogle Scholar
Federman  DGKravetz  JDTobin  DGMa  FKirsner  RS Full-body skin examinations: the patients' perspective.  Arch Dermatol 2004;140530- 534PubMedGoogle Scholar
Federman  DGKravetz  JDKirsner  RS Skin cancer screening by dermatologists: prevalence and barriers.  J Am Acad Dermatol 2002;46710- 714PubMedGoogle ScholarCrossref