Behavioral determinants of successful early melanoma detection

Role of self and physician skin examination


  • Susan M. Swetter MD,

    Corresponding author
    1. Dermatology Service, Veterans Affairs Palo Alto Health Care System, Palo Alto, California
    2. Department of Dermatology, Pigmented Lesion and Melanoma Program, Stanford University Medical Center and Cancer Institute, Stanford, California
    • Department of Dermatology/Cutaneous Oncology, Stanford University Medical Center and Cancer Institute, 900 Blake Wilbur Drive, W1013, Stanford, CA 94305-5356

    Search for more papers by this author
    • Fax: (650) 496-2573

  • Ricardo A. Pollitt MD, PhD,

    1. Department of Dermatology, Pigmented Lesion and Melanoma Program, Stanford University Medical Center and Cancer Institute, Stanford, California
    Current affiliation:
    1. Ricardo A. Pollitt's current address: Department of Dermatology, University of California, San Francisco, San Francisco, California
    Search for more papers by this author
  • Timothy M. Johnson MD,

    1. Department of Dermatology, University of Michigan Medical School, Ann Arbor, Michigan
    2. Department of Otolaryngology, University of Michigan Medical School, Ann Arbor, Michigan
    3. Department of Surgery, University of Michigan Medical School, Ann Arbor, Michigan
    Search for more papers by this author
  • Daniel R. Brooks DSc, MPH,

    1. Department of Epidemiology, Boston University School of Public Health, Boston, Massachusetts
    Search for more papers by this author
  • Alan C. Geller RN, MPH

    1. Department of Society, Human Development, and Health, Harvard School of Public Health, Boston, Massachusetts
    Search for more papers by this author

  • We acknowledge Mitzi C. Rabe, RN, BSN, OCN for her role as study coordinator at the University of Michigan. We also thank the Melanoma Prevention Working Group for its valuable contributions to the initial study design and analysis.

  • The sponsor had no role in the design or conduct of the study; in the collection, analysis, or interpretation of data; or in the preparation, review, or approval of the article.



Reduced melanoma mortality should result from an improved understanding of modifiable factors related to early detection. The authors of this report surveyed newly diagnosed patients to identify differences in prediagnosis behavioral and medical care factors associated with thinner versus thicker melanoma.


In total, 566 adults with invasive melanoma completed questionnaires within 3 months of diagnosis on demographics, health care access, skin self-examination (SSE), and physician skin examination (PSE) practices in the year before diagnosis. SSE was measured by us e of a melanoma picture aid and routine examination of some/all body sites versus none. Patient-reported partial or full-body PSE also was assessed. Melanoma thickness was dichotomized at 1 mm.


Patient ranged in age from 18 years to 99 years, and 61% were men. The median tumor thickness was 1.25 mm, and 321 tumors (57%) were >1 mm thick. Thinner tumors (≤1 mm) were associated with age ≤60 years (P = .0002), women (P = .0127), higher education level (P = .0122), and physician discovery (P ≤ .0001). Patients who used a melanoma picture aid and performed routine SSE were more likely to have thinner tumors than those who did not (odds ratio [OR], 2.66; 95% confidence interval [CI], 1.48-4.80). Full-body PSE was associated with thinner tumors (OR, 2.51; 95% CI, 1.62-3.87), largely because of the effect of PSE in men aged >60 years (OR, 4.09 95% CI, 1.88-8.89).


SSE and PSE were identified as complementary early detection strategies, particularly in men aged >60 years, in whom both partial and full-body PSE were associated with thinner tumors. Given the high rates of physician access, PSE may be a more practical approach for successful early detection in this subgroup with highest mortality. Cancer 2012. © 2011 American Cancer Society.


Known factors related to thinner versus thicker cutaneous melanomas at diagnosis involve characteristics related to the patient, tumor, and mode of discovery. Thinner tumors are associated with physician detection (vs detection by the patient or significant other),1-9 women,1, 4, 6, 10-14 younger age,10-15 higher level of education6, 8, 14-16 and socioeconomic status,17-20 increased access to specialty care,2, 5, 6, 9, 21 presence of atypical nevi,4, 8 and non-nodular or desmoplastic histologic subtypes.10-13, 22-25 However, these factors have not been studied in combination with patient health behaviors and practices (eg, skin self-examination [SSE] and health care use) and physician skin examination (PSE) during routine medical care, as a means to promote early detection of cutaneous melanoma.

We performed a multi-institutional assessment of prediagnosis behavioral and medical care/access factors in men and women aged ≥18 years with newly diagnosed, cutaneous melanoma to identify differences associated with thinner versus thicker tumors. Our objective was to investigate the associations among self- and physician examination practices and thinner melanoma at diagnosis, to more effectively formulate clinical and public health strategies for early melanoma detection. We hypothesized that both PSE and SSE practices would result in the diagnosis of thinner tumors; that their effect would differ according to age, sex, and type of skin examination performed (full-body vs partial); and that regular health care access would result in earlier detection by increasing the opportunity for PSE.


Study Participants

Approval for the study was obtained from the institutional review boards of Stanford University Medical Center (SUMC), Veterans Affairs Palo Alto Health Care System (VAPAHCS), and the University of Michigan (UM). Eligible, consecutive patients aged ≥18 years with a recent diagnosis of invasive cutaneous melanoma were surveyed at SUMC/VAPAHCS, as previously described.26 Because of the high proportion of thin melanoma at UM, which evaluates a population-based sample representative of the state (concordant with the state registrar), all eligible patients with melanoma >2 mm and a random sample of 33% of individuals with tumors ≤2 mm were surveyed. Patients were surveyed in the melanoma clinics from May 17, 2006 through March 31, 2009, within 3 months of diagnostic biopsy. Respondents were instructed in person to answer all questions for the 12 months before diagnosis, and there were reminders in multiple locations on the survey to reinforce this point.

Skin Self-Examination Measures

We used a previously published measurement of SSE incorporating the number of body sites examined (ranging from 1 to 13 sites) and the use of a picture aid illustrating a melanoma tumor.26 An earlier analysis of a subset of our study patients (321 of 566) that examined only the effect of SSE had demonstrated that routine inspection of at least some of the measured body areas and use of a melanoma picture aid were associated significantly with reduced tumor thickness, whereas the frequency of mole examination was not.

Demographic Measures

The data analyzed included patient demographics (age, sex, education level, race/ethnicity, marital/cohabitation status), previous melanoma and nonmelanoma skin cancer history, and family history of skin cancer. Nevus count and the presence of clinical atypical nevi (CAN) were assessed by the clinician at the initial melanoma visit. Age was dichotomized at age 60 years given increasing rates of thicker tumors and increased mortality among older individuals (especially men)20, 27-29 and evidence suggesting differences in melanoma awareness and skin examination practices by age.10-15

Physician Skin Examination Variables

Health care use in the year before diagnosis was assessed by questions asking where patients usually went when sick or in need of health advice (physician's office/clinic/health center vs urgent care center/emergency room vs other), whether they had a physician whom they consulted regularly for routine care, how many times they visited a physician during this year, whether a physician examined their skin for cancer during any visits, and whether the physician examined the patient's whole skin or just a particular lesion. Patients who responded “do not know” were excluded from specific analyses.

Outcome Measures

The primary study outcome was melanoma tumor depth at diagnosis. Because nearly 75% of fatal melanomas are associated with lesions thicker than 1 mm,30 this tumor cutoff point was used. We considered analyzing tumor thickness as a continuous outcome variable; however, this required log transformation to correct for the right-skewed distribution of melanoma thickness and the presentation of data with a geometric mean thickness,26 which are not easily interpretable.

Statistical Analysis

We used chi-square statistics to assess associations and conducted multivariable logistic regression analyses to assess the effects of various patient characteristics on melanoma tumor thickness at diagnosis. The odds ratio (OR) and 95% confidence interval (CI) for having a thinner (≤1 mm thick) tumor was calculated for different levels of each patient measure compared with a reference level. Decisions to include covariates as confounders in the models were based on a 10% change in the size of the independent variable's beta parameter upon inclusion of the covariate. The potential confounders examined included age, sex, race/ethnicity, education level, nevus count, and histologic subtype. Age, sex, and nevus count were retained in the models.

Statistical interaction between SSE measures and covariates was assessed by the significance of interaction terms (cutoff of P < .2 for consideration) that were entered into the models, and likelihood ratio tests were used to compare the model fit.31 Age, sex, education, histology, and the number of physician visits in the last year were examined as potential effect modifiers. Analyses were conducted using the SAS statistical software package (version 9.2; SAS Institute, Inc., Cary, NC). All tests of significance were 2-sided, and significance was defined as P < .05.



In total, 566 of 719 eligible patients completed the questionnaire (79%), including 225 from SUMC/VAPAHCS and 341 from UM. Refusals were related to patients' lack of interest/time, visual impairment, or anxiety related to diagnosis. Table 1 presents patient demographic information and tumor characteristics. Patients ranged in age from 18 years to 99 years (mean ± standard deviation, 56.4 ± 15.8 years; median, 58 years). Ninety-four percent of patients reported this as their first melanoma, and 95% denied having a prior nonmelanoma skin cancer.

Table 1. Frequencies of Sociodemographic Variables, Tumor Characteristics, and Nevus Count by Melanoma Tumor Depth at Diagnosis Dichotomized at 1 Millimeter Thickness (N = 566
 Tumor Depth Dichotomized at 1 mm: Frequency (%)  
Variable≤1 mm>1 mmRow TotalP
  1. Abbreviations: ALM, acral lentiginous melanoma; LMM, lentigo maligna melanoma; NOD, nodular melanoma; SSM, superficial spreading melanoma.

Age, y   .0002
 ≤4055 (54)46 (46)101 
 41-5048 (47)54 (53)102 
 51-6067 (50)67 (50)134 
 61-7046 (39)72 (61)118 
 ≥7129 (26)82 (74)111 
Sex   .0127
 Men135 (39)210 (61)345 
 Women110 (50)111 (50)221 
Education   .0122
 High school71 (36)127 (64)198 
 Associates' degree40 (41)57 (59)97 
 Four-year college134 (49)137 (51)271 
Race/ethnicity   .1422
 White237 (44)302 (56)539 
 Nonwhite8 (30)19 (70)27 
No. of nevi   .1590
 0-20156 (42)218 (58)374 
 20-5047 (41)69 (59)116 
 50-10031 (55)25 (45)56 
 ≥10011 (55)9 (45)20 
Histologic subtype   <.0001
 SSM188 (58)135 (42)323 
 NOD3 (4)78 (96)81 
 LMM29 (60)19 (40)48 
 ALM5 (28)13 (72)18 
 Desmoplastic2 (9)21 (91)23 
 Other/unclassified16 (24)51 (76)67 
Ulceration   <.0001
 Yes5 (5)104 (95)109 
 No239 (53)210 (47)449 
Anatomic location   <.0001
 Head/neck32 (26)91 (74)123 
 Trunk100 (48)109 (52)209 
 Extremity105 (49)111 (51)216 

Tumor Characteristics

The median tumor thickness was 1.25 mm (mean ± standard deviation, 2.14 ± 2.64 mm), and 321 patients (57%) had melanomas >1 mm thick. The median tumor thickness for patients from SUMC/VAPAHCS and from UM was 1.2 mm and 1.3 mm, respectively. Frequencies according to the American Joint Committee on Cancer 2009 tumor classification32 were 43% T1, 23% T2, 21% T3, and 13% T4. Thinner tumors were associated with superficial spreading (P < .0001) and lentigo maligna (P = .012) histologic subtypes and with visible tumor pigmentation (vs amelanotic; P = .0002), absence of ulceration (P ≤ .0001), and extremity or trunk location compared with head and neck location (P ≤ .0001).

Patient Characteristics

Thinner tumors were associated with younger age (≤60 years; P < .0001), women (P = .0127), and a higher education level (P = .0122) but not with marital/cohabitation status, prior skin cancer, or personal or family history of skin cancer. Thinner tumors were not associated with the presence of CAN, which was evident in 26% of respondents (OR, 1.06; 95% CI, 0.65-1.72), or with an increasing number of moles (OR, 1.60; 95% CI, 0.88-2.92 for patients with >50 moles vs ≤50 moles). However, thinner tumors were associated with the patient previously being told by a physician that he/she had atypical/dysplastic nevi (OR, 1.88; 95% CI, 1.15-3.06).

Mode of Melanoma Discovery

Most patients (53%) self-detected melanoma compared with detection by physicians (19%), spouse/partners (16%), and family members/friends/other (11%). Dermatologists detected 74 melanomas (13%) compared with 35 melanomas (6%) that were detected by primary care physicians. Self-detection was more common in younger patients (aged ≤60 years) than in older patients (57% vs 46%; P = .01). Tumors discovered by a physician were significantly thinner than those discovered by the spouse/partner or by the patient (OR, 4.98; 95% CI, 3.00-8.27 for physician detection vs patient detection). No significant difference was observed between discovery by a dermatologist (median thickness, 0.73 mm) versus a nondermatologist (0.72 mm). There was no clear pattern of association between how long patients waited to see a physician for a skin examination after noting a suspicious skin lesion and thickness.

Skin Self-Examination Practices and Thickness

Twenty-four percent of patients reported use of a melanoma picture for SSE in the year before diagnosis, the use of which was associated with thinner tumors, as was routine self-examination of some/all of the body compared with none (OR, 1.98; 95% CI, 1.24-3.18) (Table 2). The use of a melanoma picture aid in combination with routine SSE had an even stronger association with thinner tumors (OR, 2.66; 95% CI, 1.48-4.80) compared with neither of these practices.

Table 2. Frequencies of Patient-Associated Behaviors/Practices Related to Skin Self-Examination With Odds Ratios and 95% Confidence Intervals That the Patient Had a Thinner Tumor (≤1 mm) at Diagnosis, Adjusted for Age, Sex, and Nevus Count
Variable/ResponseNo. of PatientsOR for Tumor ≤1 mm95% CI
  • Abbreviations: CI, confidence interval; OR, odds ratio; Ref, reference group; SSE, skin self-examination.

  • a

    SSE variables pertain to the year before melanoma diagnosis.

Ever used a melanoma picture to aid in SSEa   
How often patient carefully examined his/her molesa   
 Every 1-2 mo1020.790.49-1.27
 Every 6 m941.180.72-1.93
 Every y941.000.61-1.64
Routinely examined skin on some/all of bodya   
Routinely examined skin and/or ever used a picture aid in SSEa   
 No picture aid but routinely examined some/all of skin3262.171.28-3.66
 Used picture aid for routine SSE1342.661.48-4.80
 No picture aid, did not routinely examine skin89Ref 
When patient first became concerned about the skin lesion   
 Only at time of diagnosis1711.910.97-3.77
 ≤1 y3400.650.35-1.23
 >1 y before diagnosis47Ref 
Could easily see the lesion   
 No/do not know181Ref 
Lesion color   
 Do not know621.290.73-2.27
 Skin colored320.220.08-0.58

Medical Access, Health Care Use, and Physician Skin Examination

Greater than 95% of patients had health insurance, and 82% had a regular physician. Eighty-nine percent of patients reported at least 1 physician visit in the year before diagnosis, and 47% reported receiving a skin examination. Thinner tumors were associated with having 1) a regular health care location/facility when sick/needing advice (OR, 1.64; 95% CI, 0.95-2.84), 2) a physician for routine care (OR, 1.64; 95% CI, 1.03-2.62), 3) at least 1 physician visit (OR, 3.62; 95% CI, 1.75-7.48), and 4) receiving a PSE for cancer (OR, 1.78; 95% CI, 1.24-2.56) (Table 3). Overall, patients who received a full-body skin examination were significantly more likely to have thinner tumors (OR, 2.51; 95% CI 1.62-3.87), whereas those who only had a particular lesion examined were not.

Table 3. Frequencies of Health Care-Associated Behaviors/Practices With Odds Ratios and 95% Confidence Intervals That the Patient Had a Thinner Tumor (≤1 mm) at Diagnosis, Adjusted for Age, Sex, and Nevus Count
Variable/ResponseNo. of PatientsOR for Tumor ≤1 mm95% CI
  • Abbreviations: CI, confidence interval; OR, odds ratio; Ref, reference group.

  • a

    Variables pertain to the year before melanoma diagnosis.

Had a usual place to go when sick/needed health advicea
 Physician's office/clinic/health left4881.640.95-2.84
Had a physician for routine carea
Patient states they received a physician skin examination for cancera
Type of skin examination performed by physician
 Particular lesion931.160.71-1.90
 Whole skin1472.511.62-3.87
Who first noticed the mark that turned out to be melanoma?
 Primary care physician/dermatologist1094.983.00-8.27
During the year before you found out you had melanoma, how many times did you visit a physician?
 2-3 Times2034.171.96-8.89
 >3 Times1733.361.54-7.33
During the year before you found out you had melanoma, did you visit a physician at least once?

Effect Modification by Age and by Sex

A strong statistical interaction was observed by age (dichotomized at age 60 years) and sex for the associations between multiple measures of PSE and SSE and tumor thickness (Table 4). Patients aged >60 years who had a physician examine their skin for cancer had significantly higher odds of having a thinner tumor (OR, 3.11; 95% CI, 1.65-5.86) compared with those who did not. In contrast, receipt of a PSE was not associated with thinner tumors among patients aged <60 years. Patients aged >60 years who received a full-body or partial PSE were more likely to have thinner tumors compared with those who did not (OR, 3.67 [95% CI, 1.84-7.33] and OR, 2.93 [95% CI, 1.29-6.67], respectively). Receipt of a full-body PSE was associated with thinner tumors (OR, 2.06; 95% CI, 1.14-3.71) among younger patients, but receipt of a partial PSE was not.

Table 4. Odds Ratios by Age and Sex: Frequencies of Variables With Odds Ratios and 95% Confidence Intervals That the Patient Had a Thinner Tumor (≤1 mm) at Diagnosis, Adjusted for Age and Stratified According to Age ≤60 Years Versus Age >60 Years, Sex, or Both Age and Sex
Variable/Subgroup ResponseNo. of PatientsOR for Tumor ≤1 mm (95% CI)
  • Abbreviations: CI, confidence interval; OR, odds ratio; Ref, reference group; SSE, skin self-examination.

  • a

    Variables pertain to the year before melanoma diagnosis.

Stratified by age
 Ever used melanoma picture to look at skina  
  Aged ≤60 y  
   Yes840.94 (0.57-1.57)
  Aged >60 y  
   Yes502.84 (1.46-5.49)
 Patient states they received a physician skin examination for cancera
  Aged ≤60 y  
   Yes1271.23 (0.79-1.99)
  Aged >60 y  
   Yes1313.11 (1.65-5.86)
 Type of skin examination performed by physician  
  Aged ≤60 y  
   Particular lesion520.68 (0.36-1.28)
   Whole skin682.06 (1.14-3.71)
  Aged >60 y  
   Particular lesion412.93 (1.29-6.67)
   Whole skin793.67 (1.84-7.33)
Stratified by sex
 Ever used melanoma picture to look at skina  
   Yes751.88 (1.12-3.16)
   Yes591.00 (0.53-1.87)
 Patient states they received a physician skin examination for cancera
   Yes1722.06 (1.29-3.28)
   Yes861.37 (0.76-2.46)
 Type of skin examination performed by physician  
   Particular lesion/unsure581.28 (0.67-2.42)
   Whole skin1023.13 (1.82-5.39)
   Particular lesion/unsure351.05 (0.48-2.32)
   Whole skin451.52 (0.74-3.12)
Stratified by age and sex
 Patient states they received a physician skin examination for cancera
  Men aged ≤60 y  
   Yes721.27 (0.69-2.35)
  Women aged ≤60 y  
   Yes551.24 (0.62-2.49)
  Men aged >60 y  
   Yes1004.09 (1.88-8.89)
  Women aged >60 y  
   Yes311.40 (0.43-4.53)

The associations between PSE and thinner tumors also were much stronger among men who received a PSE for skin cancer, particularly a full-body PSE (OR, 3.13; 95% CI, 1.82-5.39) versus a partial PSE (OR, 1.28; 95% CI, 0.67-2.42), and among men who had a regular physician (OR, 2.58; 95% CI, 1.35-4.94). Women demonstrated no significant differences in the odds of having a thinner tumor by these variables.

Age and sex also had a strong interaction with 1 measure of SSE. Among patients aged >60 years, those who had used a melanoma picture to aid in SSE were more likely to have thinner tumors than those who did not (OR, 2.84; 95% CI, 1.46-5.49). Among patients aged ≤60 years, however, use of a melanoma picture aid and tumor thickness were unrelated. Men who had used a picture aid were more likely to have thinner tumors (OR, 1.88; 95% CI, 1.12-3.16), but no effect was observed among women. Age and sex were not strong effect modifiers of the association between the other SSE variables and tumor thickness. An assessment of interaction between age, sex, and certain variables (such as whether patients had consulted a physician in the last year) was not possible given sample size limitations.

Simultaneous Effect Modification by Age and Sex

The presence of simultaneous effect modification by age and sex required stratifying the associations between the 3 variables addressed above and tumor thickness into 4 age/sex groups (Table 4). Among men aged >60 years, thinner tumors were strongly associated with having received a PSE for cancer (OR, 4.09; 95% CI, 1.88-8.89). No significant associations between these variables and tumor depth were observed in younger men or in women of any age. These findings persisted when analyzed by histologic subtype and by the number of physician visits in the year before diagnosis.


Disproportionate melanoma incidence and mortality have been demonstrated in middle-aged and older men worldwide.20, 27, 28, 33 Compared with other demographic groups, older men tend to have thicker, more fatal melanomas10, 13, 15; more nodular melanoma11, 22; and participate less frequently in mass screening efforts34 or in deliberate SSE.35 Our study demonstrated that thinner tumors in men aged >60 years were strongly associated with PSE, and this result supports the rigorous education of physicians regarding the benefits of full-body skin examination during routine care and of older men regarding the value of routine SSE and at least annual PSE.

Our results verify known associations with thinner melanoma at diagnosis, including younger age, female gender, location on anatomic sites other than the head or neck, patient perception of having CAN (as opposed to clinician assessment), higher educational level, visible tumor pigmentation, non-nodular or desmoplastic histologic subtype, and physician detection. More important, these result demonstrate the detection of thinner tumors in patients who, in the year before diagnosis, 1) regularly examined their own skin, 2) consulted a physician at least once, and 3) received full-body PSE for cancer, although examination of even a particular lesion was beneficial for older adults. It is noteworthy that the advantage of PSE was limited to men aged >60 years, who had 4 times the odds of a thinner tumor than older men who did not receive a PSE.

Middle-aged and older men comprise nearly half of all melanoma deaths, but only 16% of US men in this subgroup report ever receiving a physician screening.36, 37 The importance of having regular access to medical care is highlighted by the finding that 33% of study patients who reported having a usual location for medical care and a regular physician for routine care received full-body PSE compared with only 11% of those who did not, although more frequent visits did not increase the odds of having a thinner tumor. SSE practices, as measured by the use of a melanoma picture aid and routine examination of some or all of the body, also significantly increased the odds of having a thinner melanoma, although this effect was limited to men and to older patients.

The reasons for the lack of impact of SSE and PSE on tumor thickness in men aged ≤60 years or in women regardless of age are not clear, but different risks for melanoma and health surveillance behaviors/practices may play a part. In general, older men have a higher risk of developing melanoma and also participate less effectively in their health matters than younger men and women. Therefore, older men who undergo regular PSE may receive the greatest benefit from this practice. Conversely, physicians of older men may be more cognizant of the higher incidence and mortality, less frequent SSE practices, and thicker histologic subtypes in this subgroup and, thus, may perform PSEs more vigilantly. Younger patients tend to self-detect melanoma more than older patients14 (57% vs 46%, respectively, in our study), typically have thinner tumors, and may not require the same frequency of PSE as older men.

The strongest evidence to date for the increased diagnosis of thinner tumors with physician skin screening was reported in a population-based case-control study of Queensland residents ages 20 to 75 years who had histologically confirmed, first primary, invasive melanoma.38 Whole-body clinical skin examination by a physician 3 years before diagnosis was associated with a 32% increased odds of having a T1 (≤1 mm) melanoma at diagnosis. In addition, a screening program that was initiated at the Lawrence Livermore National Laboratory and conducted between 1984 and 1996 demonstrated that increased melanoma education, self-examination, and opportunity for PSE resulted in a reduction in crude incidence of thicker melanomas, although the study population was highly motivated.39 Whereas the published data in support of the association of SSE on melanoma thickness26, 40 are less robust than those in support of PSE, our findings suggest that patient education and distribution of picture aids to increase SSE practices are justified. This is especially true among older men and is complementary to promoting PSE.

Our study inquired about PSE practices and health care use in the year before diagnosis, and many questions remain regarding why patients who had a PSE during that time had thinner tumors than those who did not, especially because only 19% of melanomas were detected by physicians. Without a proven survival benefit from PSE or melanoma screening itself, we can only speculate regarding the benefit of an earlier, nondiagnostic physician visit. It is possible that, at an earlier visit, the physician told the patient to follow a particular mole that subsequently changed or that the physician made note of a mole of concern that was not suspicious enough to warrant biopsy. Whether PSE overcomes patient delay in seeking medical attention for a suspicious lesion because of a lack of awareness or lower performance of SSE, or results in the detection of intrinsically less aggressive, slow-growing tumors (length bias) warrants further study.

Likewise, it is not clear why men aged >60 years, compared with younger men and all women in our study, reported nearly 1.5 to 3 more PSEs in the year before diagnosis. One study demonstrated that men aged ≥50 years with a personal history of skin cancer were more likely to seek skin screening compared with women and younger men, although women were more likely to seek medical attention because of a particular lesion concerning for skin cancer.41 However, low rates of prior skin cancer and melanoma in our study cohort (6% and 5%, respectively) do not support this explanation for increased PSEs in older men. Other possibilities include patient or physician recognition of higher melanoma risk, prompting of the physician by the patient or his spouse/partner in requesting a skin examination, or more suspicious but noncancerous lesions triggering routine PSE. Although the reasons for the stronger association of thinner tumors with PSE in older men do not appear to be a greater number of physician visits or differences in anatomic location, these issues, as well as a more detailed analysis of the role of physician detection in finding earlier nodular and desmoplastic subtypes, should be explored in future studies.

Study limitations include reliance on self-reports of health behaviors (SSE) and practices (health care use and PSE) and recall accuracy. Although all patients completed surveys within 3 months of diagnosis, potential over-reporting of health prevention practices is a possibility. Because the recall period for SSE and PSE practices extended up to 15 months (12 months before diagnosis coupled with a survey period of up to 3 months), there was likely some misclassification of these measures. In typical studies in which exposure is ascertained after diagnosis, differential misclassification of exposure (recall bias) can result in either exaggeration or dilution of the estimated association and is an important concern. However, in our study, all individuals had been diagnosed with melanoma, and the only difference was thickness, which makes it more likely that the main mechanism of misclassification would be nondifferential and, thus, would have underestimated the association. In addition, patient perception of who discovered the melanoma may vary, particularly if the patient notices a suspicious lesion but reports that the physician who performed the diagnostic biopsy discovered it. Strengths include the geographic diversity of the data collected, low rates of case refusal, short interval between diagnosis and completion of the survey, and contemporaneous clinician validation of nevus count and presence of clinical atypical nevi.

Because 72% of fatal melanomas are >1 mm thick at diagnosis,30 reducing the burden of melanoma >1 mm should be a goal to decrease mortality. To our knowledge, our study is the first to assess the effects of SSE and PSE by age and sex and to define the strong effect that physician screening has on the detection of thinner melanoma in older men. Our findings suggest that regular SSE and PSE potentially can surmount the higher mortality associated with this subgroup. Although both partial and whole-body examinations appear to be effective in older men, full-body PSE is preferable for the detection of melanoma in less visible but common sites (eg, back and scalp) that may be missed with the examination of a particular lesion.

Facilitating the performance of PSE in clinical practice requires overcoming both patient and physician barriers (eg, time constraints, competing comorbidities, and patient embarrassment).42 Routine PSE can be accomplished through enhanced physician training, physician awareness of higher risk for melanoma in the patient, and patient request for a full-body skin examination. Because so many individuals at risk of melanoma routinely consult physicians,43 improving physician training in full-body skin examination and increasing patient self-advocacy in requesting PSE are realistic strategies to improve early detection.


This study was supported by Merck and Company, Inc. Dr. Pollitt's work was supported in part by the Stanford University Medical Scholars Research Program during his medical school tenure.


The authors made no disclosures.