Epstein DS, Lange JR, Gruber SB, Mofid M, Koch SE. Is Physician Detection Associated With Thinner Melanomas?. JAMA. 1999;281(7):640-643. doi:10.1001/jama.281.7.640
Author Affiliations: Departments of Dermatology (Drs Epstein and Mofid) and Surgery (Dr Lange), Johns Hopkins School of Medicine, Baltimore, Md; the Department of Molecular Medicine and Genetics, University of Michigan, Ann Arbor (Dr Gruber); and the Department of Dermatology, Oregon Health Sciences University, Portland (Dr Koch).
Context In cutaneous melanoma, tumor depth remains the best
biologic predictor of patient survival. Detection of prognostically
favorable lesions may be associated with improved survival in patients
Objective To determine melanoma detection patterns and relate them
to tumor thickness.
Design Interview survey.
Setting and Patients All patients with newly detected primary
cutaneous melanoma at the Melanoma Center, Johns Hopkins Medical
Institutions, between June 1995 and June 1997.
Main Outcome Measure Tumor thickness grouped according to
Results Of the 102 patients (47 men, 55 women) in the study, the
majority of melanomas were self-detected (55%), followed by detection
by physician (24%), spouse (12%), and others (10%). Physicians were
more likely to detect thinner lesions than were patients who detected
their own melanomas (median thickness, 0.23 mm vs 0.9 mm;
P<.001). When grouped according to thickness, 11 (46%) of
24 physician-detected melanomas were in situ, vs only 8 (14%) of 56
patient-detected melanomas. Physician detection was associated with an
increase in the probability of detecting thinner (≤0.75 mm) melanomas
(relative risk, 4.2; 95% confidence interval, 1.4-11.1; P =
Conclusions Thinner melanomas are more likely to have been
detected by physicians. Increased awareness by all physicians may
result in greater detection of early melanomas.
The incidence and
mortality rates associated with primary cutaneous melanoma continue to
increase in the United States.1 In 1998, it is estimated
that 41,600 individuals were diagnosed as having invasive
melanoma, and that 7300 people died due to the disease.2
Efforts to reverse these trends have focused on primary prevention
through public awareness programs and sun exposure
education.3- 5 Yet, the current lifetime risk of an American
developing malignant melanoma is expected to reach 1 in 75 by the year
In the absence of regional and systemic metastasis,
Breslow7 tumor thickness remains the most important
prognostic indicator for predicting mortality and is strongly
associated with survival. Early detection and excision of thin
melanomas is associated with excellent patient survival.8
For secondary prevention to be successful, a tumor must be detected at
an early stage of development and surgically treated.9 In
this study, we examine differences in melanoma detection patterns and
their relationship to melanoma thickness.
All patients with primary cutaneous melanoma who presented to the
Melanoma Center at the Johns Hopkins Medical Institutions, Baltimore,
Md, between June 1995 and June 1997 were interviewed and considered
eligible for this study. Patients were examined and questioned by their
treating physicians (J.R.L., S.B.G., and S.E.K.) as part of their
clinical history. All patients were interviewed no more than 6 months
after their initial diagnoses.
Patients were excluded from the study population if evidence showed
systemic metastatic disease. These patients were excluded because signs
or symptoms of regional or systemic metastases could confound the
relationship between detection and thickness of the primary lesion. A
total of 102 patients met our criteria and formed our study group.
During their evaluation at the Melanoma Center, patients were
questioned specifically about who first detected their melanoma.
Patients were queried about the interval of time between initial
awareness of a suspicious pigmented lesion and the biopsy of the lesion
by a physician. They were also interviewed on personal characteristics
including age, education, and family or personal history of melanoma.
The majority of biopsy specimens were reviewed by dermatopathologists
at the Johns Hopkins
Dermatopathology Division for histopathologic confirmation
and measurement of tumor thickness.
Differences in personal characteristics (sex, detection pattern, and
history of melanoma) were analyzed by univariate χ2
analysis. When the analysis had low expected frequencies, a Fisher
exact test was performed. Characteristics with ordered variables (age,
education, time to biopsy) were studied using trend analysis. Personal
and lesion characteristics were grouped according to median tumor
thickness and compared using the Wilcoxon rank sum test. Analysis of
tumor thickness by age and time to biopsy was completed using the
nonparametric Kruskal-Wallis 1-way analysis of variance by ranks
procedure. Tumor thickness was evaluated using χ2 trend
analysis after grouping lesions into the following categories: 0.75 mm
or less, 0.76 to 1.50 mm, 1.51 to 4.00 mm, and greater than 4.00 mm.
Finally, the variables that were significant in univariate analysis
were tested for possible confounding and interactions using logistic
A total of 102 patients with primary cutaneous melanoma were analyzed,
including 47 men (46%) and 55 women (54%). Two thirds of the patients
had at least a college education, while 9% had not received a high
school diploma. A total of 16 patients (16%) had a positive family
history of melanoma, while 4 patients (4%) had a previous personal
history of primary cutaneous melanoma.
Patients were carefully questioned regarding who first noticed their
melanoma, and the results are tabulated according to detection pattern
in Table 1. Detection was classified
into the following categories: physicians, self-detection, spouses, and
other individuals. Patients detected 55% of the melanomas in this
study, while physicians found approximately 24% of the tumors. The
remainder were divided between spouses (12%) and others (10%). There
were significant differences in self-detection rates between men and
women (40% vs 67%; P<.01). Among men, 12 (26%) of the 47
melanomas were identified by a spouse. However, among women none of the
melanomas were identified by the spouse (0/55; P<.001).
Among men and women, similar percentages of melanomas were
physician-detected (26% vs 22%). Patients with higher education
(college or graduate school) were more likely to have melanomas
detected by a physician than those with less education (32% vs 9%;
P=.01). There were no identifiable differences
by age or for patients with a personal or a family history of melanoma.
Differences according to sex for self-detected lesions and according to
education for physician-detected lesions were not explained by any of
the other variables shown in Table 1.
There were identifiable trends based on the location of the melanoma.
Of the lesions that were in a visible location to the patient (head,
arms, legs, chest), 48 (70%) of 69 were self-detected; however,
patients found only 8 (24%) of the 33 melanomas on the back or
buttocks (P<.001). Using multiple logistic regression
analysis, there were no sex differences in self-detection rates based
on the location of the melanoma.
The median time to biopsy after initial detection of a suspicious
lesion was 3 months (range, <1 week to 10 years). Among the lesions
that were detected by physicians, a biopsy had been performed within a
month in 83% of the cases, while in 4% of the cases a biopsy had not
been performed before 6 months. For patient-detected lesions, a biopsy
had been performed within a month in 16% of the cases, while in 45%
of the cases a biopsy had not been performed earlier than 6 months
after having been detected. Thus, physician-detected melanomas were
more likely to undergo biopsy within a month than patient-detected
lesions (83% vs 16%; P<.001).
Tumor thickness was analyzed in relation to personal and lesion
characteristics. Older patient age was associated with thicker
melanomas, with a median thickness of 1.6 mm in patients older than 60
years, compared with 0.49 mm (P=.002) in
patients younger than 40 years. Patients with any college or graduate
school education had a median tumor thickness of 0.4 mm, while patients
with a high school education or less had a median tumor thickness of
1.3 mm (P<.001). In univariate analysis, a
delay in biopsy of a suspicious lesion was
associated with a thicker melanoma (P=.03).
Histopathologic data were obtained for each patient and are presented
in Table 2. Of the physician-detected
melanomas, 11 (46%) of 24 physician-found lesions were in situ, while
2 (8%) of 24 were greater than 1.50 mm in thickness. Of the
patient-detected lesions, only 8 (14%) of 56 were in situ, while 19
(34%) of 56 were greater than 1.50 mm in tumor thickness. Using
χ2 trend analysis, we found that physician detection was
associated with melanomas that were thinner than patient-detected
melanomas (P=.005), or all nonphysician
detected melanomas combined (P=.002).
Melanomas were also analyzed according to Breslow7
thickness data. Physician-found melanomas had the lowest median tumor
thickness (0.23 mm), compared with 0.9 mm (P<.001, Wilcoxon
rank sum test) for self-detected melanomas, 0.75 mm
(P=.03) for spouse-detected melanomas, and 0.9
mm (P<.001) for nonphysician groups combined.
To adjust for potential confounders, physician detection and tumor
thickness were analyzed in the logistic regression model shown in
Table 3. In unadjusted crude
analysis, physician detection of a lesion was associated with a greater
probability of discovering a thinner (≤0.75 mm) melanoma (relative
risk, 4.17; P=.01). In adjusted analysis,
physician detection (P=.04), age
(P=.004), and education
(P=.002) were associated with tumor thickness.
We assessed melanoma detection patterns to investigate the hypothesis
that thinner lesions were more likely to have been physician detected.
When compared with self-detected melanomas, the median tumor thickness
was significantly less for physician-detected melanomas. The majority
of prognostically favorable lesions8 (in situ and lesions
<0.75 mm) were found by physicians. Using trend analysis, our results
support the assertion that physicians identify melanomas at an earlier
stage than do the patients themselves or all nonphysician groups
Patients with at least a college education are more likely to
have their melanomas detected by a physician, possibly because these
patients seek medical advice more often. The majority of the easily
visible lesions were patient-detected. There was a positive association
between physician detection of a tumor and its immediate biopsy. In
almost 85% of physician-detected melanomas, biopsies had been
performed within a month, indicating that suspicious lesions were
identified and removed expeditiously.
Using univariate analysis, we demonstrated that factors associated with
a thinner tumor included younger patient age, higher educational level,
and decreased time to biopsy. In univariate and multivariate logistic
regression analysis, physician detection of a melanoma was associated
with thinner tumors. Physician detection was associated with a 400%
increase in the likelihood of discovering thinner melanomas (≤0.75 mm;
relative risk, 4.17). After controlling for interactions, the study
variables of age and education remained significant. Although time to
biopsy was no longer significant in multivariate analysis, each month
of delay in biopsy was associated with a 6% decrease in the likelihood
of discovering thin melanomas (≤0.75 mm).
One limitation of our article is the inherent difficulty of accurately
estimating and analyzing the time to biopsy. We divided time to biopsy
into what we believed were relevant categories; however, these may not
represent clinically important divisions. Because little is known about
the growth rate of melanoma, we cannot determine whether the
differences observed in the time to biopsy would
contribute significantly to tumor thickness. In addition, we do not
have long-term survival information for patients in our study. Since we
have shown that physician detection is associated with thinner
melanomas, we predict that this will result in improved
survival.10 Finally, unmeasured confounding and a small
sample size also may have had an important influence on this study.
Few studies have investigated differences in melanoma detection
patterns. A study by Koh et al11 of 216 patients designed
to look at sex differences in melanoma detection revealed that women
are more likely to discover their own lesions than men (66% vs 42%).
Our results are similar. Sixty-seven percent of women and 40% of men
detected their melanomas in our study group. Hersey et al12
noted that thick primary melanomas were most likely to occur in men.
However, we did not find a significant difference in melanoma thickness
according to the sex of the patient.
In conclusion, we have described melanoma detection patterns in a
screened population of patients at a tertiary referral center.
Melanomas that were physician-detected were thinner than melanomas that
were detected by nonphysicians. Future studies that include survival
data can investigate differences in outcome according to melanoma
detection patterns. However, routine skin examination for melanoma by
all physicians may result in identification of lesions early enough to
result in favorable outcomes.