- Email: [email protected]
How well do physicians recognize melanoma and other problem lesions?
Clinical and laboratorv studies How well do physicians recognize melanoma and other problem lesions? Barrie R. Cassileth, Ph.D., Wallace H. Clark, Jr., M.D., Edward J. Lusk, Ph.D., Beth E. Frederick, B.A., C. Jean Thompson, R.N., and William P. Walsh, M.A. Philadelphia, PA The alarming increase in the incidence of cutaneous malignant melanoma in the United States emphasizes the importance of its early detection and treatment. Early detection requires accurate clinical recognition of both malignant and precancerous skin lesions (dysplastic nevi). This study presents data on dermatologists' and nondermatologists' ability to diagnose skin lesions. A total of 105 nondermatologist physicians, from first-year residents to practicing physicians, and forty-eight dermatologists were asked to identify color slides or photographs of eleven cutaneous lesions, including malignant melanomas, dysplastic nevi, and innocuous lesions such as seborrheic keratoses and common moles. Diagnosis of cutaneous lesions was generally inaccurate among nondermatologists. Only 38% correctly identified four or more of the six melanomas as melanoma of any type, and 58% were unable to diagnose dysplastic nevi. Only 17% categorized their relevant training as excellent or good. Improved training in the diagnosis of skin lesions for practicing physicians and house staff is required if mortality from malignant melanoma is to be decreased in the United States. (J AM ACAD DERMATOL 14:555560, 1986.)
The importance of detecting melanoma in its early stages has been emphasized widely in the last few years, and diagnostic guides have been made available to physicians generally.I-6* We sought to determine the ability of dermatologists and nondermatologist physicians to identify various types of melanomas and to differentiate among innocuous skin lesions, precursor lesions (dysplastic nevi), and melanomas. From The Pigmented Lesion Study Group at the Universily of Pennsylvania Cancer Center. Supported in part by NIH Grant No. CA35529 and NIH Cancer Center Core Grant No. CA16520. Accepted for publication Nov. 12, 1985. Reprint requests to: Dr. Barrie R. Cassileth, Director, Psychosocial Programs, University of Pennsylvania Cancer Center, 7 Silverstein Pavilion, 3400 Spruce St., Philadelphia, PA 19104. *Friedman RJ, Rigel DS: Identifying early malignant melanomas: The ABeD's. The Melanoma Lt;tter 3:1-3, 1985.
This survey was conducted during the spring and fall of 1985, and as such it provides an indirect measure of the effectiveness of recent public and professional efforts geared to early identification and treatment of malignant melanoma. It also provides data on the variable difficulty associated with identification of particular types of lesions. SUBJECTS AND PROCEDURE Color slides (35 rom) of eleven representative skin lesions were projected on a screen for physicians attending an Internal Medicine Board Review course. At a different time, these same slides were projected for dermatologists from the Delaware Valley who were attending a meeting at the University of Pennsylvania. Finally, 5 X 7-inch color photographs of the same lesions were placed in plastic binders for third-year residents in clinical training at the University of Pennsylvania, who did not attend the Board Review course. The following types of lesions were represented:
555
556
Journal of the American Academy of Dennato!ogy
Cassileth et al
Fig. Fig. Fig. Fig. Fig. Fig.
1. Lesion I: early superficial spreading melanoma.
2. Lesion II: late superficial spreading melanoma. 3. Lesion III: nodular melanoma. 4. Lesion IV: dysplastic nevus. 5. Lesion V: seborrheic keratosis. 6. Lesion VI: normal mole.
melanomas (two early-stage superficial spreading melanomas, two late-stage superficial spreading melanomas, and two nodular melanomas); two dysplastic nevi; and innocuous lesions (one seborrheic keratosis and two
normal moles). Each of the eleven lesions was displayed by one close-up view and one overview slide (or photograph). Overviews were included to display skin type, size, and location of the lesion. Both representations
Volume 14 Number 4 April, 1986
of dysplastic nevi, as evident in their overviews, were from patients with the dysplastic nevus syndrome. Photographs were selected jointly by a dennatologist, a dermatopathologist, a medical oncologist specializing in melanoma, and nurse clinicians, all of whom are members of the University of Pennsylvania Pigmented Lesion Study Group. Photographs that clearly and accurately represent each type of lesion were sought and utilized. Figs. I to 6 display one of each of the six lesion types used in this survey. A response sheet given to each respondent requested identification of each lesion by use of a diagnosis key, which listed ten possible responses (nine diagnoses plus "don't know"). The diagnostic response options were as follows: nodular melanoma, dysplastic nevus, normal mole, basal cell carcinoma, squamous cell carcinoma, superficial spreading melanoma, seborrheic keratosis, congenital nevus, and blue nevus. For each of the six melanomas, responses were considered correct if either superficial spreading melanoma or nodular melanoma had been selected. That is, identification of superficial spreading melanoma as nodular melanoma or nodular melanoma as superficial spreading melanoma was counted as a correct response. The response sheet also requested respondents' judgments of the adequacy of their training in differential diagnosis of skin diseases, as well as respondents' age, sex, race, and professional status. Respondent identification was not elicited; this survey was completed anonymously. The physician sample is described in Table 1. Responses were computerized and analyzed by physicians' level of training and sex, as well as by physicians' self-assessment of the adequacy of their training. Analysis of variance with Scheffe's multiple comparison procedure was applied. Tests ofproportion were used to detennine statistical differences between classifications made by dennatologists and other physicians. Analyses focused on responses to malignant melanomas versus dysplastic nevi versus innocuous lesions, as well as on physicians' ability to differentiate between problem lesions in general (melanomas, dysplastic nevi, basal cell carcinoma, squamous cell carcinoma) and innocuous lesions in general (nonnal mole, seborrheic keratosis, congenital nevus, blue nevus). Responses were not analyzed by age because 'age. differences are tantamount to differences in training (Table I). Responses were not analyzable by race because 93% who indicated their race were white.
Melanoma and other problem lesions 557
Table I. Level of training and demographic characteristics of survey participants (N = 153) Training, demography
Participants: No., age, % by sex
UP third-year residents (No,) 32 Internal medicine board review course attendees (No.) (total: 73*) Medical student, PGY 1 or 8 2 42 PGY3 Fellow, practicing physician 23 Delaware Valley dennatologists (No.) (total: 48) Resident 11 Practicing physician 37 Age (yr): mean (± SD); range N ondennatologists 32 (± 6 yr); 25-56 Dermatologists 41 (± 12 yr); 26-71 Men/women (%) Nondermatologists 72/28 Dermatologists 80/20 PGY: Postgraduate year; UP: University of Pennsylvania. *According to Board Review course administrators, approximately 90% of attendees were from hospitals or practices in the Delaware Valley other than the University of Pennsylvania.
their training in differential diagnosis of skin diseases and the number of lesions identified correctly in this survey. However, the vast majority either evaluated their training as fair (56%) or indicated that they had "no training" (35%). Only 9% of nondermatologists categorized their training as good or excellent. All dermatologists categorized their training in differential diagnosis of cutaneous lesions as excellent or good. Level of training did not influence nondermatologist physicians' ability to correctly identify lesions or to distinguish melanoma from other lesions at p < 0.05. There were no substantive response differences between men and women. Therefore data are reported for the total sample of 105 nonderrnatologists as a group and for the total sample of forty-eight dermatologists as a group.
RESULTS
Ability to identify melanoma and dysplastic nevi
There was no relationship between nondermatologist physicians' perception of the adequacy of
Whereas only 12% of nondermatologists correctly identified at least five of six melanomas as
558
Journal of the
Cassileth et al
American Academy of
Dermatology
Table m. Percent of physicians (nondermatologists and dermatologists) correctly recognizing melanoma and dysplastic nevi as problem lesions*
Table II. Percent of physicians (nondermatologists and dermatologists) correctly identifying melanoma and dysplastic nevi No. of lesions correctly identifted
% nondermatologists
eN
= 105)
Melanoma* 6 of 6 5 of 6 4 of 6 3 of 6 2 of 6
22
Dysplastic nevus 2 of 2
11
1 or 0 of 6
1 of 2
o of2
2
10
26 33
%
dermatologists
eN = 48)
27
42 23
8
7
31 58
71
25 4
SSM: Superficial spreading melanoma. *Melanomas displayed: two early 88Ms, two late 88Ms, two nodular melanomas. Response to each was counted as correct if it was called SSM or nodular melanoma.
such, 69% of dermatologists correctly identified at least five of the six melanomas (Table II). The majority of nondermatologists (62%) were able to correctly identify only three or fewer of the melanomas pictured. In 16% of cases, nondermatologists identified melanomas as innnocuous lesions, whereas 4% of dermatologists' choices involved identifying a melanoma as innocuous (difference significant at p < 0.05). Most nondermatologist physicians in this sample (58%) failed to diagnose either of the two examples of dysplastic nevi. Only 11 % correctly identified both examples, and 31 % diagnosed one representation of dysplastic nevi correctly. The majority of dermatologists (71 %) recognized both examples of dysplastic nevi; only 4% identified neither example. Recognition of problem lesions On the assumption that precise differential diagnosis is less important than the ability to recognize and act on suspicious or problem lesions, analyses were conducted to assess physicians' ability to identify a problem lesion as such, regardless of the specific subtype that they attributed to it.
No. of problem lesions correctly recognized
% nondermatologists
8 of 8 7 of 8 6 of 8 5 of 8 4 of 8 3 of 8 2 or fewer of 8
6 22
(N
= 105) 24
25 16 5
%
dermatologists (N
= 48)
52
33 10 3 2
2
*Problem lesions displayed: two early SSMs, two late SSMs, two nodular melanomas, two dysplastic nevi. Responses to each were counted as correct if they were called any of the foregoing or basal cell or squamous cell carcinoma.
Responses to problem lesions displayed (two early superficial spreading melanomas, two late superficial spreading melanomas, two nodular melanomas, and two dysplastic nevi) were counted as correct if they were recognized as any type of melanoma, as dysplastic nevi, or as basal or squamous cell carcinoma. As indicated in Table III, only 28% of nondermatologists, but 85% of dermatologists, identified at least seven of the eight problem lesions as such. How respondents identify melanoma and dysplastic nevi For this analysis, responses were averaged for each pair of same-subtype lesions: early-stage superficial spreading melanomas, late-stage superficial spreading melanomas, nodular melanomas, and dysplastic nevi. As displayed in Table IV, early superficial spreading melanoma was identified as superficial spreading melanoma by 40% of nondermatologists and 72% of dermatologists. Early superficial spreading melanoma was mistaken for dysplastic nevi, nodular melanomas, or basal cell or squamous cell carcinoma by 20% of all respondents, but the much more serious error of identifying early superficial spreading melanoma as innocuous was made by 23% of all respondents: 28% of nondermatologists and 11% of dermatologists. However, the majority of dermatologists
Volume 14 Number 4
Melanoma and other problem lesions
April, 1986
559
Table IV. Physicians' identification of melanoma and dysplastic nevi* % selecting response
Lesion
Early SSM
Late SSM
Nodular melanoma
Dysplastic nevus
Response
Nondermatologists (N 105)
=
SSM Dysplastic nevus Other problem lesiont Innocuous lesion:j: Don't know SSM Dysplastic nevus Other problem lesion Innocuous lesion Don't know Nodular Dysplastic nevus Other problem lesion Innocuous lesion Don't know Dysplastic nevus Other problem lesion Innocuous lesion Don't know
I
Dermatologists (N 48)
=
40 12 9
72
11 18 8
1 40
28
10 6 11
61
58
28
2 75 2 15 2 6 85
8 5 34 8 35 11 12
27
32 13
5 10
SSM: Superficial spreading melanoma. *Dermatologists were significantly superior to nondermatologists in each lesion category (test of proportion, p < 0.005). tOther problem lesion responses: nodular melanoma, superficial spreading melanoma, basal cell carcinoma, squamous cell carcinoma. :j:lnnocuous lesion responses: normal mole, seborrheic keratosis, congenital nevus, blue nevus.
(88%) and other physicians (61%) did identify early-stage superficial spreading melanoma as a problem lesion, presumably to be pursued with biopsy. As Table IV also indicates, late-stage superficial spreading melanoma was identified as superficial spreading melanoma by only 18% of nondermatologists and 40% of dennatologists. However, it was recognized as a problem lesion by most respondents (87% and 98%, respectively). Latestage superficial spreading melanoma was misidentified as innocuous by 8% of nondermatologists and by no dermatologists. Most dennatologists (75%) and a smaller percentage of other physicians (34%) identified nodular melanomas as such, but 2% and 11%, respectively, viewed these lesions as innocuous. The sharpest distinction between dermatologists and nondennatologists occurred with regard to dysplastic nevi. Whereas 85% of dermatologists
correctly identified these lesions, only 28% of other physicians were able to do so. Dysplastic nevi were termed innocuous by 10% of dermatologists and by 32% of other physicians. Dermatologists displayed significantly superior ability to correctly identify all lesions (p < 0.005).
DISCUSSION Physicians participating in this survey included third-year clinical residents at a major urban medical center, residents, fellows, and practicing dermatologists and other physicians from the Delaware Valley. The results of this survey are generalizable to analogous settings and are likely to be representative of the skills of similarly trained physicians generally. This study indicates that nondennatologist physicians lack a sufficient level of ability to correctly diagnose skin lesions. Only 12% identified at least
560
Journal of the American Academy of Dermatology
Cassileth et al
five of six examples of melanoma as such, regardless of subtype attributed, and only 11 % recognized both examples of dysplastic nevi. On the average, 25% of nondermatologists chose "don't know" or "innocuous lesion" in response to malignant lesions. Dermatologists, as would be expected, generally can distinguish problematic from innocuous lesions, and most are able to identify dysplastic nevi. These strikingly poor results for nondermatologists point to a failure of the educational process in this critically important area. The inability of physicians to identify lesions that are curable in their early stages and typically fatal when advanced l -3 has major clinical implications. Similarly, nondermatologists displayed poor ability to recognize precursor lesions (dysplastic nevi). Recent documentation of the familial dysplastic nevus syndrome*·7 has established that patients with this syndrome are readily identifiable and their lesions treatable prior to invasive, malignant development. Other studies have addressed physicians' ability to recognize various skin lesions. Ramsey and Fox8 assessed physicians' identification of twenty common dermatoses, including melanoma, melanocytic nevi, and seborrheic keratoses. Wagner et al9 investigated the ability of medical students, residents, and physicians' assistants to identify malignant and benign cutaneous lesions. Although these two studies and the study reported here differ in methodology and sample populations, results in each case provide evidence that primary care physicians and others lacking training in dermatologic differential diagnoses are deficient in their ability to recognize potentially serious cutaneous lesions. Despite widespread dissemination of information about the increasing incidence of melanoma and the importance of early detection and treatment,1'7 internists and other physicians, whom pa*Zehngebot LM: Malignant melanoma and the dysplastic nevus syndrome: Developmental biology and clinical management. Curr Concepts Oncology 7:3-6. 1985.
tients are likely to see more often than dermatologists, reveal a profound lack of knowledge about melanoma and its precursor lesions and show an extremely limited capacity to accurately diagnose cutaneous lesions. The recognition of melanoma and other problem lesions requires visual skills that can be learned by the general public, as indicated by the Australian experience. lOoll The availability of highquality color atlases (such as the atlas by Mihm et a14) renders self-education a realistic possibility for physicians. Improvements in the survival rates of patients with melanoma clearly require improved training of nondermatologist physicians, training that can be accomplished in a costeffective fashion. REFERENCES 1. Sober AJ, Fitzpatrick TB, Mihrn Me, et al: Early recognition of cutaneous melanoma. JAMA 242:2795-2799, 1979. 2. Sober AJ, Day CL, Kopf AW, Fitzpatrick TB: Detection of "thin" primary melanomas. CA 33:161-164,1983. 3. Ackerman AB: Clinical diagnosis of malignant melanoma in situ, in Ackerman AB, editor: The pathology of malignant melanoma. New York, 1981, Masson Publishing USA, Inc., pp. 57-60. 4. Mihm MC. Fitzpatrick TB, Lane-Brown MM. et al: Early detection of primary cutaneous melanoma: A color atlas .. N Engl J Med 289:989-996, 1973. 5. Wick MM, Sober AJ , Fitzpatrick TB, et al: Clinical characteristics of early cutaneous melanoma. Cancer 45:2684-2686, 1980. 6. Fitzpatrick TB, Rhodes AR, Sober AJ: Prevention of melanoma by recognition of its precursors. N Engl J Med 312:115-116, 1985. 7. Greene MH, Clark WH, Thcker MA, et al: Acquired precursors of cutaneous malignant melanoma: The familial dysplastic nevus syndrome. N Engl J Med 312:9197, 1985. 8. Ramsey DL, Fox AB: The ability of primary care physicians to recognize the common dermatoses. Arch Dermatol117:620-622, 1981. 9. Wagner RF, Wagner 0, Tomich JM, et al: Diagnoses of skin diseases: Dermatologists vs. nondermatologists. J Dermatol Surg Oncol 11:476-479, 1985. 10. Smith T: The Queensland melanoma project: An exercise in health education. Br Med J 1:253-254, 1979. 11. Holman COl, James lR, Gattey PH, et al: An analysis of trends in mortality from malignant melanoma of the skin in Australia. Int J Cancer 26:703-709, 1980.
The importance of detecting melanoma in its early stages has been emphasized widely in the last few years, and diagnostic guides have been made available to physicians generally.I-6* We sought to determine the ability of dermatologists and nondermatologist physicians to identify various types of melanomas and to differentiate among innocuous skin lesions, precursor lesions (dysplastic nevi), and melanomas. From The Pigmented Lesion Study Group at the Universily of Pennsylvania Cancer Center. Supported in part by NIH Grant No. CA35529 and NIH Cancer Center Core Grant No. CA16520. Accepted for publication Nov. 12, 1985. Reprint requests to: Dr. Barrie R. Cassileth, Director, Psychosocial Programs, University of Pennsylvania Cancer Center, 7 Silverstein Pavilion, 3400 Spruce St., Philadelphia, PA 19104. *Friedman RJ, Rigel DS: Identifying early malignant melanomas: The ABeD's. The Melanoma Lt;tter 3:1-3, 1985.
This survey was conducted during the spring and fall of 1985, and as such it provides an indirect measure of the effectiveness of recent public and professional efforts geared to early identification and treatment of malignant melanoma. It also provides data on the variable difficulty associated with identification of particular types of lesions. SUBJECTS AND PROCEDURE Color slides (35 rom) of eleven representative skin lesions were projected on a screen for physicians attending an Internal Medicine Board Review course. At a different time, these same slides were projected for dermatologists from the Delaware Valley who were attending a meeting at the University of Pennsylvania. Finally, 5 X 7-inch color photographs of the same lesions were placed in plastic binders for third-year residents in clinical training at the University of Pennsylvania, who did not attend the Board Review course. The following types of lesions were represented:
555
556
Journal of the American Academy of Dennato!ogy
Cassileth et al
Fig. Fig. Fig. Fig. Fig. Fig.
1. Lesion I: early superficial spreading melanoma.
2. Lesion II: late superficial spreading melanoma. 3. Lesion III: nodular melanoma. 4. Lesion IV: dysplastic nevus. 5. Lesion V: seborrheic keratosis. 6. Lesion VI: normal mole.
melanomas (two early-stage superficial spreading melanomas, two late-stage superficial spreading melanomas, and two nodular melanomas); two dysplastic nevi; and innocuous lesions (one seborrheic keratosis and two
normal moles). Each of the eleven lesions was displayed by one close-up view and one overview slide (or photograph). Overviews were included to display skin type, size, and location of the lesion. Both representations
Volume 14 Number 4 April, 1986
of dysplastic nevi, as evident in their overviews, were from patients with the dysplastic nevus syndrome. Photographs were selected jointly by a dennatologist, a dermatopathologist, a medical oncologist specializing in melanoma, and nurse clinicians, all of whom are members of the University of Pennsylvania Pigmented Lesion Study Group. Photographs that clearly and accurately represent each type of lesion were sought and utilized. Figs. I to 6 display one of each of the six lesion types used in this survey. A response sheet given to each respondent requested identification of each lesion by use of a diagnosis key, which listed ten possible responses (nine diagnoses plus "don't know"). The diagnostic response options were as follows: nodular melanoma, dysplastic nevus, normal mole, basal cell carcinoma, squamous cell carcinoma, superficial spreading melanoma, seborrheic keratosis, congenital nevus, and blue nevus. For each of the six melanomas, responses were considered correct if either superficial spreading melanoma or nodular melanoma had been selected. That is, identification of superficial spreading melanoma as nodular melanoma or nodular melanoma as superficial spreading melanoma was counted as a correct response. The response sheet also requested respondents' judgments of the adequacy of their training in differential diagnosis of skin diseases, as well as respondents' age, sex, race, and professional status. Respondent identification was not elicited; this survey was completed anonymously. The physician sample is described in Table 1. Responses were computerized and analyzed by physicians' level of training and sex, as well as by physicians' self-assessment of the adequacy of their training. Analysis of variance with Scheffe's multiple comparison procedure was applied. Tests ofproportion were used to detennine statistical differences between classifications made by dennatologists and other physicians. Analyses focused on responses to malignant melanomas versus dysplastic nevi versus innocuous lesions, as well as on physicians' ability to differentiate between problem lesions in general (melanomas, dysplastic nevi, basal cell carcinoma, squamous cell carcinoma) and innocuous lesions in general (nonnal mole, seborrheic keratosis, congenital nevus, blue nevus). Responses were not analyzed by age because 'age. differences are tantamount to differences in training (Table I). Responses were not analyzable by race because 93% who indicated their race were white.
Melanoma and other problem lesions 557
Table I. Level of training and demographic characteristics of survey participants (N = 153) Training, demography
Participants: No., age, % by sex
UP third-year residents (No,) 32 Internal medicine board review course attendees (No.) (total: 73*) Medical student, PGY 1 or 8 2 42 PGY3 Fellow, practicing physician 23 Delaware Valley dennatologists (No.) (total: 48) Resident 11 Practicing physician 37 Age (yr): mean (± SD); range N ondennatologists 32 (± 6 yr); 25-56 Dermatologists 41 (± 12 yr); 26-71 Men/women (%) Nondermatologists 72/28 Dermatologists 80/20 PGY: Postgraduate year; UP: University of Pennsylvania. *According to Board Review course administrators, approximately 90% of attendees were from hospitals or practices in the Delaware Valley other than the University of Pennsylvania.
their training in differential diagnosis of skin diseases and the number of lesions identified correctly in this survey. However, the vast majority either evaluated their training as fair (56%) or indicated that they had "no training" (35%). Only 9% of nondermatologists categorized their training as good or excellent. All dermatologists categorized their training in differential diagnosis of cutaneous lesions as excellent or good. Level of training did not influence nondermatologist physicians' ability to correctly identify lesions or to distinguish melanoma from other lesions at p < 0.05. There were no substantive response differences between men and women. Therefore data are reported for the total sample of 105 nonderrnatologists as a group and for the total sample of forty-eight dermatologists as a group.
RESULTS
Ability to identify melanoma and dysplastic nevi
There was no relationship between nondermatologist physicians' perception of the adequacy of
Whereas only 12% of nondermatologists correctly identified at least five of six melanomas as
558
Journal of the
Cassileth et al
American Academy of
Dermatology
Table m. Percent of physicians (nondermatologists and dermatologists) correctly recognizing melanoma and dysplastic nevi as problem lesions*
Table II. Percent of physicians (nondermatologists and dermatologists) correctly identifying melanoma and dysplastic nevi No. of lesions correctly identifted
% nondermatologists
eN
= 105)
Melanoma* 6 of 6 5 of 6 4 of 6 3 of 6 2 of 6
22
Dysplastic nevus 2 of 2
11
1 or 0 of 6
1 of 2
o of2
2
10
26 33
%
dermatologists
eN = 48)
27
42 23
8
7
31 58
71
25 4
SSM: Superficial spreading melanoma. *Melanomas displayed: two early 88Ms, two late 88Ms, two nodular melanomas. Response to each was counted as correct if it was called SSM or nodular melanoma.
such, 69% of dermatologists correctly identified at least five of the six melanomas (Table II). The majority of nondermatologists (62%) were able to correctly identify only three or fewer of the melanomas pictured. In 16% of cases, nondermatologists identified melanomas as innnocuous lesions, whereas 4% of dermatologists' choices involved identifying a melanoma as innocuous (difference significant at p < 0.05). Most nondermatologist physicians in this sample (58%) failed to diagnose either of the two examples of dysplastic nevi. Only 11 % correctly identified both examples, and 31 % diagnosed one representation of dysplastic nevi correctly. The majority of dermatologists (71 %) recognized both examples of dysplastic nevi; only 4% identified neither example. Recognition of problem lesions On the assumption that precise differential diagnosis is less important than the ability to recognize and act on suspicious or problem lesions, analyses were conducted to assess physicians' ability to identify a problem lesion as such, regardless of the specific subtype that they attributed to it.
No. of problem lesions correctly recognized
% nondermatologists
8 of 8 7 of 8 6 of 8 5 of 8 4 of 8 3 of 8 2 or fewer of 8
6 22
(N
= 105) 24
25 16 5
%
dermatologists (N
= 48)
52
33 10 3 2
2
*Problem lesions displayed: two early SSMs, two late SSMs, two nodular melanomas, two dysplastic nevi. Responses to each were counted as correct if they were called any of the foregoing or basal cell or squamous cell carcinoma.
Responses to problem lesions displayed (two early superficial spreading melanomas, two late superficial spreading melanomas, two nodular melanomas, and two dysplastic nevi) were counted as correct if they were recognized as any type of melanoma, as dysplastic nevi, or as basal or squamous cell carcinoma. As indicated in Table III, only 28% of nondermatologists, but 85% of dermatologists, identified at least seven of the eight problem lesions as such. How respondents identify melanoma and dysplastic nevi For this analysis, responses were averaged for each pair of same-subtype lesions: early-stage superficial spreading melanomas, late-stage superficial spreading melanomas, nodular melanomas, and dysplastic nevi. As displayed in Table IV, early superficial spreading melanoma was identified as superficial spreading melanoma by 40% of nondermatologists and 72% of dermatologists. Early superficial spreading melanoma was mistaken for dysplastic nevi, nodular melanomas, or basal cell or squamous cell carcinoma by 20% of all respondents, but the much more serious error of identifying early superficial spreading melanoma as innocuous was made by 23% of all respondents: 28% of nondermatologists and 11% of dermatologists. However, the majority of dermatologists
Volume 14 Number 4
Melanoma and other problem lesions
April, 1986
559
Table IV. Physicians' identification of melanoma and dysplastic nevi* % selecting response
Lesion
Early SSM
Late SSM
Nodular melanoma
Dysplastic nevus
Response
Nondermatologists (N 105)
=
SSM Dysplastic nevus Other problem lesiont Innocuous lesion:j: Don't know SSM Dysplastic nevus Other problem lesion Innocuous lesion Don't know Nodular Dysplastic nevus Other problem lesion Innocuous lesion Don't know Dysplastic nevus Other problem lesion Innocuous lesion Don't know
I
Dermatologists (N 48)
=
40 12 9
72
11 18 8
1 40
28
10 6 11
61
58
28
2 75 2 15 2 6 85
8 5 34 8 35 11 12
27
32 13
5 10
SSM: Superficial spreading melanoma. *Dermatologists were significantly superior to nondermatologists in each lesion category (test of proportion, p < 0.005). tOther problem lesion responses: nodular melanoma, superficial spreading melanoma, basal cell carcinoma, squamous cell carcinoma. :j:lnnocuous lesion responses: normal mole, seborrheic keratosis, congenital nevus, blue nevus.
(88%) and other physicians (61%) did identify early-stage superficial spreading melanoma as a problem lesion, presumably to be pursued with biopsy. As Table IV also indicates, late-stage superficial spreading melanoma was identified as superficial spreading melanoma by only 18% of nondermatologists and 40% of dennatologists. However, it was recognized as a problem lesion by most respondents (87% and 98%, respectively). Latestage superficial spreading melanoma was misidentified as innocuous by 8% of nondermatologists and by no dermatologists. Most dennatologists (75%) and a smaller percentage of other physicians (34%) identified nodular melanomas as such, but 2% and 11%, respectively, viewed these lesions as innocuous. The sharpest distinction between dermatologists and nondennatologists occurred with regard to dysplastic nevi. Whereas 85% of dermatologists
correctly identified these lesions, only 28% of other physicians were able to do so. Dysplastic nevi were termed innocuous by 10% of dermatologists and by 32% of other physicians. Dermatologists displayed significantly superior ability to correctly identify all lesions (p < 0.005).
DISCUSSION Physicians participating in this survey included third-year clinical residents at a major urban medical center, residents, fellows, and practicing dermatologists and other physicians from the Delaware Valley. The results of this survey are generalizable to analogous settings and are likely to be representative of the skills of similarly trained physicians generally. This study indicates that nondennatologist physicians lack a sufficient level of ability to correctly diagnose skin lesions. Only 12% identified at least
560
Journal of the American Academy of Dermatology
Cassileth et al
five of six examples of melanoma as such, regardless of subtype attributed, and only 11 % recognized both examples of dysplastic nevi. On the average, 25% of nondermatologists chose "don't know" or "innocuous lesion" in response to malignant lesions. Dermatologists, as would be expected, generally can distinguish problematic from innocuous lesions, and most are able to identify dysplastic nevi. These strikingly poor results for nondermatologists point to a failure of the educational process in this critically important area. The inability of physicians to identify lesions that are curable in their early stages and typically fatal when advanced l -3 has major clinical implications. Similarly, nondermatologists displayed poor ability to recognize precursor lesions (dysplastic nevi). Recent documentation of the familial dysplastic nevus syndrome*·7 has established that patients with this syndrome are readily identifiable and their lesions treatable prior to invasive, malignant development. Other studies have addressed physicians' ability to recognize various skin lesions. Ramsey and Fox8 assessed physicians' identification of twenty common dermatoses, including melanoma, melanocytic nevi, and seborrheic keratoses. Wagner et al9 investigated the ability of medical students, residents, and physicians' assistants to identify malignant and benign cutaneous lesions. Although these two studies and the study reported here differ in methodology and sample populations, results in each case provide evidence that primary care physicians and others lacking training in dermatologic differential diagnoses are deficient in their ability to recognize potentially serious cutaneous lesions. Despite widespread dissemination of information about the increasing incidence of melanoma and the importance of early detection and treatment,1'7 internists and other physicians, whom pa*Zehngebot LM: Malignant melanoma and the dysplastic nevus syndrome: Developmental biology and clinical management. Curr Concepts Oncology 7:3-6. 1985.
tients are likely to see more often than dermatologists, reveal a profound lack of knowledge about melanoma and its precursor lesions and show an extremely limited capacity to accurately diagnose cutaneous lesions. The recognition of melanoma and other problem lesions requires visual skills that can be learned by the general public, as indicated by the Australian experience. lOoll The availability of highquality color atlases (such as the atlas by Mihm et a14) renders self-education a realistic possibility for physicians. Improvements in the survival rates of patients with melanoma clearly require improved training of nondermatologist physicians, training that can be accomplished in a costeffective fashion. REFERENCES 1. Sober AJ, Fitzpatrick TB, Mihrn Me, et al: Early recognition of cutaneous melanoma. JAMA 242:2795-2799, 1979. 2. Sober AJ, Day CL, Kopf AW, Fitzpatrick TB: Detection of "thin" primary melanomas. CA 33:161-164,1983. 3. Ackerman AB: Clinical diagnosis of malignant melanoma in situ, in Ackerman AB, editor: The pathology of malignant melanoma. New York, 1981, Masson Publishing USA, Inc., pp. 57-60. 4. Mihm MC. Fitzpatrick TB, Lane-Brown MM. et al: Early detection of primary cutaneous melanoma: A color atlas .. N Engl J Med 289:989-996, 1973. 5. Wick MM, Sober AJ , Fitzpatrick TB, et al: Clinical characteristics of early cutaneous melanoma. Cancer 45:2684-2686, 1980. 6. Fitzpatrick TB, Rhodes AR, Sober AJ: Prevention of melanoma by recognition of its precursors. N Engl J Med 312:115-116, 1985. 7. Greene MH, Clark WH, Thcker MA, et al: Acquired precursors of cutaneous malignant melanoma: The familial dysplastic nevus syndrome. N Engl J Med 312:9197, 1985. 8. Ramsey DL, Fox AB: The ability of primary care physicians to recognize the common dermatoses. Arch Dermatol117:620-622, 1981. 9. Wagner RF, Wagner 0, Tomich JM, et al: Diagnoses of skin diseases: Dermatologists vs. nondermatologists. J Dermatol Surg Oncol 11:476-479, 1985. 10. Smith T: The Queensland melanoma project: An exercise in health education. Br Med J 1:253-254, 1979. 11. Holman COl, James lR, Gattey PH, et al: An analysis of trends in mortality from malignant melanoma of the skin in Australia. Int J Cancer 26:703-709, 1980.