RESEARCH LETTERS

Perception of ABC (asymmetry, borders, We designed a study to separately analyze judg-

and color) parameters in the screening
for melanoma: Model exercise with
experienced dermatologists
To the Editor: The ABCD rule (A for asymmetric
shape, B for irregular borders, C for variation in
color, and D for diameter [6 mm) has been pro-
posed as a reliable and affordable screening method
in melanoma campaigns.1,2 However, little is known
about the ability of examiners to agree on ABC
parameters when faced with pigmentary lesions.3
ments on ABC features, avoiding, as much as
possible, the influence of the potentially confound-
ing effect of the observer’s overall diagnostic
impression.
Twelve photographs of melanocytic nevi and 11
photographs of melanomas with Breslow thickness
#1.5 mm were randomly selected from our file of
excised lesions. For judgment regarding symmetry
and borders, the profiles of the selected lesions
were traced, and the resultant silhouette was filled
in with black. To minimize the effect of dimension

Fig 1. Example of test materials prepared for the ABC (asymmetry, border, color) perception
exercise. A, Symmetry and borders were evaluated using a black colored-in blot representing
the shape of the segmented lesion. B, Judgment of color was performed using an ovoid
depiction of the original segmented lesion. C, Original clinical photographs were used for the
final classification of lesions as melanocytic nevi or melanoma.

996 MAY 2018 J AM ACAD DERMATOL

J AM ACAD DERMATOL
VOLUME 78, NUMBER 5
Table I. Interobserver agreement on ABC param-
eters and clinical diagnoses
Kappa values
Session Symmetry Borders Color Clinical diagnosis*
1st 0.21 0.52 0.36
2nd 0.32 0.62 0.32
ABC, Asymmetry, boarders, color.
*Assessed in third/final exercise.
on the judgment, all silhouettes were given the
same area. For judgment regarding color, the
borders of the original photographs were con-
cealed and the lesions were given a regular circular
border (Fig 1).
We organized 2 evaluation sessions involving 7
experienced dermatologists 3 weeks apart from
each other. We projected slides obtained from the
test material and asked participants to indepen-
dently express their judgment on symmetry, border,
and color with dichotomous answers (eg, color
mainly homogeneous or nonhomogeneous). We
also organized a third session during which the
original clinical photographs were shown and
participants were asked to classify lesions as mel-
anocytic nevi or melanoma. Cohen kappa and
Fleiss kappa4 were used to assess intraobserver
and interobserver agreement with multiple raters,
respectively. The relation between the scores on
ABC parameters and the final diagnosis reached by
dermatologists (ie, nevus vs melanoma) was exam-
ined by logistic regression analysis and expressed
as odds ratios (ORs) with their 95% confidence
intervals (CI).
Variations were observed for intraobserver
agreement on ABC parameters. The judgments on
borders had the highest values of agreement, with
Cohen kappa ranging 0.64-1.00. Less satisfactory
was the intraobserver agreement for judgment on
symmetry and color, Cohen kappa ranging 0.38-
0.81 for symmetry and 0.26-0.72 for color. Table I
presents data on interobserver agreement. The
agreement was moderate to substantial for judg-
ment on borders and for the final diagnosis made
by the dermatologists. On the basis of logistic
regression analysis, a diagnosis of melanoma was
significantly associated with judgment of irregular
borders (OR 2.4, 95% CI 1.1-5.4) and haphazard
color (OR 2.6, 95% CI 1.4-4.8) but not with
judgment of asymmetric lesion (OR 1.3, 95% CI
0.6-2.9).
Our data point to the existence of variability
in the interpretation of ABC parameters. We
Research Letters 997
documented high interobserver agreement on the
final diagnosis in the face of unsatisfactory intra-
observer and interobserver agreement on specific
features. Even if not a direct proof of the cognitive
mechanism involved, our data support the concept
0.63
of an automatic pattern recognition modality in the
diagnosis of pigmentary lesions.5 They also point
to the need for a more standardized terminology to
describe the clinical features of pigmented
lesions.6
We wish to thank Gillian Jarvis for editorial assistance.
Luigi Naldi, MD,a Giuseppe Falgheri, MD,b
Antonello De Bitonto, MD,c Anna Di Landro,
MD,d Luigi Foiadelli, MD,b Gianlorenzo Imberti,
MD,a Francesco Riva, MD,b and Simone
Cazzaniga, DMathd,e
From the Department of Dermatology, Azienda
Socio-Sanitaria Territoriale Ospedale Papa Gio-
vanni XXIII, Bergamo, Italya; Agenzia per la
Tutela della Salute, Bergamo, Italyb; Policlinico
S. Marco, Zingonia, Bergamo, Italyc; Centro
Studi Gruppo Italiano Studi Epidemiologici in
Dermatologia, Bergamo, Italyd; Department of
Dermatology, Inselspital University Hospital,
Bern, Switzerlande
Funding sources: Supported by the Italian National
Council on Research (no. 92.02221.PF39).
Conflicts of interest: None disclosed.
Reprints not available from the authors.
Correspondence to: Luigi Naldi, MD, Centro Studi
GISED, Via Garibaldi 13/15 - 24122 Bergamo,
Italy
E-mail: luigi.naldi@gised.it
REFERENCES
1. Abbasi NR, Shaw HM, Rigel DS, et al. Early diagnosis of
cutaneous melanoma: revisiting the ABCD criteria. JAMA. 2004;
292(22):2771-2776.
2. American Academy of Dermatology Ad Hoc Task Force for the
ABCDEs of Melanoma, Tsao H, Olazagasti JM, Cordoro KM,
et al. Early detection of melanoma: reviewing the ABCDEs. J
Am Acad Dermatol. 2015;72(4):717-723.
3. Aldridge RB, Zanotto M, Ballerini L, et al. Novice identification
of melanoma: not quite as straightforward as the ABCDs. Acta
Derm Venereol. 2011;91(2):125-230.
4. Fleiss JL. Statistical Methods for Rates and Proportions. 2nd ed.
New York, NY: John Wiley ed; 1981:38-46.
5. Girardi S, Gaudy C, Gouvernet J, et al. Superiority of a cognitive
education with photographs over ABCD criteria in the education
of the general population to the early detection of melanoma: a
randomized study. Int J Cancer. 2006;118(9):2276-2280.
998 Research Letters
6. Flood KS, Martin GJ, Robinson JK. Uniform nomenclature to
describe clinical features of pigmented lesions. JAMA Derma-
tol. 2017;153:973-975.
https://doi.org/10.1016/j.jaad.2017.12.035
Racial disparities and insurance status:
An epidemiological analysis of Ohio
melanoma patients
To the Editor: Recent research has shown that
minority patients with melanoma have a 1.96- to
3.01-fold greater risk of disease-specific mortality
compared with white patients,1 but the question as to
why remains. Given that insurance status varies
significantly across racial groups,2 it is possible that
insurance status has contributed to the later-stage
diagnoses and consequent poorer outcomes of
melanoma in minority patients. This analysis con-
siders the effect of race and insurance status on
melanoma stage at diagnosis and uniquely empha-
sizes the significant association of Medicaid with late-
stage diagnoses within a single race.
We performed an epidemiological assessment of
26,958 patients with cutaneous melanoma by using the
Ohio Cancer Incidence Surveillance System registry
from 1996 to 2009. The cancer incidence data used in
these analyses were obtained from the Ohio
Department of Health’s Ohio Cancer Incidence
Surveillance System, which is a cancer registry partially
supported by the National Program of Cancer Registries
at the Centers for Disease Control and Prevention
through cooperative agreement 1U58DP003936-01.
Use of these data does not imply that the Ohio
Department of Health and Centers for Disease
Control and Prevention agree or disagree with the
analyses, interpretations, or conclusions in this article.
We collected information on insurance status
( primary payer at diagnosis), race, and stage at
diagnosis. Chi-square analyses were performed and
risk ratios were determined with a significance of P
of less than .05. The categories of race for the chi-
square analyses were white, black, and other
(‘‘other’’ included Asian/Pacific Islander, American
Indian, and other race entries in the database).
Chi-square analysis revealed a statistically signif-
icant relationship between insurance status and race
(P \ .001) (Fig 1). The proportions of race differed
significantly within Medicaid (P \ .001) and private
insurance (P ¼ .025).
Black patients were more likely to present at a late
stage, which included stage III or IV at diagnosis
(P \.001) (Table I). Further, patients with Medicaid
were significantly more likely to present at a late
stage compared with patients with other insurance
types, including those who were uninsured
J AM ACAD DERMATOL
MAY 2018
Fig 1. Distribution of insurance status by race. Insurance
status presented by race. Percentages are of each racial
group separately. Other includes Asian, American Indian,
and ‘‘other’’ entries in the database.
(P \.001). Of note, patients with Medicaid remained
significantly more likely to present at a late stage
(P \ .001) compared with patients in all insurance
status categories when only white patients were
considered (Table I).
These findings suggest that insurance type alone,
specifically Medicaid, may be involved in the dis-
parities of melanoma stage at diagnosis. Potential
explanations for this are decreased utilization of
cancer screening in Medicaid subscribers and pro-
vider discrimination based on insurance status. Prior
research has shown that Medicaid patients face
lower acceptance rates for dermatology appoint-
ments (32% acceptance for those with Medicaid, 85%
for those with Medicare, and 87% for those with
private insurance) and longer wait times (50 days
average for those with Medicaid and 37 days average
for those with Medicare and private insurance),3
possibly contributing to poorer outcomes.
A limitation of our study is that the data do not
incorporate changes in insurance status before or
after the time of diagnosis, given that length of time
enrolled in Medicaid has been shown to be a
significant variable affecting melanoma stage at
diagnosis.4
Patients with cancer who are on Medicaid have
been shown to have comparatively later-stage di-
agnoses,5 but few studies have considered patients
with melanoma specifically. This study provides
evidence that insurance type alone is associated
with stage at melanoma diagnosis. Given that a
larger proportion of patients from minority popula-
tions have Medicaid,2 it is likely that insurance status
is contributing to the current racial disparities in
melanoma outcomes. In light of the frequently
debated and shifting configuration of health insur-
ance in this country, it is important that providers