Screening program reduced melanoma mortality
at the Lawrence Livermore National
Laboratory, 1984 to 1996
Jeffrey S. Schneider, MD,a,bDan H. Moore II, PhD,b,cand Mortimer L. Mendelsohn, MD, PhDb
San Rafael, Livermore, and San Francisco, California
Background: Worldwide incidence of cutaneous malignant melanoma has increased substantially, and no
screening program has yet shown reduction in mortality. We evaluated results of an educational campaign
designed to promote self-examination and targeted screening at the Lawrence Livermore National
Methods: Thickness and crude incidence of melanomas detected during 3 phases of increasing melanoma
surveillance were studied. These periods were: (1) preawareness (1969-1975), (2) early awareness of
increased melanoma risk (1976-1984); and (3) screening program (1984-1996). Melanoma mortality was
derived from data recorded in the National Death Index search. The expected annual number of deaths
from melanoma among LLNL employees was calculated by using California mortality data matched by age,
sex, and race/ethnicity and adjusted to exclude deaths from melanoma diagnosed before the program
began or before employment at LLNL.
Results: Crude incidence of melanomas thicker than 0.75 mm decreased during the 3 periods from 22.1 to
15.13 to 4.62 cases per 100,000 person-years (P = .001 by chi-square for trend) with the larger decrease from
the active screening program. The crude incidence of melanoma measuring less than 0.75 mm in thickness
increased and then decreased slightly without a significant linear trend, and crude incidence of in situ
melanoma increased substantially. No eligible melanoma deaths occurred among LLNL employees during
the screening period, whereas the expected number of deaths was calculated to be 3.39 deaths (P = .034).
Limitations: The study design was not randomized or controlled. The methodology for adjusting expected
mortality for the exclusion of employees diagnosed with melanoma before the screening period was
devised for this study.
Discussion: Increasing community awareness of melanoma was associated with a progressive decreasing
incidence of thicker melanoma. The education, self-examination, and selective program generated the
larger reduction in incidence of melanoma thicker than 0.75 mm. This campaign decreased the melanoma-
related mortality to zero. The statistically significant decrease in mortality persisted for at least 3 years after
employees retired or otherwise left the laboratory. ( J Am Acad Dermatol 2008;58:741-9.)
were undertaken to reduce melanoma mortality.
Most consisted of brief, no-charge skin screenings
or publicity campaigns; these efforts showed in-
creased numbers of thin tumors but did not evaluate
mortality.2Population-based timeseries in England,3
Scotland,4,5and Trentino, Italy6,7involved education
campaigns to teach identification of suggestive le-
sions to the local populations and medical commu-
nity, encouraging people to be examined by medical
personnel and helping to expedite referral and
diagnosis. The English study3showed an increase
incidence increased 137%, and mortality in-
creased 29%.1Many early-detection campaigns
From the Department of Dermatology, Kaiser Permanente Medical
Center, San Rafaela; Biomedical Sciences Division, Lawrence
Livermore National Laboratoryb; and Research Institute, Cal-
ifornia Pacific Medical Center, San Francisco.c
Research support, including travel, was provided by the US
Department of Energy, which had no role in the study design.
The Kaiser Foundation Research Institute provided a grant
139-9858 for partial salary support from 1994 to 1997.
Conflicts of interest: None declared.
Accepted for publication October 28, 2007.
Reprint requests: Jeffrey Schneider, MD, Kaiser Permanente
Medical Center, 99 Montecillo Road, San Rafael, CA 94903.
Published online December 10, 2007.
ª 2007 by the American Academy of Dermatology, Inc.
in thin tumors and no effect on mortality. The Italian
program showed an increase in the proportion of
thin tumors6and suggested a possible reduction in
mortality.7The Scottish study showed a temporary
reduction in thick tumors and a stabilization of
mortality among women and men.4,5A large, ran-
domized controlled trial of melanoma screening in
Australia is in its preliminary stage.8
In 1981, a 3- to 4-fold excess incidence of mela-
noma during the period 1972 to 1977 was reported
among the approximately 5100 persons then em-
ployed at the Lawrence Livermore National Labora-
tory (LLNL) in Northern California. In response, a
secondary melanoma prevention campaign began in
1984. This campaign consists of employee educa-
tion, self-examination, and targeted onsite screening
conducted by a dermatologist.
We evaluated the LLNL melanoma screening and
education program by investigating the thickness of
diagnosed melanoma from 1969 through 1996 and
deaths from melanoma among LLNL employees and
former employees from July 1984 through 1996.
Melanoma education and screening program
In 1976 a preliminary study by the medical direc-
tor of LLNL suggested an increased incidence of
melanoma from 1969 through 1975, which is con-
sidered the preawareness period. The early aware-
ness period from 1976 to June 1984 was effected by
the publicity surrounding this announcement and
otherstudies9,10confirming the3- to4-fold increased
incidence. This lead to the LLNL education and
screening program, which began in July 1984 with
instructions to employees about sun protection,
signs of melanoma, and melanoma risk factors.
This information was disseminated by direct mail-
ings, worksite news articles, meetings, and lectures
to employees and local physicians. The local news
media responded with articles about the program.
Employees were encouraged to examine themselves
for suggestive lesions; if self-examination showed a
suggestive lesion, a visit was arranged to the onsite
screening facility for full-body examination, dermo-
scopy, and biopsy if indicated. Alternatively, per-
sonnel could be examined by their personal
physicians, in which case the employees were en-
couraged to report results to the LLNL medical staff.
All employees were also sent a form to report their
mole counts at the outset of the program, and new
employees received this form as part of their new-
employee orientation at LLNL. Program participants
who counted 5 or more moles greater than or equal
to 5 mm in diameter or a single mole greater than or
equal to 18 mm in diameter were offered a screening
were considered for pathologic confirmation even if
melanoma was not suggested; in this regard, evalu-
ators followed recommendations promulgated at the
October 1983 National Institutes of Health (NIH)
Consensus Conference, published online as Precur-
NIH Consensus Conference,13new guidelines were
followed with biopsy performed only on atypical
lesions to evaluate for melanoma. All examinations
were performed by one of the authors (J. S. S.) or by
fellows from the University of California San Fran-
cisco Melanoma Clinic. During the study period,
virtually all melanoma in LLNL employees was
diagnosed in the onsite screening program. After
dermatologic evaluation, employees with melanoma
(invasive or in situ), dysplastic nevi, 50 or more
moles, or family history of melanoma were offered
often with full-body photography and dermoscopy,
according to level of melanoma risk.14
Melanoma thickness and crude incidence data
Institutional review board approval was obta-
ined for all aspects of the study. Biopsy specimens
obtained by the screening program at LLNL were
evaluated at the University of California San Fran-
cisco Melanoma Center, and the thickness of each
diagnosed melanoma was recorded. Any employees
who reported an outside diagnosis of melanoma had
their slides reviewed at University of California San
Francisco. Crude incidence of melanoma was calcu-
lated for the preawareness, early awareness, and
screening periods stratified into 4 thickness cate-
gories: in situ, less than or equal to 0.75 mm, 0.75 to
1.50 mm, and greater than 1.50 mm. The preaware-
ness period was January 1969 through December
1975, the early awareness period was from January
1976 to June 1984, and the screening campaign was
July 1984 through December 1996. The denominator
was the number of people employed at LLNL during
the respective time periods.
Mortality data models and statistical analysis
A roster was compiled listing all LLNL employees
with at least 6 months’ tenure who worked at LLNL
LLNL: Lawrence Livermore National Laboratory
NIH: National Institutes of Health
NDI: National Death Index
SEER: Surveillance, Epidemiology, and End
J AM ACAD DERMATOL
742 Schneider et al
sometime during the period July 1984 through 1996.
Roster information about employees and former
employeesefull name, date of birth, sex, and US
Social Security numberewas sent to the National
Center for Health Statistics for matching with records
in the National Death Index (NDI), a computerized
database of records from the vital statistics offices of
all US states and territories. Matches were confirmed
independently using LLNLrecords. TheNDI provided
the coded cause of death for each probable match.
Annual California melanoma mortality for 1989
through 1996 was obtained online from Expert
Health Data Programming Inc.15Because the num-
ber of deaths in each subcategory was small, the
annual melanoma-related mortality was highly var-
iable when subcategorized by sex, race, and 5-year
age subgroups. The mortality was smoothed, using a
Poisson regression model, before being applied to
population at-risk data. Mortality was extrapolated
back from 1989 through 1984 using the same model.
The smoothed deaths based on the smoothed mor-
tality showed good agreement with the observed
mortality (chi-square for fit = 33.8 with 33 degrees of
freedom P = .43).
Total number of melanoma deaths expected for
the LLNL population was obtained by summing
the products of calendar year/sex/race/age group-
smoothed mortality and multiplying this product by
the LLNL employee population. This calculation is
shown mathematically by the equation:
LLNL melanoma deaths ¼ +
where Rxrepresents the calendar year/sex/race/age
group-smoothed mortality and Px represents the
corresponding calendar year/sex/race/age group-
specific employee populations at LLNL. The summa-
tion includes all calendar year/sex/race/age groups.
We calculated cumulative mortality,3which ex-
cluded deaths among patients given the diagnosis
before July 1984 or new hires who developed
melanoma before employment at LLNL, because
they could not have benefited from the screening
program. To account for this exclusion of pre-1984
case deaths in the study group, we adjusted the
expected number of deaths by subtracting those
likely to occur among patients given the diagnosis
before 1984. This adjustment was based on observed
melanoma mortality for the 5 San Francisco-Oakland
Bay area counties as reported to the Surveillance,
Epidemiology, and End Results (SEER) program.
Table I shows the adjustment percentages for each
year of the study. We did not need to correct for new
hires after 1984, who were given a diagnosis of
melanoma, before employment at LLNL. Assuming
persons who left the laboratory who had melanoma
would be the same or more (because the laboratory
incidence rate was higher than the community rate)
than the new employees. Because all employees
who left LLNL were followed up for melanoma
mortality and would have been included in observed
mortality, we excluded from observed mortality any
hires after 1984 who died from melanoma diagnosed
periods of LLNL employment or until 1996. In this
computation, a number of years of risk beyond
termination of LLNL employment was added for
each employee. Employees and former employees
who died from any cause were not included in the
at-risk population beyond the year of the death.
Because the observed numbers of melanoma-
related deaths can be considered rare events, these
numbers can be expected to follow a Poisson distri-
bution when compared with the adjusted expected
number of melanoma-related deaths. The probabil-
ity that 0 events are observed is equal to e?lwhen l
events are expected. This probability is approxi-
mately .05 when l = 3.
Decreased incidence of high-risk melanoma
From January 1969 through December 1975 14
melanomas were diagnosed; of these 9 (64%)
Table I. Distribution of deaths from melanoma
(1984-1996) in 5 San Francisco Bay area counties
stratified by diagnosis date before July 1984
Deaths in cases diagnosed
before July 1984
Year Total deaths, No.No. (%) Smoothed percentage*
*Used to calculate number of deaths expected in Lawrence
melanoma before July 1984.
J AM ACAD DERMATOL
VOLUME 58, NUMBER 5
Schneider et al743
measured greater than 0.75-mm thick. From January
1976 through June 1984, 34 melanomas were diag-
nosed; of these, 9 (26%) measured greater than 0.75-
incidence of these high-risk lesions to be 22.11 and
1975 and 1976 to 1984 periods, respectively. The
reduction is 32%. From July 1984 through December
which measured greater than 0.75-mm thick; these
results yielded a crude incidence of 4.62 lesions per
100,000 person-years (Table II). This is a 69% reduc-
tion from the 1976 to 1984 rate. This series showed
that the number of higher-risk melanomas ([0.75
mm) decreased markedly during the time period
1969 to 1996 and particularly markedly after screen-
ing began in July 1984 (P = .001 by chi-square test for
and P = .024 by chi-square test for the last two time
periods). We noted with particular interest that 3 of
the 5 lesions thicker than 0.75 mm were diagnosed
during the first year and that the other two lesions
were diagnosed by the sixth year of the 12-year
screening campaign (Table II). The crude incidence
of melanoma measuring less than 0.75 mm in thick-
ness increased and then decreased slightly without a
significant linear trend. We observed a large increase
in number of ‘‘in situ’’ melanoma with the increasing
common use of the term in the late 1970s.
Decreased mortality from melanoma
A search of the NDI showed 3 melanoma-related
deaths, all in people known to have the disease
before either July 1984 or their date of being hired at
LLNL. Therefore, the eligible mortality was zero with
an upper 95% limit of 2.6 deaths.
Table III shows employee population by year
and expected numbers of melanoma-related deaths
among LLNL employees by sex, race, and year.
During this period, no melanoma-related deaths
were observed among LLNL employees who were
not already given a diagnosis of melanoma before
July 1984; and the probability of observing zero
deaths is P = .034 when 3.39 deaths are expected.
Long-term effect of screening campaign
The original study plan was to extend the analysis
year-by-year beyond termination of LLNL employ-
ment. This was to account for possible melanoma
deaths among LLNL employees who terminated their
employment because of illness from melanoma.
Secondarily this allowed for the analysis of possible
decay of the education effect as terminees escaped
LLNL melanoma-prevention methods. This em-
ployeeawareness might not diminish among termin-
ees who retired with benefits from LLNL and
continued to receive LLNL publications. However,
no such melanoma deaths were observed, and
terminees survived remarkably well from all causes
Table II. Number and crude incidence of melanoma during preawareness (1969-1975), early awareness
(1976-1984), and education and screening program (1984-1996) at Lawrence Livermore National Laboratory
Year In situ
No. of lesions
1984-1996 (linear trend)
24.01 (P = .0003) 21.24 (P = .32) 4.62 (P = .04) 0.00 (P = .001)
J AM ACAD DERMATOL
744 Schneider et al
during the study period. We, therefore, calculated
the effect on P value of 3 additional years of survival
into retirement or until 1996. This calculation
method increased the number of person-years under
study from 113,715 to 149,474. Because no mela-
noma deaths were observed and the expected
number of melanoma deaths was 5.26, the proba-
bility of chance being an explanation was reduced
to P = .005.
The study confirmed a reduction in melanoma
mortality in the LLNL workforce from July 1984
through December 1996 with 3.39 deaths expected
from the California mortality data and no eligible
cases observed. This result was markedly different
from the high, but not significant, increase in deaths
noted in a previous national death certificate search
atLLNL from 1964through1979,when 6deaths were
observed and 4 deaths were expected.10
Several potential limitations of this study were
associated with its design and with the relatively
small occupational setting. Ideally, intervention
would have been evaluated by a randomized con-
trolled study. Individual randomization within LLNL
was impossible given the goals of the program; and
although comparison with outside sites was consid-
ered, no satisfactory comparable site was found.
Therefore, California, with a very large population
base that can provide stable mortality and the local
SEER San Francisco-Oakland Bay area incidence
records, was used for comparison with the LLNL
data. Possible sources of error include failure to
detect a melanoma death, comparability of the LLNL
workforce with the local SEER and general California
population, difficulty of properly adjusting for ex-
clusion of the pre-July 1984 melanoma cases, small
number of expected melanoma deaths, and inclu-
sion of LLNL terminees.
Failure to detect a melanoma death because of an
error in matching the LLNL employee roster with the
NDI was unlikely. The matching algorithm was
robust using abest-fit analysis in case no exact match
existed for name, sex, date of birth, and social
security number. Moreover, all known patients at
LLNL given a diagnosis of melanoma during the
study period have received follow-up, and none of
these employees have died from melanoma. The
number of melanoma cases shown in an all-cancer
incidence study conducted among LLNL employees
in cooperation with the local SEER registry was
similar to the number of melanoma cases shown by
our study.16This finding makes remote the possibil-
ity of an undiscovered melanoma death in an LLNL
employee given the diagnosis of melanoma during
the study period.
In the absence of a matched or randomized
control group, we used age-, sex-, and race-specific
mortality for California as reasonable comparison
data for analyzing mortality among LLNL employees.
The large size of the California database provided
much more stable mortality, especially for the low-
risk groups. The local SEER database, which might
have been more comparable, had a higher mortality,
so using the state data would not have unfairly
increased the significance of the result. Some risk
factors for melanomaenumber and type of nevi
along with family and personal history of melanoma
Table III. Population and expected deaths from melanoma at Lawrence Livermore National Laboratory
19841985 1986 19871988 198919901991 1992 19931994 1995 1996
J AM ACAD DERMATOL
VOLUME 58, NUMBER 5
Schneider et al745
and other types of skin cancerecould not be
assessed, because these data were unavailable; how-
ever, we were able to make some estimates on
ultraviolet light exposure and social class.17Sun
to exposure among the general California popula-
tion, given that most LLNL employees live in the
sunny north-central section of the state, where
the climate resembles the climate found in most of
the heavily populated areas of the state.
Socioeconomic status and proportion of doctor of
philosophy awardees and college graduates in the
LLNL workforce is much higher than in the general
California population. Socioeconomic status and
education are strong risk factors for melanoma inci-
dence. Two population-based studies18,19showed
a 2- to 4-fold effect of socioeconomic status on
incidence of melanoma. The effect on melanoma
mortality is smaller. One study18showed that highly
educated persons had a 55% higher melanoma-
The high education level and high socioeconomic
status among LLNL employees would produce a
higher melanoma-related mortality than expected
from age-, sex-, and race-based mortality in the
general California population. This difference makes
the achieved mortality reduction more impressive.
To ensure inclusion of workers who terminated
their LLNL employment because of metastatic mela-
noma, surveillance of melanoma-related deaths in-
cluded all persons who had terminated their LLNL
employment until the end of the study period in
1996. In our calculation of expected mortality among
retirees we used 3 years of follow-up only. Because
screening-derived benefit may be lost after employ-
ment, we initially planned to watch for an increase in
mortality in the years after termination from LLNL;
however, no additional deaths were observed in all
the NDI data. We chose a limited postemployment
period to calculate additional expected deaths, be-
cause we initially predicted that the effect of screen-
ing would not be long lasting. This 3-year point
makes only a conservative increase in person-years
of follow-up and is useful because more than 90% of
mortality from systemically metastatic melanoma
would probably occur by that point.20Inclusion of
LLNL workforce terminees strengthens the compre-
hensiveness of our study. The extended follow-up
added person-years to the study and thus added
statistical significance to the reduction in mortality.
Moreover, the statistical significance of the result
would persist even if one death had occurred in the
screened population during the study. This fact
makes the result more robust, even with the small
number of expected deaths.
The spike of melanoma cases, which prompted
creation of the screening program, might have
caused a decline in number of melanoma cases as
an after-effect. Instead, cases of melanoma (in situ
plus invasive) continued to be diagnosed at a varia-
ble, but somewhat increasing rate during the 3
periods, while showing the striking reduction in
thicker lesions (Table II). The striking and persistent
reduction in the number of thick lesions with higher
provides the strongest support for concluding that
the program resulted in real reduction of melanoma
mortality. This conclusion contrasts with other stud-
ies3-5of melanoma screening, which did not show
any decrease in the absolute number of lesions
thicker than 0.75 mm. These studies report a de-
creased percentage of thick tumors, because of the
even larger increase in thin tumors. Unfortunately,
without a decrease in thick tumors, mortality would
not be expected tochange.Furthermore, two studies
using SEER melanoma incidence data by thickness
through 1987 and the other using national SEER
data24from 1988 through 1994 show the LLNL
decrease in the rate of thick tumors was a marked
showed an increased incidence rate in all thickness
categories more than 0.75 mm. Among men who
make up more than 70% of the laboratory, the San
national study showed a similar increase from 1988
through 1994. This is consistent with the increased
melanoma mortality in men seen during this period.
In women, who increased from 20% to 30% of the
workforce during the study, the trend of increasing
end of the 1988 to 1994 study there was a decrease in
thick tumors. This more recent change in female
melanoma would have only had a small effect on the
overall dramatic decrease in thick lesions at the labo-
ratory and was controlled for in the mortality study.
There was an increase in nonwhite workers from
12% to 19% during the time period of the LLNL study.
Although more darkly pigmented races have a lower
melanoma incidence, their increased proportion of
the workforce would have been expected to pro-
duce the opposite effect in terms of tumor thickness.
A recent study of Hispanics in California25showed
tumors greater than 1.5 mm increased significantly
from 1988 to 2001 at a rate of 11.6% per year.
Nationally, from 1992 to 2002 melanomas in darkly
pigmented ethnic groups had a significantly in-
creased median thickness and other high-risk factors
such as Clark level IV and tumor ulceration
J AM ACAD DERMATOL
746 Schneider et al
compared with whites.26This again makes our
reduction in thick tumors more notable and should
not have had an effect on the mortality study, where
race was taken into account.
Another possible bias influencing decreased thick
tumor and mortality could be possible changes and
improvements in health care for LLNL employees.
The LLNL health benefits were excellent, but no
different from most employees at large firms or
government workers. LLNL used the University of
Permanente, which cared for a large number of
laboratory employees, compared dermatologic care
for LLNL employees and a control group.27Both
groups had equal access, but the number of visits to
dermatologists and biopsies among LLNL employees
were elevated with the biopsies significantly in-
creased. Furthermore, the biopsy rate difference only
became significant after 1976. This supports the con-
elevated at LLNL and increased after 1976. After 1984
virtuallyall melanomaswerediagnosedonsiteas part
of the screening program. We agree this easier access
to care was a major factor in the reduced mortality.
One might also argue that the observed reduction
in melanoma mortality was a result of elimination of
an occupational cause of melanoma at LLNL.
Intensive efforts were made before and during the
agent was ever found,28and no preventive occupa-
tional measures (other than this surveillance pro-
gram) wereinstituted.Perhaps most importantisthat
our mortality comparison was applied to concurrent
California mortality and not to earlier LLNL mortality.
At least 3 factors reduced the number of thick
melanoma lesions: (1) prompt recognition and re-
moval of melanoma lesions by the medical staff; (2)
routine follow-up of several hundred employees
who were at increased risk for melanoma; and (3)
LLNL employees knowing the warning signs of
melanoma, actively practicing skin awareness (eg,
counting moles), and presenting themselves for
examination after noting suggestive lesions. The
importance of rapid recognition and biopsy of sug-
gestive lesions is clear, and the pathology data con-
firm the success of this part of the program. Virtually
all the melanoma diagnosed during the screening
program was diagnosed by the onsite dermatologic
staff. Close follow-up of families with dysplastic
nevus and melanoma is well documented to reduce
the thickness of newly diagnosed melanoma le-
sions29and should also be successful for routinely
screened LLNL employees with moderately elevated
risk. Because most of the LLNL workforce was not
examined in the voluntary screening program and
because most of those who participated were seen
examination helped to prevent development of
thicker, higher-risk melanoma lesions. This same
effect may have occurred among retirees.
A large, population-based case-control study of
melanoma suggested that self-examination reduced
the relative risk of melanoma mortality by 63% and
predicted a simultaneous 34% reduction of invasive
melanoma incidence.30Most screening programs2
and the early awareness second phase of the LLNL
experience show amarked increased rate of invasive
melanoma. Berwick’s data predicts that a program
with a marked mortality reduction would show a
reduction in the mean thickness of cases and,
surprisingly, a reduction in all invasive cases.30Our
screening, education, and self-examination program
45 cases per 100,000 person-years during the early
study. The study of Berwick et al30reported a mean
Breslow thickness of 1.09 mm among patients with
melanoma who performed self-examination com-
pared with 1.65 mm among patients with melanoma
who did not examine their skin. LLNL employees
showed a similar difference in mean lesion thick-
ness: 1.05 mm before the intervention and 0.59 mm
during the study period. The quality of the derma-
tologic staff at LLNL was probably similar to that of
the motivated practitioners in other studies, and
screening by itself has not been successful in other
studies; therefore, the similarity between Berwick’s
findings30and ours suggests that the high degree of
melanoma awareness and self-examination was a
major factor that led to the reduction in incidence of
thick lesionsand melanoma-related
Achieving this level of interest in the general
populationein which many diseases have higher
incidence and mortalityemay be difficult.
Worksites are different from the community at
large, and LLNL is a particularly unusual occupa-
tional setting because of the specialized nature and
high level of security of the high-energy physics
research taking place there. Located on a compact
campus, the highly educated population was well
informed about critical events. Two factorsea 1976
report describing local increase in melanoma cases
and publication of several studiesemarkedly in-
creased employee anxiety and interest in melanoma.
Even before the program began, workforce aware-
ness and early diagnosis may have surpassed the
levels observed in the surrounding community.31,32
At the outset of the study, which was accompanied
by massive local publicity in 1984, 96% of LLNL
J AM ACAD DERMATOL
VOLUME 58, NUMBER 5
Schneider et al747
regarding melanoma. Throughout the study, the
program staff performed almost 11,600 dermatologic
examinations and received more than 7000 com-
pleted self-examination sheets. Providing this well-
motivated, intelligent workforce with easily accessi-
ble, free-of-charge, onsite screening showed that
melanoma-related mortality can be reduced through
a program of education, self-examination, and lim-
The mortality reduction we observed probably
resulted from a preventive strategy consisting of 3
components: hands-on examination of about half
the employees being screened; long-term screening
of a small, moderate-risk population; and sensitizing
nearly all employees and their spouses and care-
givers to the menace of melanoma and to the
importance of noting suggestive lesions and count-
ing moles. Our result emphasizes the usefulness of
education at all levels in any future screening pro-
gram. We look forward to larger, more broadly
based, controlled studies to confirm this result.
Editorial assistance was provided by the Medical
Editing Service of The Permanente Medical Group
Physician Education and Development Department.
The authors thank Richard W. Sagebiel, MD, of the
University of California, San Francisco Melanoma Center,
who helped set up the screening program and evaluated
almost all the pathology specimens and Robert A. Hiatt,
MD, PhD, of the University of California, San Francisco,
who reviewed the manuscript. The authors also thank
Karen Martin, RN, who was the nurse coordinator for the
project and the many University of California, San
Francisco melanoma fellows who assisted in the clinical
1. National Cancer Institute. Surveillance, epidemiology, and end
results. Available from: http://seer.cancer.gov/csr/1975_2001/
results_merged/sect_16_melanoma.pdf. p. 2-3. Accessed July
2. Koh HK, Norton LA, Geller AC, Sun T, Rigel DS, Miller DR, et al.
Evaluation of the American Academy of Dermatology’s
national skin cancer early detection and screening program.
J Am Acad Dermatol 1996;34:971-8.
3. Melia J, Moss S, Coleman D, Frost T, Graham-Brown R, Hunter
JA, et al. The relation between mortality from malignant
melanoma and early detection in the cancer research cam-
paign mole watcher study. Br J Cancer 2001;85:803-7.
4. MacKie RM, Hole D, Hunter JA, Rankin R, Evans A, McLaren K,
et al. Cutaneous malignant melanoma in Scotland: incidence,
survival, and mortality, 1979-94. The Scottish melanoma
group. BMJ 1997;315:1117-21.
5. MacKie RM, Bray CA, Hole DJ, Morris A, Nicolson M, Evans R,
et al. Scottish Melanoma Group. Incidence of and survival from
malignant melanoma in Scotland: an epidemiological study.
6. Cristofolini M, Zumiani G, Boi S, Piscioli F. Community detec-
tion of early melanoma [letter]. Lancet 1986;1:156.
7. Cristofolini M, Bianchi R, Boi S, DeCarli A, Micciolo R,
Cristofolini P, et al. Effectiveness of the health campaign for
the early diagnosis of cutaneous melanoma in Trentino, Italy.
J Dermatol Surg Oncol 1993;19:117-20.
8. Aitken JF, Elwood JM, Lowe JB, Firman DW, Balanda KP, Ring
IT. A randomized trial of population screening for melanoma.
J Med Screen 2002;9:33-7.
9. Austin DF, Reynolds PJ, Snyder MA, Biggs MW, Stubbs HA.
Malignant melanoma among employees of Lawrence Liver-
more National Laboratory. Lancet 1981;2:712-6.
10. Moore DH II, Bennett DE. Mortality among the employees of
the Lawrence Livermore National Laboratory. Energy Technol
11. Lawson DD, Moore DH II, Schneider JS, Sagebiel RW. Nevus
with count by physician. J Am Acad Dermatol 1994;31:438-44.
12. Precursors to malignant melanoma. National Institutes of Health
consensus statement online. October 24-26, 1983;4(9):1-14.
Available from: http://www.ncbi.nlm.nih.gov/books/bv.fcgi?
rid=hstat6.chapter.48887. Accessed November 23, 2007.
13. Diagnosis and treatment of early melanoma. National Insti-
tutes of Health consensus statement online. January 27-29,
1992;10(1):1-26. Available from: http://www.ncbi.nlm.nih.gov/
books/bv.fcgi?rid=hstat4.chapter.10336. Accessed November
14. Schneider JS, Moore DH II, Sagebiel RW. Risk factors for
melanoma incidence in prospective follow-up: the importance
of atypical (dysplastic) nevi. Arch Dermatol 1994;130:1002-7.
15. Expert Health Data Programming Inc. Health data software
and health statistics. Available from: URL:http://www.ehdp.
com. Accessed April 1, 2001.
16. Whorton MD, Moore DN, Seward JP, Noonan KA, Mendelsohn
ML. Cancer incidence rates among Lawrence Livermore Na-
tional Laboratory (LLNL) employees: 1974-1997. Am J Ind Med
17. Berwick M. Epidemiology: current trends, risk factors, and
environmental concerns. In: Balch CM, Houghton AN, Sober
AJ, Soong SJ, editors. Cutaneous melanoma. 3rd ed. St Louis:
Quality Medical Publishing; 1998. pp. 551-71.
and survival of patients with cutaneous malignant melanoma
in relation to socioeconomic status. BMJ 1996;312:1125-8.
19. Geller AC, Miller DR, Lew RA, Clapp RW, Wenneker MB, Koh HK.
Cutaneous melanoma mortality among the socioeconomically
disadvantaged in Massachusetts. Am J Public Health 1996;
20. National Cancer Institute. Surveillance, epidemiology, and end
results. Available from: http://seer.cancer.gov/faststats/sites.
cessed May 6, 2007.
21. Balch CM, Buzaid AC, Soong SJ, Atkins MB, Cascinelli N, Coit
DG, et al. Final version of the American Joint Committee on
Cancer staging system for cutaneous melanoma. J Clin Oncol
Cascinelli N, et al. Prognostic factors analysis of 17,600 mela-
noma patients: validation of the American Joint Committee on
23. Horn-Ross PL, Holly EA, Brown SR, Aston DA. Temporal trends
in the incidence of cutaneous malignant melanoma among
Caucasians in the San Francisco-Oakland MSA. Cancer Causes
24. Hall HI, Miller DR, Roger JD, Bewerse B. Update on the
incidence and mortality from melanoma in the United States.
J Am Acad Dermatol 1999;40:35-42.
J AM ACAD DERMATOL
748 Schneider et al
25. Cockburn MG, Zadnick J, Deapen D. Developing epidemic of
melanoma in the Hispanic population of California. Cancer
26. Cormier JN, Xing Y, Ding M, Lee JE, Mansfield PF,
Gershenwald JE, et al. Ethnic differences among patents
with cutaneous melanoma. Arch Intern Med 2006;166:
27. Haitt RA, Fireman B. The possible effect of increased surveil-
lance on the incidence of malignant melanoma. Prev Med
28. Moore DH, Patterson HW, Hatch F, Discher D, Schneider JS,
Bennett D, et al. Case control study of malignant melanoma
among employees of the Lawrence Livermore National Lab-
oratory. Am J Ind Med 1997;32:377-91.
29. Masri GD, Clark WH Jr, Guerry D IV, Halpern A, Thompson CJ,
Elder DE. Screening and surveillance of patients at high risk for
malignant melanoma result in detection of earlier disease.
J Am Acad Dermatol 1990;22:1042-8.
30. Berwick M, Begg CB, Fine JA, Roush GC, Barnhill RL. Screening
for cutaneous melanoma by skin self-examination. J Natl
Cancer Inst 1996;88:17-23.
31. Schneider JS, Moore DH II, Sagebiel RW. Early diagnosis of
cutaneous malignant melanoma at Lawrence Livermore Na-
tional Laboratory. Arch Dermatol 1990;126:767-9.
32. Hiatt RA, Krieger N, Sagebiel RW, Clark WH Jr, Mihm MC Jr.
Surveillance bias and the excess risk of malignant melanoma
among employees of the Lawrence Livermore National Lab-
oratory. Epidemiology 1993;4:43-7.
J AM ACAD DERMATOL
VOLUME 58, NUMBER 5
Schneider et al749