Trends in the incidence of cutaneous malignant melanoma in New South Wales, 1983–1996


  • Loraine D. Marrett,

    Corresponding author
    1. Division of Preventive Oncology, Cancer Care Ontario, Toronto, Ontario, Canada
    2. Department of Public Health Sciences, University of Toronto, Toronto, Ontario, Canada
    • Cancer Care Ontario, 620 University Avenue, Toronto, Ontario, M5G 2L7, Canada
    Search for more papers by this author
    • Fax: +1-416-971-6888

  • Huu L. Nguyen,

    1. Cancer Research and Registers Division, New South Wales Cancer Council, New South Wales, Australia
    Search for more papers by this author
  • Bruce K. Armstrong

    1. Cancer Research and Registers Division, New South Wales Cancer Council, New South Wales, Australia
    Search for more papers by this author


The incidence of cutaneous malignant melanoma (CMM) has been rising in fair-skinned populations throughout the world for decades. The upward trend may, however, finally be slowing in some of these populations. Recent (1983–1996) CMM incidence trends for a high incidence area (New South Wales, Australia) have been examined according to gender, age group, body site and tumour thickness. Despite continuing upward trends in older age groups, particularly among men (e.g., 7.20% increase per year in men aged 75+), incidence for younger ages is stabilizing (in men) or declining (in women): average annual percentage changes of −3.03 and −0.88 were observed for women aged 15–34 and 35–54, respectively. Patterns suggest a birth-cohort effect, with those born since 1945 or 1950 having lower (females) or similar (males) rates to those born earlier. For each gender, all-ages incidence rose by a similar amount for each of the main body sites except the leg in women, where incidence fell by 0.49% per year. In men, the incidence of both thin (≤75 mm) and thick (>75 mm) melanomas increased (significantly, by 2.63% per year and non-significantly, by 0.93% per year, respectively) between 1989 and 1996. In women, incidence remained stable for both thickness subgroups. These data are consistent with a stabilization or reduction in either total sun exposure or intermittency of exposure among New South Wales cohorts born since about 1950. Because incidence rates are still much higher than they were a few decades ago, however, efforts to reduce sun exposure, particularly in children and youth, must continue. © 2001 Wiley-Liss, Inc.


AAPC, average annual percentage change; CI, confidence interval; CL, confidence limits; CMM, cutaneous malignant melanoma; ICD9, International Classification of Diseases, 9th Revision; NSW, New South Wales; SEER, Surveillance, Epidemiology and End Results.

The incidence of cutaneous malignant melanoma (CMM) has been rising in fair-skinned populations throughout the world for several decades.1 However, incidence may finally have begun to stabilize or even to fall in younger people (and in recently born cohorts) in some of these populations.2–4

As a result of the increasing incidence, CMM is now one of the more common cancers in white populations. It ranks 4th in men and 3rd in women in high incidence areas such as Australia5 and New Zealand (non-Maoris)6 and about 6th in medium incidence areas like the U.S. (whites), Scandinavia and parts of Canada.6 Further, CMM has a predilection for youth: it is one of the most common cancers in those age 15–34 in many high (e.g., Australia)6 and middle (e.g., Ontario, Canada)7 incidence areas. Because of these features, high incidence areas in particular have made efforts to educate the public and health professions about “sun smart behavior” and early detection of melanoma.

Monitoring trends in incidence of melanoma in a variety of settings is important to improve our understanding of its natural history, to monitor the possible effects of preventive or early detection strategies, to predict health service requirements, and to assist in setting and re-evaluating priorities within the health care system.

This report presents a detailed analysis of the trends in melanoma incidence for a recent time period (1983–1996) in a high incidence area. The state of New South Wales, Australia, had the highest incidence in males, and the second highest in females (after non-Maori New Zealanders) among all regions reported in the most recent volume of Cancer Incidence in Five Continents.6


Data sources

Information about new cases of cutaneous malignant melanoma was available from the population-based New South Wales Central Cancer Registry.8 New South Wales (NSW) is the most populous state in Australia, with a population of 6.2 million in 1996. Hospitals and radiotherapy departments in NSW have been required to report invasive cancers to the Registry since 1972, whereas pathology laboratories and day procedure centres have been required to notify all cancers since 1985 and 1991, respectively. Before 1985, notification from laboratories was voluntary. As part of a special study, however, all melanoma pathology reports for 1983 and 1984 were identified and entered into the Registry. Hence, melanoma data are believed to be quite complete since 1983. There has also been a very high rate of microscopic verification for melanoma since 1983 (98% in 1983–87 and 99% in 1988–92).6, 9

Data for all cases of invasive cutaneous malignant melanoma (International Classification of Diseases, 9th revision (ICD9), code 172)10 for the period 1983 to 1996 were extracted from the Registry for analysis; in situ or level I melanomas were excluded. Data items included year and age at diagnosis, gender, body site of melanoma (fourth digit of ICD9 code), and tumour thickness. Analyses involving tumour thickness were restricted to the period 1989–1996, when thickness was available for 87-89% of melanomas. Before 1989 it was infrequently recorded.

During the period 1983–1996, 28,867 cases of melanoma were diagnosed in residents of NSW, ranging from an annual low of 1,433 (in 1983) to a high of 2,479 (in 1996).


Estimated NSW populations by gender and 5-year age group were used.11 These constituted the denominators in the calculation of incidence rates for each gender and year in 5-year age groups to age 85+. From these, age-standardized rates were computed for broad age groups (15–34, 35–49, 50–64, 65–74, and 75+), body site groups (ICD9 172.0–172.4, head and neck; 172.5, trunk; 172.6, upper limb including shoulder; 172.7, lower limb, including hip; and 172.8–172.9, other and unspecified) and tumour thickness (≤0.75 mm; >0.75 mm; unkown). Age-standardization employed the direct method and the 1991 Australian population.11

Trends in age-standardized incidence rates for the period 1983–1996 were analyzed by Poisson regression with a linear term for year of diagnosis by use of the STATA software:12

equation image

The average annual percentage change (AAPC) in age-standardized incidence rates (% change per year) and its 95% confidence interval (CI) were calculated for subgroups of interest from the maximum-likelihood estimates of the parameter for year of diagnosis. Quadratic effects were not examined.

To assess cohort patterns, age-specific rates (in 5-year age groups) were calculated for 3 intervals of year-of-diagnosis (1983–87; 1988–92; and 1993–96) and displayed for 9 birth cohorts constructed from these (mid-years of birth 1935 through 1975).


The age-standardized incidence of CMM increased in both genders over the 14-year period of study (Fig. 1), to a much greater degree in men than in women. A prominent feature of the trends is the steep increase between 1986 and 1988, followed by a fall to rates somewhat higher than those experienced previously. In women, the rates were comparatively stable after 1989.

Figure 1.

Age-standardized incidence rates of melanoma by gender, NSW, 1983–1996.

The incidence in males rose by an average of 3.17% per annum (95% CI 2.78–3.57) over all ages, whereas in females the average annual percent change was very small (0.45% per annum, 95% CI 0.02–0.90) (Table I). The male:female incidence rate ratio rose steadily from 1.0 in 1983 to 1.4 in 1996.

Table I. Average Annual Percent Change (AAPC) and 95% Confidence Limits (CL), 1983–1996, by Gender and Age Group, Incidence of Malignant Melanoma, New South Wales
GenderAge groupAAPC (%)95% CL
Male15–34−0.81−1.98, 0.36
35–491.300.42, 2.17
50–642.051.33, 2.77
65–744.533.69, 5.38
75+7.206.07, 8.34
All ages3.172.78, 3.57
Female15–34−3.03−4.04, −2.02
35–49−0.88−1.72, −0.04
50–641.680.82, 2.55
65–742.471.42, 3.52
75+3.872.65, 5.10
All ages0.450.02, 0.90

Age groups

The trends differ by age group within gender (Fig. 2 and Table I; note that Fig. 2 displays incidence rates using a logarithmic scale). Although the AAPCs increased with age, they were generally 2–3 percentage points less in females than males in each age group. In males, increases were sizable in the older age groups (4.53% and 7.20% per annum in men aged 65–74 and 75+ respectively), whereas incidence actually fell slightly (but not significantly) over time in the youngest age group (−0.81%, 95% CI −1.98 to 0.36). In females, incidence in the 2 youngest age groups (those age 15–34 and 35–49) fell significantly (AAPC = −3.03% and −0.88%, respectively). Rates seem to have stabilized in the most recent few years in most age-gender groups despite overall increasing trends.

Figure 2.

(a) Age-standardized incidence rates of melanoma by age group, males, NSW, 1983–1996. (b) Age-standardized incidence rates of melanoma by age group, females, NSW, 1983–1996.

The net effect of these age- and gender-specific changes in incidence is that the most recent rates were higher in men than in women at every age from 35 years. Although the male to female incidence rate ratio was only slightly greater than 1 up to age 50–54, it increased markedly with age thereafter (exceeding 2.5 for ages 75 and older). This is in contrast to the situation in the earlier years of the period when the incidence in women was higher up to age 50–54, and the male to female rate ratio increased less at older ages, never exceeding 2 (data not shown).

The pattern of change with age and time is consistent with a birth cohort effect. Although there are only 3 points (5-year age groups) for each 5-year birth cohort, it is evident from Figures 3 and 4 that incidence increased with year of birth in earlier born cohorts (more so in males than in females) then began to decrease (females) or stabilize (males) in later born cohorts. These changes occurred subsequent to the 1945 birth cohort for females and the 1950 cohort for males.

Figure 3.

Melanoma incidence by mid-year of birth cohort, males, NSW, 1983–1996.

Figure 4.

Melanoma incidence by mid-year of birth cohort, females, NSW, 1983–1996.

Body site

The highest gender- and site-specific incidence (unadjusted for body surface area) was for trunk melanomas in males. The lower limb was the most common melanoma site in females throughout the time period. “Other and unspecified” site represented 8% of melanomas in males and 6% in females (data not shown).

For males, incidence increased significantly at each specific site (Table II). The AAPCs varied little across sites, ranging from 2.82% per annum for lower limb lesions to 3.98% for upper limb melanomas. In females, the incidence of melanoma of the lower limb decreased (non-significantly), whereas that for the other sites increased significantly, at a rate of 1–2% per annum, with the highest rate of increase for trunk melanomas. Only for “other and unspecified” melanomas (almost entirely “unspecified” body site) did the incidence decrease in both genders during the study period. This category included 9–13% of the melanomas diagnosed in earlier years and 3–7% of those diagnosed in the mid-1990s.

Table II. Average Annual Percent Change (AAPC) and 95% Confidence Limits (CL), 1983–1996, by Gender and Body Site, Incidence of Melanoma, New South Wales
GenderBody siteAAPC (%)95% CL
MaleHead and neck3.802.89, 4.72
Trunk3.432.82, 4.06
Upper limbs3.983.06, 4.90
Lower limbs2.821.75, 3.91
Other and unspecified−1.05−2.37, 0.29
All sites3.172.78, 3.57
FemaleHead and neck1.280.11, 2.47
Trunk1.860.88, 2.84
Upper limbs1.720.81, 2.64
Lower limbs−0.49−1.22, 0.24
Other and unspecified−5.68−7.39, −3.94
All sites0.450.02, 0.90

Incidence increased more rapidly for males than for females at every site.

There was little evidence of an interaction between age and site relationships with incidence trends in either gender (data not shown). In males, though, incidence fell in the youngest age group (15–34) only for melanomas on the trunk (AAPC = −0.86%; 95% CI −2.57 to 0.87) and upper limb (AAPC = −1.47%; 95% CI −4.31 to 1.45), and for “other and unspecified” melanomas. For all other site-age-group combinations, the incidence in males increased. In females, incidence fell significantly in the youngest age group for every site and in the age group 35–49 for lower limb melanoma (AAPC = −2.42%; 95% CI −3.7 to −1.08). AAPCs and their 95% CIs for females aged 15–34 were as follows: −2.0% (−3.80 to −0.17) for trunk; −2.69% (−4.42 to −0.94) for lower limbs; −2.41% (−4.63 to −0.13) for upper limbs; and −3.79% (−7.05 to −0.41) for head and neck.


Trends in melanoma according to thickness of the lesion were examined only for the period 1989 to 1996. Within gender, the incidence rates for thin (≤0.75 mm) and thick (>0.75 mm) melanomas were about the same, although the rates of both were higher in males than in females (data not shown). The incidence of melanomas of unknown thickness was low, accounting for about 10% of newly diagnosed cases, and changed little over time (AAPC = −0.21%, 95% CI −2.05 to 2.53 in males; AAPC = −1.03%, 95% CI −3.99 to 2.03 in females).

The AAPCs for thin and thick melanomas over the 8-year period are shown by age group and gender in Table III.

Table III. Average Annual Percent Change (AAPC) and 95% Confidence Limits (CL) for Thin (≤0.75 mm) and Thick (>0.75 mm) Lesions, 1989–1996, by Gender and Age Group, Incidence of Melanoma, New South Wales
GenderAge groupThin (≤0.75 mm)Thick (>0.75 mm)
AAPC (%)95% CLAAPC (%)95% CL
Males15–34−3.04−6.69, 0.75−2.23−6.55, 2.29
35–490.17−2.43, 2.84−2.80−5.77, 0.27
50–642.840.44, 5.29−2.06−4.40, 0.33
65–744.772.00, 7.622.03−0.47, 4.59
75+7.103.31, 11.026.813.74, 9.96
All ages2.631.33, 3.940.93−0.33, 2.21
Females15–34−3.61−6.69, −0.43−3.38−7.64, 1.07
35–49−2.71−5.13, −0.22−1.12−4.32, 2.19
50–643.120.35, 5.97−1.21−4.21, 1.88
65–743.56−0.02, 7.262.80−0.56, 6.27
75+0.84−3.71, 5.614.030.67, 7.49
All ages−0.26−1.66, 1.170.30−1.27, 1.91

In males, the overall average annual rate of increase in the incidence of thin melanomas was significantly greater than zero (AAPC = 2.63%, 95% CI 1.33–3.94). Significant increases, however, occurred only for age groups 50 years and over. The incidence of thick lesions in males rose slightly overall, with an AAPC of 0.93% (non-significant). Only for men aged 75+ years was there a significant increase in incidence of thick melanomas (AAPC = 6.81%, 95% CI 3.74–9.96). There was a net downtrend in incidence of thin melanomas in men 15–34 years and of thick melanomas in men 15–64 years.

In females, the AAPC for thin melanomas was −0.26% (95% CI −1.66 to 1.17), indicating a small but not significant downtrend. Whereas young females (age groups 15–34 and 35–49) experienced significant falls in the incidence of thin melanoma, older women experienced rises (significant only in the age group 50–64). The same general pattern occurred for thick melanomas, although downward trends in the younger age groups were non-significant, and the upward trend in the oldest age group (75+) was significant.

The incidence rate ratio of thin to thick melanoma remained stable over the study period; it was close to 1 for males, and slightly in excess of 1 for females (data not shown).


Summary of results

There are several important features of the recent CMM incidence trends for New South Wales. First, the trends differ significantly according to gender: female incidence increased only very slightly over the 14 year study period (0.45% on average per year) whereas male incidence increased by 3.17% per year. The result is a substantial male preponderance of CMM, where previously there was approximate equality (though rates were higher in women at younger ages and men at older ages). Second, there are substantial differences in the trends according to age: incidence increased most in the oldest age groups, particularly in men, but actually fell significantly in young women (age groups 15–34 and 35–49) and non-significantly in the youngest men (age 15–34). Third, the patterns of change with age and time are suggestive of a birth cohort effect, with cohorts born between 1935 and 1950 experiencing rising incidence, and subsequent cohorts experiencing falling (females) or stable (males) incidence. Fourth, for each gender, there were similar rising trends across body sites, with the exception of the lower limb in females where incidence fell slightly. Finally, incidence rates were similar for thin (≤0.75 mm) and thick (>0.75 mm) melanomas within each gender, but their trends differed somewhat. Incidence of thin melanoma rose in all except the youngest age groups whereas incidence of thick melanoma rose only in the oldest age groups.

International trends

There are few other reports of recent (i.e., since the mid-1980s) melanoma incidence trends. Results from data in populations of mainly European origin suggest 2 major patterns: one of falling or stable incidence in recent years, seen first in the youngest age groups, and the other of continuing increases across all age groups.

The first pattern, as observed in New South Wales, was evident in Hawaiian whites as early as the beginning of the 1980s.1 It has also become apparent more recently in several other North American populations: Canada, since the mid-1980s4 and the U.S. areas in the SEER (Surveillance, Epidemiology and End Results) program since the early 1990s.13 In these populations, increases have generally been greater in males than in females (or decreases greater in females than in males), so that the male to female incidence rate ratio has risen. In both North America and Australasia, there is good evidence that the secular trends are the result of cohort patterns, with more recently born cohorts at lower risk.2–4, 14–16

The second pattern is evident in most populations of Europe. In regions with a large enough population base to produce stable age-specific incidence rates (in the age groups used here), rates were rising up to the time of the most recent data from Cancer Incidence in Five Continents (generally 1988–92) for both males and females.6, 9, 17 There was no consistency across regions in terms of which gender or age groups showed the greatest rate of increase.

There are even fewer data concerning recent trends according to body site and tumour thickness. For body site, 2 reports have findings similar to ours. For the U.S. SEER areas, Hall et al.13 reported proportional increases in incidence of 12% to 25% across sites for males and 3% to 15% for females between 1985–89 and 1990–94; the lowest rate of increase was for lower limb melanomas in females (3% rise). Gaudette and Gao4 noted a fall in the incidence of leg melanomas in women in recent years in Canada, whereas rates for other subsites in women seemed to be level. The actual size of these incidence changes cannot be determined because data were presented only graphically.

The incidence of both thin (≤0.75 mm) and thicker (>0.75 mm) melanomas increased in SEER areas between 1988 and 1994,13 but with much greater increases in males than females; in females, increases were greater for thin melanomas, whereas in males thicker melanomas increased more. Using different cut points on the same data set, Dennis18 reported slightly larger increases for lesions 1.0 mm or more thick than for thinner ones, for both genders combined.


The major cause of cutaneous malignant melanoma is exposure to the sun,19 with an intermittent pattern of exposure considered particularly harmful. The trend of continuously increasing incidence over several decades in populations of European origin, largely in a cohort fashion (with those born before the middle of the century at ever-increasing risk), has been attributed to changing sun exposure. There are several aspects of sun-related behavior that may have been responsible for increases. They include adoption of briefer clothing styles, including less hat use, and increasing recreational exposure to sun, both on vacations and at other times, with increasing affluence and correspondingly, a reduction in work-related sun exposure; the latter has been associated with lower than average risks of melanoma.20, 21

In New South Wales, there is evidence that the stabilization or fall in incidence has, like the increase, happened in a cohort-wise fashion, with those born since about 1950 at the same or lower risk compared with earlier cohorts. What changes have occurred to turn around the pattern of increase, particularly among women? It would seem reasonable to postulate either reduction in amount of sun exposure without a countervailing increase in intermittency, or a fall in intermittency of exposure without an increase in amount. Australia has had programs to increase public awareness about the hazards of sun exposure and to educate the public concerning use of effective sun-protective behaviors for nearly 20 years.22 There is evidence from repeated surveys starting in the late 1980s in the Australian state of Victoria that important shifts in behavior have occurred toward increased use of sun protection, including sunscreens,23 and reduced levels of sunburn.24 The latter may be an indication that people are either reducing their amount of sun exposure or moving away from the more hazardous intermittent pattern of exposure. Further support comes from the systematic observation of attendees at two major sporting events in Melbourne, Australia that noted that a high percentage of people were covered or had applied sunscreen; few, however, were considered to be optimally protected from the sun.25, 26 Of some interest is the anecdotal observation by Borland and Theobold that the Swedes in attendance at one of the sporting events (at least those who were obviously Swedes by their flags or face paint) had considerably less sun protection than the crowd average.26

It may be argued that it is too early to be seeing falls in melanoma incidence in people born after 1950 resulting from sun protection programs that started only in the 1970s or 80s, especially given the apparent importance of early life sun exposure on melanoma risk. There are a number of lines of evidence, however, pointing toward later stage effects of sun exposure on melanoma risk.27 We could postulate, perhaps, that early life exposure to the sun determines the lifetime potential for melanoma and later exposure determines how much of that potential is realized. The trends are consistent with that possibility but direct empirical data are lacking.

Why do women seem to have fared better than men? Empirically, changes in sunburn and protective behaviors have been greater in women than in men.24 It is also relevant to note that women have experienced major social changes in Australia and many other countries over the past 50 years. In particular, their participation in the workforce has greatly increased. Data on labour force participation in Australia between 1978 and 199528 show a substantial increase in participation by women 15–64 years of age (13.4 percentage points) beginning in 1984, with the largest increases being in women 25–54 years of age (between 15.2 and 20.4 percentage points). In relative terms, participation in these age groups increased by between 27% and 43%. Over the same period, men aged 15–64 years showed a small fall in participation (−2.8 percentage points); this fall was greatest in men 55–59 and 60–64 (−7.9 and −14.4 percentage points respectively). Both men and women 15–19 years of age showed a fall in participation over the period, greater in men (−6 percentage points) than in women (−0.6 percentage points). These data suggest an appreciable fall in opportunity for recreational sun exposure in women over the period of our study, due to competing work commitments, but a small increase in men, particularly in their 50s and 60s.

Additionally, special tabulations of data from the Australian Time Use Surveys of 1992 and 1997,29 supplied by the Australian Bureau of Statistics, indicate that at every age, men spent more time outdoors, during both employment and other activities. Somewhat trivially, perhaps, it may be noted that the fall in incidence of melanoma on the legs in women is consistent with the trend to much greater wearing of trousers by women over the past couple of decades.

Our results beg two other questions. First, why is the incidence of thick melanoma now increasing only in older people (65 years and older)? Presumably, this is a reflection of the cohort nature of changing incidence. Those who were recently aged 65 and over were born well before 1950 and so are now experiencing the higher rates of their cohort, for both thin and thick melanomas.

Second, do these data for New South Wales as a whole support the hypothesis, based on data from one part of the State, that increased and persisting high incidence rates in the late 1980s suggest increasing detection of a clinically inconspicuous form of melanoma with low metastatic potential?30 In the age-gender subgroups with the greatest increase in incidence, both thin and thick melanomas have been rising, whereas in those subgroups with declining incidence, both thin and thick melanomas have been falling. This does not support either a medical surveillance effect or a permanent increase in thin melanomas. The apparent persistence noted by Burton et al.30 may have been due to the effects of extended early detection, particularly after the airing of “Good-bye Sunshine”.31 “Good-bye Sunshine” was a television production that aired initially in late 1987 and featured a young man dying from melanoma.


Our results suggest that in New South Wales the increasing incidence of CMM is real (i.e., not due to increased surveillance and detection of thin lesions) but that this rising trend is gradually stabilizing or falling, particularly in females. We can therefore expect all ages trends for females, at least, to soon turn downward as those cohorts who are at lower risk age, carrying their reduced risk of melanoma with them. In males, it seems likely that incidence will stabilize, but it is not clear whether it will then fall as cohorts born since 1950 age. Because of the very strong increases in older men, it will be a longer time before overall incidence stabilizes in men. Thus, CMM will become even more of a male cancer.

Although the dramatic increases experienced by the oldest age groups are cause for concern, close examination of the age-specific trend lines (Fig. 2) suggests that even in these age groups the incidence trend may have flattened (or even started to decline) in the most recent few years. There is thus cause for optimism.

Complacency, however, is not warranted. The current incidence of melanoma is much higher than formerly because of the decades-long increases experienced by fair-skinned populations around the world.1 Concerted efforts to reduce sun exposure, particularly in children and youth, must continue.


This work would not have been possible but for the hard work and dedication to their task of all staff of the NSW Central Cancer Registry.