Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Female breast cancers are getting smaller, but socio‐demographic differences remain

Female breast cancers are getting smaller, but socio‐demographic differences remain Abstract Objective: To investigate trends towards early detection of infiltrating ductal carcinomas, possible effects on patients’ prognosis, and characteristics of women still at high risk of late detection. Methods: South Australian Cancer Registry data were analysed to compare breast tumour diameters for the 1980-86 and 1997-2002 diagnostic periods by age. Relative survivals for 1980-86 were compared with corresponding survival estimates for 1997-2000, obtained by weighting diameter-specific survivals for 1980-86 to equate with the diameter distribution for 1997-2002. Multivariable logistic regression was used to determine socio-demographic predictors of large diameters (≥30 mm) in 1997-2002. Results: The proportion of tumours with diameters smaller than 15 mm increased from 13.0% in 1980-86 to 36.7% in 19972002, whereas the proportion with large diameters reduced from 43.0% to 18.6%. Estimated changes in 20-year survivals equated with a 33% reduction in breastcancer mortality among patients aged 50-69 years at diagnosis. Data for 19972002 indicate that early diagnosis is not evenly distributed, with large diameters more common in age ranges outside the 50-69 year target for mammography screening; low socio-economic areas; nonCaucasians; patients born in northern/ eastern Europe and potentially Asia/Middle East; and in some country locations. Conclusions and implications: Increased emphasis on early detection should be directed at sectors of the population where delays in diagnosis and poorer prognosis are evident. Projected reductions in breastcancer mortality among patients are indicative of effects of earlier detection on patients’ prognosis, but require confirmation with follow-up data. More particularly, parallel studies of effects on populationbased mortality are warranted. (Aust N Z J Public Health 2004; 28: 312-16) Colin Luke, Anh-Minh Nguyen, Kevin Priest Epidemiology Branch, Department of Human Services, South Australia David Roder The Cancer Council South Australia ncreased emphasis has been placed on the early detection of breast cancer in Australian women over the past two decades. This was principally through the promotion of breast self-examination in the early 1980s, followed by the introduction of a National Program for the Early Detection of Breast Cancer from 1991.1-4 Although directed primarily at 50-69 year olds, the national program also has screened women aged 40-49 and 70-79 years. By 2000/01, approximately 57% of Australian women aged 50-69 years and 37% of those aged 40 years or more were participating in the national program of mammography screening (renamed BreastScreen).5 Corresponding proportions of South Australian women were 64% and 36% respectively.5 A Commonwealth-funded study of national breast-cancer data indicated that about 59% of Australian women diagnosed in 1997 had tumour diameters smaller than 20 mm.6 For cancers diagnosed through BreastScreen, 75% had diameters smaller than 20mm.6 National time trends could not be analysed in this study, however, since comparable data were not available for periods prior to 1997. By comparison, there were limited data for earlier years in South Australia. From 1980 to 1986, the State Cancer Registry recorded diameters for multiple population-based series of infiltrating ductal carcinomas of the breast.7,8 Corresponding data are now available from the registry for the period from 1997 to 2002. Infiltrating ductal carcinomas were the predominant histological type of invasive female breast cancer during these periods, accounting for 83% of all cancers. In addition, relative survivals of women diagnosed in 1980-86 were compared for periods up to 20 years from diagnosis with corresponding survival estimates for the 1997-2002 diagnostic period. The latter estimates were obtained by weighting diameter-specific survivals for 1980-86 cases according to the diameter distribution of carcinomas diagnosed in 1997-2002. The purpose was to investigate possible changes in prognosis of patients at diagnosis, rather than effects of earlier detection on population-based mortality. Tumour sizes for 1997-2002 also were analysed by age, socio-economic status, country of birth, race and region of residence to identify groups of women at elevated risk for whom additional emphasis on early detection may be warranted. Methods Data collection The South Australian Cancer Registry has received statutory notifications of cancer since 1977.9 The registry is population-based and covers all regions of the State. Its procedures have been described previously.9 The registry collects death data through routine notifications, electronic searching of official State death records and the National Death Index at the Australian Institute of Health and Welfare, and from interstate cancer registries.9 Under-ascertainment of death Submitted: February 2004 Revision requested: April 2004 Accepted: May 2004 Correspondence to: Dr Colin Luke, Epidemiology Branch, Department of Human Services, PO Box 6, Rundle Mall, Adelaide, South Australia 5001. Fax: (08) 8226 6291; e-mail: Colin.Luke@dhs.sa.gov.au AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 4 Cancer Socio-demographic differences in breast cancer information has been checked periodically through active followup and with deaths reported independently and found to be minimal, with little effect on calculated case survivals.9,10 The present study included 4,912 infiltrating ductal carcinomas of the female breast (SNOMED II histological code 85003), diagnosed in 1980-86 or 1997-2002.11 They comprised all infiltrating ductal carcinomas with a recorded tumour diameter. This information was not recorded in the intervening years. Socio-demographic descriptors used in the study were: age at diagnosis; region of residence, classified as metropolitan (four metropolitan subdivisions of the State capital of Adelaide) or nonmetropolitan (16 statistical subdivisions outside Adelaide);9 country of birth (expressed as Australia; the United Kingdom/Ireland; other English-speaking countries; southern Europe – mostly Italy, Malta, Greece, former Yugoslav states, or other; northern/eastern Europe; Asia/Middle East; or other); race (classified as Caucasian, Aboriginal, Asian or other); and place of residence (coded into four ordinal categories by socio-economic status of postcode, using the SEIFA index12). Diameters were obtained from the microscopic section of the pathology report, as described previously.6,7 This applied to 3,777 infiltrating ductal carcinomas (83% of the total) diagnosed in 1997-2002. Cancers with recorded diameters appeared representative of the total, in that they did not differ from those without a recorded diameter by the woman’s metropolitan or non-metropolitan place of residence (p=0.803); socio-economic status (p=0.153);12 race expressed as Caucasian or non-Caucasian (p=0.551); country of birth (p=0.482); or age in the age range under 80 years (p=0.850). However, a lower proportion of women with recorded diameters were aged 80 years or more (p<0.001). Diameters of infiltrating ductal carcinomas were similarly recorded during 1980-86.4 This was undertaken for a series of consecutively notified carcinomas during periods when staffing levels permitted this practice. Women with recorded diameters appeared representative of the total, in that they did not differ from women without recorded diameters by place of residence (p=0.144); socioeconomic status (p=0.578);12 race (p=0.456); country of birth (p=0.346); or age in the age range under 80 years (p=0.906). As for 1997-2002, a lower proportion with recorded diameters were aged 80 years or more (p<0.001). (Spearman correlation); race (summarised as Caucasian or nonCaucasian, due to small numbers in non-Caucasian subcategories); and metropolitan or non-metropolitan place of residence (Mann-Whitney U test).13,14 A multiple logistic regression analysis also was undertaken of 1997-2002 data in which presence or absence of a large tumour diameter (≥30 mm) was the outcome and age, socio-economic status, country of birth and race were entered as predictors, with backwards elimination of those whose elimination did not reduce model fit (p>0.050 for change in chi-square goodness-of-fit).13 Model requirements, including collinearity assumptions, were checked and found to be met.13,14 A Hosmer-Lemeshow chi-square of 2.15 (df=4) indicated that the model was well calibrated (p=0.707).13 Survivals Five, 10, 15 and 20 year relative survivals (+/- standard error) were calculated for carcinomas diagnosed in 1980-86 by tumour size.9,13 Life tables by calendar year for all South Australian women in the corresponding age categories (i.e. 0-4, 5-9 . . . ≥85 years) were used to produce these relative survivals. Probabilities of surviving breast cancer were inferred from these survivals. Actual relative survivals for all 1980-86 cases with a recorded tumour size were compared with estimated survivals for 19972002. The estimates for 1997-2002 were obtained by weighting 1980-86 diameter-specific survivals according to the diameter distribution for 1997-2002.14 Results Diameters Bivariate analyses Statistical analysis A de-identified registry file was extracted and analysed inhouse, with legal authority from Section 42a of the Public and Environmental Health Act, using STATA 8.0 software.13 Two sets of analyses were undertaken, the first directed at socio-demographic risk indicators of larger tumour diameters during 19972002, and the second at secular gains in survival between the 1980-86 and 1997-2002 diagnostic periods. Diameters Diameter distributions during 1997-2002 were compared by age at diagnosis, country of birth, and statistical subdivision of residence (Kruskal-Wallis ANOVA); socio-economic status 2004 VOL. 28 NO. 4 Tumour size differed by age at diagnosis, socio-economic status, country of birth, and race, but not by metropolitan/non-metropolitan place of residence (see Table 1). The proportion of cancers with diameters of 30 mm or more was higher in age groups under 50 years (21.6%) and in women aged 70 years or more (22.0%) than among 50-69 year olds (15.7%); in low socioeconomic areas (22.6%) than other localities (17.1%); in women born in northern/eastern Europe or Asia/Middle East (28.7%) than elsewhere (18.1%); and in non-Caucasians (32.7%) than Caucasians (18.7%). Data for non-Caucasian women are presented collectively in Table 1 because most were Asian and no difference was evident in diameter distribution between Asian and other non-Caucasian women (Mann-Whitney p=0.523). More specifically, mean diameters (+/- standard errors) were 24.3 mm (+/-2.1) for 44 Asian women, and 27.8 mm (+/-4.7) for four Aboriginal women. In addition, there was one female classified to ‘other’ race who presented a tumour diameter of 24.0 mm. Further analyses by statistical subdivision did not indicate differences in diameter distribution in the metropolitan area (MannWhitney p=0.189). A difference was suggested in the non-metropolitan area (Mann-Whitney p=0.075), with the Yorke AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Luke et al. Article Subdivision having a higher proportion of cancers with diameters of 30 mm or more (29.2%; 26 of 89) than other subdivisions (18.8%; 157 of 836). Multivariable analyses Multiple logistic regression indicated that the risk of a large tumour (diameter ≥30 mm) was higher for women under 50 years of age, and those aged 70 years or more, than for 50-69 year olds; women residing in low than higher socio-economic areas; women born in northern/eastern Europe; and non-Caucasians (see Table 2). This model was a result of the backwards elimination process. When the analysis was repeated, retaining all four socio-economic categories and using high socio-economic status as the reference category, the relative odds (95% confidence limits) was 1.02 (0.771.36) for mid-high, 1.18 (0.92-1.50) for mid-low, and 1.44 (1.151.80) for low socio-economic status. It was not possible, in addition, to retain subcategories of non-Caucasian race because of small numbers. When the analysis in Table 2 was repeated, excluding socioeconomic status of residential area and substituting statistical subdivision of residential area (as dummy variables), the same variables were retained as significant predictors (i.e. age, race, and country of birth (p≤0.015), plus the Yorke Statistical Subdivision (p=0.008). The relative odds (95% confidence limits) of a large diameter for women resident in Yorke, as compared with other statistical subdivisions, were 1.89 (1.18-3.02). Survivals Twenty-year relative survivals for women diagnosed in 198086 ranged from a high of 88.1%, when tumour diameters were under 10 mm, to a low of 25.0% when tumour diameters were 50 mm or more, with a collective survival for all diameters combined of 48.9% (see Table 3). Diameter distributions differed between 1980-86 and 1997-2002 (Mann-Whitney p<0.001). The proportion smaller than 15 mm increased from 13.0% in 1980-86 to 36.7% in 1997-2002, while a corresponding reduction occurred in the proportion of large cancers (≥30 mm) from 43.0% to 18.6%. When diameter-specific survivals for 1980-86 were weighted according to the diameter distribution for 1997-2002 (see Table 1), the weighted 20-year survival for women diagnosed in 19972002 was 60.4%, which exceeded the corresponding figure of 48.9% for 1980-86 (see Table 4). This would equate with a reduction in case fatality of 22.5%. This process was repeated independently for each age stratum, Table 1: Percentage distribution of infiltrating ductal carcinomas of the female breast by diameter and sociodemographic characteristic; South Australia, diagnostic period circa 1997-2002. Characteristic Age at diagnosis (years) Under 50 (n=962) 50-69 (n=1,973) 70+ (n=842) Socio-economic statusb Low (n=1,036) Mid-low (n=835) Mid-high (n=731) High (n=1,175) Country of birthc Australia (n=2,545) United Kingdom/Ireland (n=558) Other (English speaking) (n=37) Southern Europe (n=165) Northern/eastern Europe (n=183) Asia/Middle East (n=82) Other (n=25) Raced Caucasian (n=3,539) Non-Caucasian (n=49) Place of residence Metropolitan (n=2,852) Non-metropolitan (n=925) Under 10 (n=550) 10.2 18.3 10.7 (n=550) 13.2 13.4 15.7 15.8 (n=512) 14.3 15.2 21.6 15.2 8.7 11.0 20.0 (n=515) 14.4 8.2 (n=550) 14.9 13.5 10-14 (n=836) 18.1 25.0 20.1 (n=836) 21.5 22.0 21.9 22.9 (n=786) 22.4 20.1 24.3 20.0 22.4 23.2 8.0 (n=774) 21.6 16.3 (n=836) 22.4 21.3 15-19 (n=746) 22.2 19.4 17.8 (n=746) 18.9 21.0 18.3 20.5 (n=715) 20.3 19.2 27.0 17.6 19.1 18.3 8.0 (n=714) 19.9 18.4 (n=746) 19.5 20.4 Diameter (mm) 20-29 30-39 (n=942) 27.9 21.6 29.5 (n=942) 23.7 25.1 27.1 24.5 (n=904) 24.8 27.2 21.6 29.7 21.9 17.1 40.0 (n=906) 25.3 24.5 (n=942) 24.9 25.0 (n=373) 11.7 8.2 11.6 (n=373) 12.2 9.8 8.1 9.0 (n=358) 9.6 9.0 5.4 12.7 12.6 15.9 16.0 (n=358) 9.9 14.3 (n=373) 10.0 9.6 40-49 (n=143) 4.9 2.9 4.6 (n=143) 4.4 3.4 3.4 3.7 (n139) 3.9 2.5 0.0 3.6 4.4 13.4 8.0 (n=140) 3.7 16.3 (n=143) 3.8 3.7 50+ (n=187) 5.0 4.6 5.7 (n=187) 6.0 5.3 5.5 3.5 (n=181) 4.7 6.8 0.0 1.2 10.9 1.2 0.0 (n=181) 5.1 2.0 (n=187) 4.5 6.5 p valuea KW p<0.001 Sp p<0.001 KW p=0.015 MW p=0.020 MW p=0.130 Notes: (a) KW = Kruskal-Wallis ANOVA; Sp = Spearman correlation; MW = Mann-Whitney U test.13, 14 (b) Derived from postcode characteristics (see text).14 (c) Excludes 4.8% of cancers (n=182) due to missing values. (d) Excludes 5.0% of cancers (n=189) due to missing values. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 4 Cancer Socio-demographic differences in breast cancer giving 20-year survival figures for 1980-86 and 1997-2000 respectively of 58.2% and 67.1% for women under 50 years at diagnosis; 57.8% and 71.7% for 50-69 year olds; and 35.9% and 44.0% for women aged 70 years or more (see Table 4). Confidence ranges of 20-year survivals overlapped for women in individual age categories, but not for all ages combined. Table 2: Relative odds (95% confidence limits) of infiltrating ductal carcinomas of the female breast being diagnosed when large (diameter ≥30 mm) by sociodemographic characteristic: South Australia, diagnostic period 1997-2002.a Characteristic Multiple logistic regression Relative odds 1.43 (1.17-1.75) 1.00 1.49 (1.21-1.84) 1.00 1.35 (1.13-1.63) 1.00 1.66 (1.18-2.32) 1.00 2.07 (1.13-3.81) Discussion Results indicate a marked reduction in size of infiltrating ductal carcinomas between the 1980-86 and 1997-2002 diagnostic periods. Were diameter-specific survivals for 1980-86 to apply for the 1997-2002 diagnostic period, this size reduction would be associated with a reduction in case fatality from breast cancer during the 20 years from diagnosis of about 23%. Among 50-69 year olds, the target age range for mammography screening, the reduction would be about 33%, whereas it would approximate 21% for younger women and 13% for those aged 70 years or more. The reductions in case fatality computed between 1980-86 and 1997-2002 are indicative of the improvements in prognosis that might be expected in breast-cancer patients as a result of earlier detection. Allowances were not made in these calculations for any effects of treatment advances, as may have resulted, for example, from adjuvant therapy.15,16 Also, there is evidence from a recent study of higher survivals from mammographically detected than other tumours of equivalent stage, which may affect future survival trends.17 Survival trends may not translate directly into population-based mortality reductions, since this would depend on differences in lead-time, length-time, and related effects that may have occurred between diagnostic periods.4 The Commonwealth Government is planning to fund an evaluation study to investigate these population-based screening effects more directly. The present data for 1997-2002 indicate that early diagnosis is not distributed equally in the population. Women outside the target age range for mammography screening appear to be disadvantaged in this regard. An earlier analysis did not show this age Age at diagnosis (years) Under 50 (n=911) 50-69 (reference) (n=1,845) 70+ (n=781) Socio-economic statusb Not low (reference) (n=2,560) Low (n=977) Country of birth Other (reference) (n=3,354) Northern/eastern Europe (n=183) Race Caucasian (n=3,488) Non-Caucasian (n=49) Notes: (a) Comparison group = cancers with diameters of under 30 mm (n=2,865). Excludes 6.4% of cases (n=240) due to missing values on race or country of birth. (b) Derived from postcode characteristics (see text).12 pattern for 1980-86.16 The rationale for excluding these age groups from the screening target will warrant ongoing re-evaluation as new research evidence presents. Meanwhile, low socio-economic groups and non-Caucasian women also appear to be disadvantaged and may warrant special emphasis in early-detection promotion. In addition, there is evidence that women born in northern/ eastern Europe or potentially in Asia/Middle East, and those residing in some country locations, may be at increased risk. Race is generally regarded as a poorly recorded data item in Australian cancer registries. This is reflected in the approximate 5% of women with missing values for this item in the present study. While the potential for bias from missing values, or unreliable recording, cannot be discounted, we suspect that this would Table 3: Relative survivals by percentage (95% confidence limits) for infiltrating ductal carcinomas of the female breast by period from diagnosis; South Australia, diagnostic period 1980-86.a Diameter (mm) Under 10 (n=30) 10-14 (n=117) 15-19 (n=173) 20-29 (n=327) 30-39 (n=183) 40-49 (n=109) 50+ (n=196) Total (n=1,135) 100 95.3 (89.5-100) 88.4 (82.1-94.7) 83.0 (78.1-87.9) 79.4 (72.4-86.3) 72.7 (63.1-82.2) 51.5 (43.9-59.1) 78.7 (75.9-81.5) Period from diagnosis (years) 10 15 100 84.7 (75.9-93.6) 76.5 (68.3-84.7) 64.9 (58.8-71.0) 58.6 (50.3-67.0) 59.5 (48.6-70.4) 36.5 (29.0-44.1) 63.4 (60.0-66.7) 88.1 (69.2-100) 76.7 (65.9-87.5) 69.4 (59.9-78.9) 52.9 (46.2-59.5) 52.7 (43.8-61.6) 44.1 (32.8-55.4) 30.8 (23.2-38.3) 54.2 (50.5-57.8) 88.1 (64.4-100) 70.1 (55.7-84.5) 61.9 (49.2-74.5) 50.6 (42.8-58.4) 47.1 (36.3-57.9) 35.1 (21.1-49.1) 25.0 (17.0-33.0) 48.9 (44.5-53.3) Note: (a) Date of censoring: 31 December 2001. 2004 VOL. 28 NO. 4 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Luke et al. Article Table 4: Actual relative survivals by percentage (95% confidence limits) for infiltrating ductal carcinomas of the female breast diagnosed in 1980-86 compared with estimated relative survivals (95% confidence limits) for corresponding cancers diagnosed in 1997-2002; South Australia.a Age at diagnosis (years) All Under 50 50-69 70+ Diagnostic period 1980-86 (n=1,135) 1997-2002 (n=3,777) 1980-86 (n=293) 1997-2002 (n=962) 1980-86 (n=580) 1997-2002 (n=1,973) 1980-86 (n=262) 1997-2002 (n=842) 78.7 (75.9-81.5) 87.6 (85.3-90.0) 79.5 (74.8-84.2) 85.8 (81.9-89.7) 79.7 (76.1-83.4) 89.3 (86.0-92.6) 77.1 (69.4-84.7) 84.8 (76.3-93.2) Period from diagnosis (years) 10 15 63.4 (60.0-66.7) 74.9 (71.4-78.3) 69.4 (63.9-74.9) 77.6 (72.8-82.5) 65.5 (61.0-70.0) 77.6 (72.4-82.9) 59.7 (49.8-69.6) 67.6 (55.4-79.8) 54.2 (50.5-57.8) 65.0 (60.5-69.6) 61.6 (55.7-67.5) 70.0 (64.2-75.7) 60.5 (55.4-65.7) 74.5 (67.9-81.0) 42.6 (30.6-54.6) 44.3 (29.9-58.7) 48.9 (44.5-53.3) 60.4 (54.6-66.2) 58.2 (51.7-64.7) 67.1 (60.3-73.9) 57.8 (51.0-64.6) 71.7 (61.7-81.8) 35.9 (13.7-58.2) 44.0 (20.1-68.0) Notes: (a) Date of censoring for actual survivals: 31 December 2001. Estimated survivals calculated by weighting actual survivals for 1980-86 to the 1997-2002 diameter distribution (see text).13,14 have been haphazard and would tend to have diminished rather than increased differences in recorded diameter by race. The present data revealed an elevated proportion of large tumours among women from the Yorke Statistical Subdivision. This is predominantly a rural area approximately 200 kilometres from Adelaide,9 which may warrant additional attention in the promotion of breast screening and other early-detection initiatives. The results of this study indicate the extent to which earlier diagnosis is occurring in South Australia, possible prognostic effects, and sections of the population at elevated risk. These findings warrant confirmation with follow-up data and data from other Australian registries. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Female breast cancers are getting smaller, but socio‐demographic differences remain

Download PDF
5 pages

Loading next page...
 
/lp/wiley/female-breast-cancers-are-getting-smaller-but-socio-demographic-ncttUiD85m

References (9)

Publisher
Wiley
Copyright
Copyright © 2004 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.2004.tb00435.x
Publisher site
See Article on Publisher Site

Abstract

Abstract Objective: To investigate trends towards early detection of infiltrating ductal carcinomas, possible effects on patients’ prognosis, and characteristics of women still at high risk of late detection. Methods: South Australian Cancer Registry data were analysed to compare breast tumour diameters for the 1980-86 and 1997-2002 diagnostic periods by age. Relative survivals for 1980-86 were compared with corresponding survival estimates for 1997-2000, obtained by weighting diameter-specific survivals for 1980-86 to equate with the diameter distribution for 1997-2002. Multivariable logistic regression was used to determine socio-demographic predictors of large diameters (≥30 mm) in 1997-2002. Results: The proportion of tumours with diameters smaller than 15 mm increased from 13.0% in 1980-86 to 36.7% in 19972002, whereas the proportion with large diameters reduced from 43.0% to 18.6%. Estimated changes in 20-year survivals equated with a 33% reduction in breastcancer mortality among patients aged 50-69 years at diagnosis. Data for 19972002 indicate that early diagnosis is not evenly distributed, with large diameters more common in age ranges outside the 50-69 year target for mammography screening; low socio-economic areas; nonCaucasians; patients born in northern/ eastern Europe and potentially Asia/Middle East; and in some country locations. Conclusions and implications: Increased emphasis on early detection should be directed at sectors of the population where delays in diagnosis and poorer prognosis are evident. Projected reductions in breastcancer mortality among patients are indicative of effects of earlier detection on patients’ prognosis, but require confirmation with follow-up data. More particularly, parallel studies of effects on populationbased mortality are warranted. (Aust N Z J Public Health 2004; 28: 312-16) Colin Luke, Anh-Minh Nguyen, Kevin Priest Epidemiology Branch, Department of Human Services, South Australia David Roder The Cancer Council South Australia ncreased emphasis has been placed on the early detection of breast cancer in Australian women over the past two decades. This was principally through the promotion of breast self-examination in the early 1980s, followed by the introduction of a National Program for the Early Detection of Breast Cancer from 1991.1-4 Although directed primarily at 50-69 year olds, the national program also has screened women aged 40-49 and 70-79 years. By 2000/01, approximately 57% of Australian women aged 50-69 years and 37% of those aged 40 years or more were participating in the national program of mammography screening (renamed BreastScreen).5 Corresponding proportions of South Australian women were 64% and 36% respectively.5 A Commonwealth-funded study of national breast-cancer data indicated that about 59% of Australian women diagnosed in 1997 had tumour diameters smaller than 20 mm.6 For cancers diagnosed through BreastScreen, 75% had diameters smaller than 20mm.6 National time trends could not be analysed in this study, however, since comparable data were not available for periods prior to 1997. By comparison, there were limited data for earlier years in South Australia. From 1980 to 1986, the State Cancer Registry recorded diameters for multiple population-based series of infiltrating ductal carcinomas of the breast.7,8 Corresponding data are now available from the registry for the period from 1997 to 2002. Infiltrating ductal carcinomas were the predominant histological type of invasive female breast cancer during these periods, accounting for 83% of all cancers. In addition, relative survivals of women diagnosed in 1980-86 were compared for periods up to 20 years from diagnosis with corresponding survival estimates for the 1997-2002 diagnostic period. The latter estimates were obtained by weighting diameter-specific survivals for 1980-86 cases according to the diameter distribution of carcinomas diagnosed in 1997-2002. The purpose was to investigate possible changes in prognosis of patients at diagnosis, rather than effects of earlier detection on population-based mortality. Tumour sizes for 1997-2002 also were analysed by age, socio-economic status, country of birth, race and region of residence to identify groups of women at elevated risk for whom additional emphasis on early detection may be warranted. Methods Data collection The South Australian Cancer Registry has received statutory notifications of cancer since 1977.9 The registry is population-based and covers all regions of the State. Its procedures have been described previously.9 The registry collects death data through routine notifications, electronic searching of official State death records and the National Death Index at the Australian Institute of Health and Welfare, and from interstate cancer registries.9 Under-ascertainment of death Submitted: February 2004 Revision requested: April 2004 Accepted: May 2004 Correspondence to: Dr Colin Luke, Epidemiology Branch, Department of Human Services, PO Box 6, Rundle Mall, Adelaide, South Australia 5001. Fax: (08) 8226 6291; e-mail: Colin.Luke@dhs.sa.gov.au AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 4 Cancer Socio-demographic differences in breast cancer information has been checked periodically through active followup and with deaths reported independently and found to be minimal, with little effect on calculated case survivals.9,10 The present study included 4,912 infiltrating ductal carcinomas of the female breast (SNOMED II histological code 85003), diagnosed in 1980-86 or 1997-2002.11 They comprised all infiltrating ductal carcinomas with a recorded tumour diameter. This information was not recorded in the intervening years. Socio-demographic descriptors used in the study were: age at diagnosis; region of residence, classified as metropolitan (four metropolitan subdivisions of the State capital of Adelaide) or nonmetropolitan (16 statistical subdivisions outside Adelaide);9 country of birth (expressed as Australia; the United Kingdom/Ireland; other English-speaking countries; southern Europe – mostly Italy, Malta, Greece, former Yugoslav states, or other; northern/eastern Europe; Asia/Middle East; or other); race (classified as Caucasian, Aboriginal, Asian or other); and place of residence (coded into four ordinal categories by socio-economic status of postcode, using the SEIFA index12). Diameters were obtained from the microscopic section of the pathology report, as described previously.6,7 This applied to 3,777 infiltrating ductal carcinomas (83% of the total) diagnosed in 1997-2002. Cancers with recorded diameters appeared representative of the total, in that they did not differ from those without a recorded diameter by the woman’s metropolitan or non-metropolitan place of residence (p=0.803); socio-economic status (p=0.153);12 race expressed as Caucasian or non-Caucasian (p=0.551); country of birth (p=0.482); or age in the age range under 80 years (p=0.850). However, a lower proportion of women with recorded diameters were aged 80 years or more (p<0.001). Diameters of infiltrating ductal carcinomas were similarly recorded during 1980-86.4 This was undertaken for a series of consecutively notified carcinomas during periods when staffing levels permitted this practice. Women with recorded diameters appeared representative of the total, in that they did not differ from women without recorded diameters by place of residence (p=0.144); socioeconomic status (p=0.578);12 race (p=0.456); country of birth (p=0.346); or age in the age range under 80 years (p=0.906). As for 1997-2002, a lower proportion with recorded diameters were aged 80 years or more (p<0.001). (Spearman correlation); race (summarised as Caucasian or nonCaucasian, due to small numbers in non-Caucasian subcategories); and metropolitan or non-metropolitan place of residence (Mann-Whitney U test).13,14 A multiple logistic regression analysis also was undertaken of 1997-2002 data in which presence or absence of a large tumour diameter (≥30 mm) was the outcome and age, socio-economic status, country of birth and race were entered as predictors, with backwards elimination of those whose elimination did not reduce model fit (p>0.050 for change in chi-square goodness-of-fit).13 Model requirements, including collinearity assumptions, were checked and found to be met.13,14 A Hosmer-Lemeshow chi-square of 2.15 (df=4) indicated that the model was well calibrated (p=0.707).13 Survivals Five, 10, 15 and 20 year relative survivals (+/- standard error) were calculated for carcinomas diagnosed in 1980-86 by tumour size.9,13 Life tables by calendar year for all South Australian women in the corresponding age categories (i.e. 0-4, 5-9 . . . ≥85 years) were used to produce these relative survivals. Probabilities of surviving breast cancer were inferred from these survivals. Actual relative survivals for all 1980-86 cases with a recorded tumour size were compared with estimated survivals for 19972002. The estimates for 1997-2002 were obtained by weighting 1980-86 diameter-specific survivals according to the diameter distribution for 1997-2002.14 Results Diameters Bivariate analyses Statistical analysis A de-identified registry file was extracted and analysed inhouse, with legal authority from Section 42a of the Public and Environmental Health Act, using STATA 8.0 software.13 Two sets of analyses were undertaken, the first directed at socio-demographic risk indicators of larger tumour diameters during 19972002, and the second at secular gains in survival between the 1980-86 and 1997-2002 diagnostic periods. Diameters Diameter distributions during 1997-2002 were compared by age at diagnosis, country of birth, and statistical subdivision of residence (Kruskal-Wallis ANOVA); socio-economic status 2004 VOL. 28 NO. 4 Tumour size differed by age at diagnosis, socio-economic status, country of birth, and race, but not by metropolitan/non-metropolitan place of residence (see Table 1). The proportion of cancers with diameters of 30 mm or more was higher in age groups under 50 years (21.6%) and in women aged 70 years or more (22.0%) than among 50-69 year olds (15.7%); in low socioeconomic areas (22.6%) than other localities (17.1%); in women born in northern/eastern Europe or Asia/Middle East (28.7%) than elsewhere (18.1%); and in non-Caucasians (32.7%) than Caucasians (18.7%). Data for non-Caucasian women are presented collectively in Table 1 because most were Asian and no difference was evident in diameter distribution between Asian and other non-Caucasian women (Mann-Whitney p=0.523). More specifically, mean diameters (+/- standard errors) were 24.3 mm (+/-2.1) for 44 Asian women, and 27.8 mm (+/-4.7) for four Aboriginal women. In addition, there was one female classified to ‘other’ race who presented a tumour diameter of 24.0 mm. Further analyses by statistical subdivision did not indicate differences in diameter distribution in the metropolitan area (MannWhitney p=0.189). A difference was suggested in the non-metropolitan area (Mann-Whitney p=0.075), with the Yorke AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Luke et al. Article Subdivision having a higher proportion of cancers with diameters of 30 mm or more (29.2%; 26 of 89) than other subdivisions (18.8%; 157 of 836). Multivariable analyses Multiple logistic regression indicated that the risk of a large tumour (diameter ≥30 mm) was higher for women under 50 years of age, and those aged 70 years or more, than for 50-69 year olds; women residing in low than higher socio-economic areas; women born in northern/eastern Europe; and non-Caucasians (see Table 2). This model was a result of the backwards elimination process. When the analysis was repeated, retaining all four socio-economic categories and using high socio-economic status as the reference category, the relative odds (95% confidence limits) was 1.02 (0.771.36) for mid-high, 1.18 (0.92-1.50) for mid-low, and 1.44 (1.151.80) for low socio-economic status. It was not possible, in addition, to retain subcategories of non-Caucasian race because of small numbers. When the analysis in Table 2 was repeated, excluding socioeconomic status of residential area and substituting statistical subdivision of residential area (as dummy variables), the same variables were retained as significant predictors (i.e. age, race, and country of birth (p≤0.015), plus the Yorke Statistical Subdivision (p=0.008). The relative odds (95% confidence limits) of a large diameter for women resident in Yorke, as compared with other statistical subdivisions, were 1.89 (1.18-3.02). Survivals Twenty-year relative survivals for women diagnosed in 198086 ranged from a high of 88.1%, when tumour diameters were under 10 mm, to a low of 25.0% when tumour diameters were 50 mm or more, with a collective survival for all diameters combined of 48.9% (see Table 3). Diameter distributions differed between 1980-86 and 1997-2002 (Mann-Whitney p<0.001). The proportion smaller than 15 mm increased from 13.0% in 1980-86 to 36.7% in 1997-2002, while a corresponding reduction occurred in the proportion of large cancers (≥30 mm) from 43.0% to 18.6%. When diameter-specific survivals for 1980-86 were weighted according to the diameter distribution for 1997-2002 (see Table 1), the weighted 20-year survival for women diagnosed in 19972002 was 60.4%, which exceeded the corresponding figure of 48.9% for 1980-86 (see Table 4). This would equate with a reduction in case fatality of 22.5%. This process was repeated independently for each age stratum, Table 1: Percentage distribution of infiltrating ductal carcinomas of the female breast by diameter and sociodemographic characteristic; South Australia, diagnostic period circa 1997-2002. Characteristic Age at diagnosis (years) Under 50 (n=962) 50-69 (n=1,973) 70+ (n=842) Socio-economic statusb Low (n=1,036) Mid-low (n=835) Mid-high (n=731) High (n=1,175) Country of birthc Australia (n=2,545) United Kingdom/Ireland (n=558) Other (English speaking) (n=37) Southern Europe (n=165) Northern/eastern Europe (n=183) Asia/Middle East (n=82) Other (n=25) Raced Caucasian (n=3,539) Non-Caucasian (n=49) Place of residence Metropolitan (n=2,852) Non-metropolitan (n=925) Under 10 (n=550) 10.2 18.3 10.7 (n=550) 13.2 13.4 15.7 15.8 (n=512) 14.3 15.2 21.6 15.2 8.7 11.0 20.0 (n=515) 14.4 8.2 (n=550) 14.9 13.5 10-14 (n=836) 18.1 25.0 20.1 (n=836) 21.5 22.0 21.9 22.9 (n=786) 22.4 20.1 24.3 20.0 22.4 23.2 8.0 (n=774) 21.6 16.3 (n=836) 22.4 21.3 15-19 (n=746) 22.2 19.4 17.8 (n=746) 18.9 21.0 18.3 20.5 (n=715) 20.3 19.2 27.0 17.6 19.1 18.3 8.0 (n=714) 19.9 18.4 (n=746) 19.5 20.4 Diameter (mm) 20-29 30-39 (n=942) 27.9 21.6 29.5 (n=942) 23.7 25.1 27.1 24.5 (n=904) 24.8 27.2 21.6 29.7 21.9 17.1 40.0 (n=906) 25.3 24.5 (n=942) 24.9 25.0 (n=373) 11.7 8.2 11.6 (n=373) 12.2 9.8 8.1 9.0 (n=358) 9.6 9.0 5.4 12.7 12.6 15.9 16.0 (n=358) 9.9 14.3 (n=373) 10.0 9.6 40-49 (n=143) 4.9 2.9 4.6 (n=143) 4.4 3.4 3.4 3.7 (n139) 3.9 2.5 0.0 3.6 4.4 13.4 8.0 (n=140) 3.7 16.3 (n=143) 3.8 3.7 50+ (n=187) 5.0 4.6 5.7 (n=187) 6.0 5.3 5.5 3.5 (n=181) 4.7 6.8 0.0 1.2 10.9 1.2 0.0 (n=181) 5.1 2.0 (n=187) 4.5 6.5 p valuea KW p<0.001 Sp p<0.001 KW p=0.015 MW p=0.020 MW p=0.130 Notes: (a) KW = Kruskal-Wallis ANOVA; Sp = Spearman correlation; MW = Mann-Whitney U test.13, 14 (b) Derived from postcode characteristics (see text).14 (c) Excludes 4.8% of cancers (n=182) due to missing values. (d) Excludes 5.0% of cancers (n=189) due to missing values. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 4 Cancer Socio-demographic differences in breast cancer giving 20-year survival figures for 1980-86 and 1997-2000 respectively of 58.2% and 67.1% for women under 50 years at diagnosis; 57.8% and 71.7% for 50-69 year olds; and 35.9% and 44.0% for women aged 70 years or more (see Table 4). Confidence ranges of 20-year survivals overlapped for women in individual age categories, but not for all ages combined. Table 2: Relative odds (95% confidence limits) of infiltrating ductal carcinomas of the female breast being diagnosed when large (diameter ≥30 mm) by sociodemographic characteristic: South Australia, diagnostic period 1997-2002.a Characteristic Multiple logistic regression Relative odds 1.43 (1.17-1.75) 1.00 1.49 (1.21-1.84) 1.00 1.35 (1.13-1.63) 1.00 1.66 (1.18-2.32) 1.00 2.07 (1.13-3.81) Discussion Results indicate a marked reduction in size of infiltrating ductal carcinomas between the 1980-86 and 1997-2002 diagnostic periods. Were diameter-specific survivals for 1980-86 to apply for the 1997-2002 diagnostic period, this size reduction would be associated with a reduction in case fatality from breast cancer during the 20 years from diagnosis of about 23%. Among 50-69 year olds, the target age range for mammography screening, the reduction would be about 33%, whereas it would approximate 21% for younger women and 13% for those aged 70 years or more. The reductions in case fatality computed between 1980-86 and 1997-2002 are indicative of the improvements in prognosis that might be expected in breast-cancer patients as a result of earlier detection. Allowances were not made in these calculations for any effects of treatment advances, as may have resulted, for example, from adjuvant therapy.15,16 Also, there is evidence from a recent study of higher survivals from mammographically detected than other tumours of equivalent stage, which may affect future survival trends.17 Survival trends may not translate directly into population-based mortality reductions, since this would depend on differences in lead-time, length-time, and related effects that may have occurred between diagnostic periods.4 The Commonwealth Government is planning to fund an evaluation study to investigate these population-based screening effects more directly. The present data for 1997-2002 indicate that early diagnosis is not distributed equally in the population. Women outside the target age range for mammography screening appear to be disadvantaged in this regard. An earlier analysis did not show this age Age at diagnosis (years) Under 50 (n=911) 50-69 (reference) (n=1,845) 70+ (n=781) Socio-economic statusb Not low (reference) (n=2,560) Low (n=977) Country of birth Other (reference) (n=3,354) Northern/eastern Europe (n=183) Race Caucasian (n=3,488) Non-Caucasian (n=49) Notes: (a) Comparison group = cancers with diameters of under 30 mm (n=2,865). Excludes 6.4% of cases (n=240) due to missing values on race or country of birth. (b) Derived from postcode characteristics (see text).12 pattern for 1980-86.16 The rationale for excluding these age groups from the screening target will warrant ongoing re-evaluation as new research evidence presents. Meanwhile, low socio-economic groups and non-Caucasian women also appear to be disadvantaged and may warrant special emphasis in early-detection promotion. In addition, there is evidence that women born in northern/ eastern Europe or potentially in Asia/Middle East, and those residing in some country locations, may be at increased risk. Race is generally regarded as a poorly recorded data item in Australian cancer registries. This is reflected in the approximate 5% of women with missing values for this item in the present study. While the potential for bias from missing values, or unreliable recording, cannot be discounted, we suspect that this would Table 3: Relative survivals by percentage (95% confidence limits) for infiltrating ductal carcinomas of the female breast by period from diagnosis; South Australia, diagnostic period 1980-86.a Diameter (mm) Under 10 (n=30) 10-14 (n=117) 15-19 (n=173) 20-29 (n=327) 30-39 (n=183) 40-49 (n=109) 50+ (n=196) Total (n=1,135) 100 95.3 (89.5-100) 88.4 (82.1-94.7) 83.0 (78.1-87.9) 79.4 (72.4-86.3) 72.7 (63.1-82.2) 51.5 (43.9-59.1) 78.7 (75.9-81.5) Period from diagnosis (years) 10 15 100 84.7 (75.9-93.6) 76.5 (68.3-84.7) 64.9 (58.8-71.0) 58.6 (50.3-67.0) 59.5 (48.6-70.4) 36.5 (29.0-44.1) 63.4 (60.0-66.7) 88.1 (69.2-100) 76.7 (65.9-87.5) 69.4 (59.9-78.9) 52.9 (46.2-59.5) 52.7 (43.8-61.6) 44.1 (32.8-55.4) 30.8 (23.2-38.3) 54.2 (50.5-57.8) 88.1 (64.4-100) 70.1 (55.7-84.5) 61.9 (49.2-74.5) 50.6 (42.8-58.4) 47.1 (36.3-57.9) 35.1 (21.1-49.1) 25.0 (17.0-33.0) 48.9 (44.5-53.3) Note: (a) Date of censoring: 31 December 2001. 2004 VOL. 28 NO. 4 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Luke et al. Article Table 4: Actual relative survivals by percentage (95% confidence limits) for infiltrating ductal carcinomas of the female breast diagnosed in 1980-86 compared with estimated relative survivals (95% confidence limits) for corresponding cancers diagnosed in 1997-2002; South Australia.a Age at diagnosis (years) All Under 50 50-69 70+ Diagnostic period 1980-86 (n=1,135) 1997-2002 (n=3,777) 1980-86 (n=293) 1997-2002 (n=962) 1980-86 (n=580) 1997-2002 (n=1,973) 1980-86 (n=262) 1997-2002 (n=842) 78.7 (75.9-81.5) 87.6 (85.3-90.0) 79.5 (74.8-84.2) 85.8 (81.9-89.7) 79.7 (76.1-83.4) 89.3 (86.0-92.6) 77.1 (69.4-84.7) 84.8 (76.3-93.2) Period from diagnosis (years) 10 15 63.4 (60.0-66.7) 74.9 (71.4-78.3) 69.4 (63.9-74.9) 77.6 (72.8-82.5) 65.5 (61.0-70.0) 77.6 (72.4-82.9) 59.7 (49.8-69.6) 67.6 (55.4-79.8) 54.2 (50.5-57.8) 65.0 (60.5-69.6) 61.6 (55.7-67.5) 70.0 (64.2-75.7) 60.5 (55.4-65.7) 74.5 (67.9-81.0) 42.6 (30.6-54.6) 44.3 (29.9-58.7) 48.9 (44.5-53.3) 60.4 (54.6-66.2) 58.2 (51.7-64.7) 67.1 (60.3-73.9) 57.8 (51.0-64.6) 71.7 (61.7-81.8) 35.9 (13.7-58.2) 44.0 (20.1-68.0) Notes: (a) Date of censoring for actual survivals: 31 December 2001. Estimated survivals calculated by weighting actual survivals for 1980-86 to the 1997-2002 diameter distribution (see text).13,14 have been haphazard and would tend to have diminished rather than increased differences in recorded diameter by race. The present data revealed an elevated proportion of large tumours among women from the Yorke Statistical Subdivision. This is predominantly a rural area approximately 200 kilometres from Adelaide,9 which may warrant additional attention in the promotion of breast screening and other early-detection initiatives. The results of this study indicate the extent to which earlier diagnosis is occurring in South Australia, possible prognostic effects, and sections of the population at elevated risk. These findings warrant confirmation with follow-up data and data from other Australian registries.

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Aug 1, 2004

There are no references for this article.