Epidemiology o f pancreatic carcinoma in Canada: 1931 to 1985

: Pancreatic cmc1noma accounts for approximately one of every 20 deaths from cancer and one out of every 30 new cases of cancer in Canada. Ir is among the 10 leading causes of death from cancer and the JO leading sires of newly diagnosed cases for a number of .ige groups. Age standardized mortality rares have risen sharply in both males (2.0 to 9.5 deaths per 100,000 population per year) and females 12.2 to 5. 7 deaths per 100.000 population per year) during 1931 ro 1985 (P < 0.0001 ). The increase in srnndarized rates, 0.14 and 0.07 additional deaths per 100,000 population per year in males and females, respectively, is arrributable to significant increases in age specific rates for males and females aged 35 to 44. 45 to 54. 55 co 64, 65 to 74. 7 5 co 84 and over 85 years old ( P ~ 0.0005) in whom rares have risen by as much as 2. 5 additional deaths per 100,000 population per year. Although age stand ardized incidence rnte5 have risen marginally in males (P = 0.085), age specific rares in mab aged Oto 24 years have risen significantly (P = 0.01 ). In contrast, standarized incidence races have risen sharply in females (0.12 additional new cases per 100,000 population per year; P - 0.0007), which was also characteristic of age specific rates for women aged 45 to 54, 55 ro 64. 65 to 74 and 75 to 84 years old (P < 0.03) where rates have risen by 0.15 to 1.65 additional new cases per 100,000 population per year. Recent age specific incidence and mortality rates indicate rhar morbidity and mortal ity rates rise sharply after age 45, when rates double between successive 10-year age groups. and peak for males and females aged more than 85 years. Can J Gastro cnterol [988;2(2):71-74

Mortality from pancreatic carcinoma is ranked as high as the fourth leading ca use of death from cancer for a numbe r of age groups in ma les and females. It is among the 10 leading causes of death from cancer in mnles aged 35 to 39 and A) 11)~1A~11 fl~ mab and females aged 40 ro 44. 45 ro 49. 50 to 54, 55 co 59, 60 to 64. 65 to 69. 70 to 74, 75 co 79, 80 to 84 and 85 years anJ older ,ind accounts for as many as 7.2'\, of all Jearhs due to ca ncer ( 3). T he lifetime probability of dying from ch is neoplastic disease in Canad,1 is 1.27'\, for male~ and 1.23"(, in females, (unpublished data).
Projections hy the Canadi,ln Cancer Sociery indicate that approximately 2600 Canadians will die from pancreatic carcinoma in 1987 and ch is neoplastic disease will be diagnosed in another 2600 during chat yc::ir.
Over 3750 hospital admissions during 1982 were for che diagnosis of rhis neopl.1sm ( 17.2 admissions in males and 13.4 admissions in females per 100,000 population per year). Admiss ion races were least in males and females under the age of 45. helnw 5 :1dmi ssions per 100,000 population per year. However, race~ rose quickly with increasing age and were greatest in males and females aged 7 5 to 84 anJ 85 years and older where they cxcecdcJ 150 :md 7 5 ad mission~ per L00,000 population per year, respectively. The mean IV"~pital stay wa~ 20.8 and 25. I day~ per admission for males and fe-male~. rcspl'Ctivcly ( > ). Statistics  The epidemiology of pancreatic carcinoma has heen described for Australia (4). England and Wales (5 ) and Japan (6). However, similar derailed analyses based on Canadian data have heen lacking and the purpose of this study was to add ress th is deficiency.

MATERIALS AND METHODS
A database consisting of the annual nu mber of repor ted new primary cases and deaths from pancreatic carcim1ma ns well as esti mates of mid-year popu lations was assembled using the annual publications of Statistics Canada ( 1-2) and the rubrics of the International C las-72 sificarion of Diseases ( ninth revision). Data were collected by age grou p (0 to 24, 25 to 34, 35 to 44, 45 to 54, 55 to 64, 65 m 74, 7 5 to 84 and 85 years and older) and sex for the periods 1970-82 (morbidity figures) and 1931-85 (mortality figures).
Simple b ivariate plots of age specific and age standardized morb idity (or incidence) and mortality rates as a function of calendar year indicated that linear regression analysis was suitable for the assessment of secular trends as both age specific and age standardized races did not demonstrate any curvilinear relationships. Age specific and age standardized incidence and mortality rates were used as the dependent variable in the regression analysis whereas the year from 1970 ( morbidity rates) and from 1931 ( mortality rates) was used as the independent variable, thus allowing for the estimation of the significance of annual rates of change in these rares. All parametric analyses were confirmed using Spearman's (nonparametric) rank-order correlation coefficient. Statistics with associated levels of significance of 0.05 or less were deemed significan t.
Other details regarding data manage-

RESULTS
A ge s tan d a rized morbidity ra tes (ASMRs): Standarizcd morbidity rates for males during l970-82 h ave consistently exceeded corresponding rates in females by 49% to 88% (Table I)   The stabillly of ASMRs noted for males is characteristic of the seven age specific rates studied which d id not increase or decrease significantly during the period ~tu died ( P ?: 0 10). However, rates in males increased significantly during this period ( P 0.01>) ( Table 2 ). In contrast, the rise in female A M Rs is associated with corresponding significan t increases in age specific rates for females aged 45 Lo 54. 5 5 to 64. 65  Age s tand a ri zed m or tality rates (ASDRs) : The increase in A DRs was dramatic in both males and fema les (Table I) where rates increased from approximately 2 to as many as 9 4 deaths per 100.000 populatio n per year in males dunng 1931 to 1985 and from about 2 to slightly under 6 deaths per 100.000 population per year in females during the same period . The sign ificant rise in ASDRs for both males. 0. 14 additional deaths per 100.000 population per year, and females, 0.07 addiuonal deaths per 100,000 population per year, is highl y significant (P < 0.0001) ( Table 3).
Exam ination of age-sex-specific mortality rates inc.licc1tcs that the apprec iable rise m male ASDRs 1s attributable to significant increase:, in mortality in males aged 35 to 44. 45 to 54, 55 to 64. 65 to i4. 75 to84 and 85 years and older where rates increased by as much as 2.5 addinonal deaths per 100,000 populatio n per year (P < 0.0005 ). Similar resu lts were Vol 2 No. 2.June 1988 noted 111 females as rates for the same six age groups rose significantly (0.01 to 1.8 addnional deaths per 100,000 p0pulatio n per year; P < 0.0001) As was the case with morbidity rates, standardized mo rtality rates were consistently higher in males and they generally exceeded rates in females by 50'\, to 90"[,.
The rapid inc rease 111 morbidity and

DISCUSSION
Using the model proposed by Zdeb (9), the probability that a Canadian will die from carcinoma of the pancreas is 1.27% in males and I 2 3% in fomales.
Although rhis probability is low, particularly when compared to probabilities for dying from other neoplastic diseases, pancreatic carcinoma is co n sisten cly ranked among the 10 leading causes and sites of cancer in Canada for many age groups for both males and females, and accounts for at least 6°{, of all deaths from cancer for many age-sex groups ( 3 ).
The sim ilarity between morbidity and mortality rates in Table 4 (most recent years for which incidence and mortality rares are available) is reflective of the low survival rare among patients afflicted with this neoplastic disease. Five-year surviv,i l rates based on a large series of patients from England and Wales indicates that survival is most likely in males and females age 15 to 44. with rates of 7 a nd 6'\,, respectively. Survival rates arc 4% or less for all other age groups ( 5 ).
The marked increase in m ortality noted for Canada in this study is characteristic of the pattern of th.is disease in other parts of the world. In England and Wales, mortality rates have risen from 2.9 to 10.6 deaths per 100,000 population per year in males during 1911-15 to 1976-78and from 2.9 to 8.6 deaths per 100,000 population per year in females.
Inspection of age-sex-specific m ortality rates indicates that dramatic increases in mortality rates have occurred for older members of the population. Rares have risen from 12.6 and 12. 1 deaths per L 00.000 population in males and females aged 75 years a nd older ( 1911-15) co 94. l and 66.9 deaths per 100,000 population ( 1976-78). Equally impres~ive increases arc noted for males and females aged 45 to 54. 5'i co 64 and 65 to 74 where rates have risen by factors of approximately 1 or more and 2 or more, respectively.
Further analysis of British mortality rates by metropolitan coun ties and rural and intermediate regions failed to produce any major difference in mortality between rural a nd urban dwellers in eithe r males or females. S imilarly, analysis by social class fai led to produce any dramatic differences between the five major classes for males and (married) females.
The epidemiology of pancreatic carcinoma in Australia has followed a similar course. Holman and Armstrong (4) found mortaliry rates in males to h ave increased more than fourfold ( l.84 to 7.84 deaths per 100.000 population per year) and in females by threefold ( 1.57 to 4. 58 deaths per 100,000 population per year).
lncidence rates around the world vary within and between nations ( 10). In Canada, rates in males vary from 3. 7  Holman and Armstrong (4) note that the dramatic rise in pancreatic carcinoma is perhaps a reflection of both crosssectional and cohort-based effects and may involve an improvement in the abi lity co diagnose the disease. However. improvements in the reporting and/or diagnosis of this neoplastic disease is not totally supported by Canadian rates ('T:'lble 1) since 1970 as, for example. standardized incidence rates have been relatively stable during the period 1970-82 for males w hereas rates in females h ave continued to rise. Furthermore, the relative consistency in mortality races be-