Until now, direct comparisons of cancer survival between high-income and low-income countries have not generally been available. The information provided here might therefore be a useful stimulus for change. The findings should eventually facilitate joint assessment of international trends in incidence, survival, and mortality as indicators of cancer control.

Four approaches to estimating a regression model for relative survival using the method of maximum likelihood are described and compared. The underlying model is an additive hazards model where the total hazard is written as the sum of the known baseline hazard and the excess hazard associated with a diagnosis of cancer. The excess hazards are assumed to be constant within pre-specified bands of follow-up. The likelihood can be maximized directly or in the framework of generalized linear models. Minor differences exist due to, for example, the way the data are presented (individual, aggregated or grouped), and in some assumptions (e.g. distributional assumptions). The four approaches are applied to two real data sets and produce very similar estimates even when the assumption of proportional excess hazards is violated. The choice of approach to use in practice can, therefore, be guided by ease of use and availability of software. We recommend using a generalized linear model with a Poisson error structure based on collapsed data using exact survival times. The model can be estimated in any software package that estimates GLMs with user-defined link functions (including SAS, Stata, S-plus, and R) and utilizes the theory of generalized linear models for assessing goodness-of-fit and studying regression diagnostics.

Prediction of the future number of cancer cases is of great interest to society. The classical approach is to use the age-period-cohort model for making cancer incidence predictions. We made an empirical comparison of different versions of this model, using data from cancer registries in the Nordic countries for the period 1958-1997. We have applied 15 different methods to 20 sites for each sex in Denmark, Finland, Norway and Sweden. Median absolute value of the relative difference between observed and predicted numbers of cases for these 160 combinations of site, sex and country was calculated. The medians varied between 10.4 per cent and 15.3 per cent in predictions 10 years ahead, and between 15.1 per cent and 32.0 per cent for 20 year predictions. We have four main conclusions: (i) projecting current trends worked better than assuming that future rates are equal to present rates; (ii) the method based on the multiplicative APC model often overestimated the number of cancer cases due to its exponential growth over time, but using a power function to level off this growth improved the predictions; (iii) projecting only half of the trend after the first 10 years also gave better long-term predictions; (iv) methods that emphasize trends in the last decade seem to perform better than those that include earlier time trends.

Key words: prostate cancer; 25(OH)-vitamin D 3 ; serum bankVitamin D deficiency has been implicated as risk factor for prostate cancer. 1 In cell culture studies, vitamin D metabolites have had protective action against cancer development (for review see Ylikomi et al. 2 ). Normal and malignant prostate cells contain vitamin D receptor (VDR), 3-5 which mediates the antiproliferative action of 1,25(OH) 2 -vitamin D 3 . 6 In addition to the antiproliferative action of 1,25(OH) 2 -vitamin D 3 , it can cause apoptosis, 7 induce differentiation, 8 inhibit telomerase expression, 9 inhibit tumor cell invasiveness 10 and suppress tumor-induced angiogenesis. 11 Several epidemiologic studies have reported that high serum vitamin D levels or sunlight may protect against prostate cancer. 3,4,[12][13][14][15] Factors that affect prostate cancer include age, dark skin and environment, e.g., latitude and diet. 16 These factors might be linked to vitamin D availability. 17,18 Furthermore, high fish (rich in vitamin D) consumption appears to correlate with lower prostate cancer risk. 19 In addition, VDR gene polymorphism may contribute to the risk of prostate cancer. 20 -24 There is also a study showing no correlation between serum vitamin D metabolites and prostate cancer in Maryland (USA), 25 but the authors concluded that the power of their study was limited. In another study on U.S. male physicians, only a weak protection against prostate cancer was found with the highest quartile of serum 1␣,25(OH) 2 -vitamin D 3 . 26 Similarly, no correlation was found in Hawaii. 27 There are 2 physiologically interesting metabolites of vitamin D, 1␣,25(OH) 2 -vitamin D 3 , regulating calcium homeostasis for bones and muscles in extremely narrow limits, and 25(OH)-vitamin D 3 , regulating target (prostate) cell proliferation and differentiation through activation to 1␣,25(OH) 2 -vitamin D 3 in the target (prostate) cell. Serum 25(OH)-vitamin D 3 is produced by liver 25-hydroxylase, the rate of the synthesis being directly proportional to vitamin D 3 serum concentration. 28 Therefore, serum 25(OH)-vitamin D 3 reflects vitamin D availability in the organism. Serum concentration of 25(OH)-vitamin D 3 is so high that it might possess a significant biologic activity in target cells, but it is also a precursor for the biologically more active 1␣,25(OH) 2 -vitamin D 3 . Prostate as well as many other target organs can activate 25(OH)-vitamin D 3 through 1␣-hydroxylation 29,30 and inactivate it through 24-hydroxylation. 31 In an epidemiologic study, we found that low concentrations (Ͻ40 nmol/l) of 25(OH)-vitamin D in serum were associated with a 1.7-fold increased risk of prostate cancer. 3,4 Since the power of our study was limited, preventing extensive analysis of the data, and we are partners in the Nordic Specimen Banks for Cancer Causes and Control, we had an opportunity to extend our study to other Scandinavian countries located geographically at the same latitude. Our aim was to determine whether our finding could be replicated in a la...

Survival from cancer over a certain time period is often measured by the 'relative survival rate'. This is the ratio of the observed survival rate in the group of patients to the survival rate expected in a group of people in the general population, who are similar to the patients with respect to all of the possible factors affecting survival at the beginning of the period, except for the disease under study. When patterns of patient withdrawal differ for a number of subgroups of patients with equal relative survival rates, the current method of derivation of the relative survival rate is biased. A method based on the concept of an 'expected life table' is proposed for removal of the bias. Examples based on material from the Finnish Cancer Registry suggest that the practical performance of the proposed method is better than that of other alternatives, even when the relative survival rates in the subgroups are not equal.

The increasing trend in testicular cancer risk observed for these six populations follows a birth cohort pattern. This distinct risk pattern provides a framework for the identification of specific etiologic factors.

We used multiple regression models to assess the influence of disease stage at diagnosis on the 5-year relative survival of 4,478 patients diagnosed with breast cancer in 1990 -1992. The cases were representative samples from 17 population-based cancer registries in 6 European countries (Estonia, France, Italy, Netherlands, Spain and UK) that were combined into 9 regional groups based on similar survival. Five-year relative survival was 79% overall, varying from 98% for early, node-negative (T1N0M0) tumours; 87% for large, node-negative (T2-3N0M0) tumours; 76% for node-positive (T1-3N؉M0) tumours and 55% for locally advanced (T4NxM0) tumours to 18% for metastatic (M1) tumours and 69% for tumours of unspecified stage. There was considerable variation across Europe in relative survival within each disease stage, but this was least marked for early nodenegative tumours. Overall 5-year relative survival was highest in the French group of Bas-Rhin, Cô te d'Or, Hé rault and Isè re (86%), and lowest in Estonia (66%). These geographic groups were characterised by the highest and lowest percentages of women with early stage disease (T1N0M0: 39% and 9%, respectively). The French, Dutch and Italian groups had the highest percentage of operated cases. The number of axillary nodes examined, a factor influencing nodal status, was highest in Italy and Spain. After adjusting for TNM stage and the number of nodes examined, survival differences were greatly reduced, indicating that for these women, diagnosed with breast cancer in Europe during 1990 -1992, the survival differences were mainly due to differences in stage at diagnosis. However, in 3 regional groups, the relative risks of death remained high even after these adjustments, suggesting less than optimal treatment. Screening for breast cancer did not seem to affect the survival patterns once stage had been taken into account.

No abstract