Survival of ovarian cancer patients in Denmark: Excess mortality risk analysis of five-year relative survival in the period

Acta Obstetricia et Gynecologica. 2008; 87: 13531360 ORIGINAL ARTICLE Survival of ovarian cancer patients in Denmark: Excess mortality risk analysis of five-year relative survival in the period 1978 2002 CHARLOTTE GERD HANNIBAL 1, RIKKE CORTES 2, GERDA ENGHOLM 2 & SUSANNE KRÜGER KJAER 1,3 1 Department of Viruses, Hormones and Cancer, Institute of Cancer Epidemiology, Danish Cancer Society, Strandboulevarden 49, 2100 Copenhagen, Denmark, 2 Department of Cancer Prevention and Documentation, Danish Cancer Society, Strandboulevarden 49, 2100 Copenhagen, Denmark, 3 Gynecologic Clinic, Juliane Marie Centre, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen, Denmark Abstract Objective. To explore the variation in ovarian cancer survival in Denmark in the period 19782002 in relation to time since diagnosis, age at diagnosis, period of diagnosis, stage and histology. Design. Register-based cohort study. Setting. Denmark in the period 19782002. Population. Using the nationwide Danish Cancer Registry, we included a total of 13,035 women diagnosed with invasive ovarian cancer in Denmark in the period 19782002. Methods. Excess mortality risk analyses of fiveyear relative survival of ovarian cancer patients diagnosed in the period 19782002 with follow-up through 2006 were made based on data from the NORDCAN database. Main outcome measures. Five-year relative survival, excess mortality rate (ER) and relative excess mortality risk (RER) after an ovarian cancer diagnosis. Results. The relative survival of Danish ovarian cancer patients slightly increased in the period 19782002. The ERs were highest in the first year following diagnosis, in particular in the first three months, and among older patients, even for localized and regional tumors. The pattern remained the same when stratified by histological subgroup. Older age at diagnosis, earlier period of diagnosis, more advanced stage at diagnosis and being diagnosed with undifferentiated carcinoma predicted poorer survival among Danish ovarian cancer patients diagnosed in the period 19782002. Conclusions. The survival of Danish ovarian cancer patients has slightly increased from 1978 through 2002. Despite this, the mortality rate of ovarian cancer in Denmark is still higher than in the other Nordic countries. Explanations for these differences are still to be identified. Key words: Ovarian cancer, cancer registry, population based, excess mortality risk analysis Abbreviations: CI: confidence interval, DCO: death certificate only, ER: excess rate, FIGO: International Federation of Gynecology and Obstetrics, GLM: General Linear Models, RER: relative excess risk Introduction Worldwide, ovarian cancer is the sixth most frequent cancer in women, with 200,000 new cases diagnosed every year, and the most lethal of all being gynecological cancers (1). Given that ovarian cancer in 75% of cases is diagnosed in advanced stage, the disease has a fairly poor prognosis with an overall five-year survival of around 40%, ranging from 10% for patients with tumors with distant metastases to 80% for patients with localized tumors (24). There is marked geographical variation in agestandardized incidence and mortality rates of ovarian cancer, with the highest rates observed in Northern and Western Europe, notably Scandinavia, and in North America (5). Even within the Nordic countries, there are striking differences in the mortality rates of ovarian cancer despite similarities in the incidence rates and management of health care systems (57). In Denmark, the age-standardized incidence rate of ovarian cancer has decreased during the last 25 years, Correspondence: Susanne Krüger Kjaer, Department of Viruses, Hormones and Cancer, Institute of Cancer Epidemiology, Danish Cancer Society, Strandboulevarden 49, 2100 Copenhagen, Denmark. E-mail: susanne@cancer.dk (Received 4 July 2008; accepted 16 September 2008) ISSN 0001-6349 print/issn 1600-0412 online # 2008 Informa UK Ltd. (Informa Healthcare, Taylor & Francis AS) DOI: 10.1080/00016340802483000

1354 C.G. Hannibal et al. whereas a more substantial decrease has been observed in the mortality rate (8). Similar patterns have been seen in the other Nordic countries; nevertheless, Denmark has had the highest mortality rate in the world during the entire period (5,8). The decreasing mortality rate may be due to earlier diagnosis and improved treatment. There is no active screening for ovarian cancer yet, but increasing access to more specific diagnostic methods, such as ultrasound and tumor markers (e.g. CA125), may explain the trend towards diagnosis at earlier stages (8). In this paper, we explored the variation in ovarian cancer survival in Denmark during the period 1978 2002 in relation to time since diagnosis, age at diagnosis, period of diagnosis as well as stage and histological subgroup. Material and methods Material From the Danish Cancer Registry we identified a total of 14,660 women diagnosed with ovarian cancer (ICD-10: C56, C57.0-4) in the period 19782002. The ICD-10 codes are based on a conversion from the registration of diagnosis according to ICD-O-1 in the Danish Cancer Registry to the NORDCAN database. NORDCAN contains cancer incidence and mortality from the five Nordic countries. Different classification systems have been used over time in the Nordic countries, and in NORD- CAN conversions to groups of diagnoses based on ICD-10 have been formed for better comparability between counties over time (6). Patients with a previous cancer diagnosis (n 1,062), disregarding non-melanoma skin cancer, a simultaneous cancer at the time of the ovarian cancer diagnosis (n 318) and patients with ovarian cancer only, but detected at autopsy or identified by death certificate only (DCO) (n 245), were excluded from analysis. We included a total of 13,035 women with ovarian cancer recorded as the first invasive cancer among women aged 1589 years in the period 19782002. Follow-up for death or emigration was made through 2006, but only follow-up in the first five years after diagnosis was used and for those diagnosed in 2002 only to the end of 2006, i.e. follow-up between four and five years. Variables available to the analysis were: month and year of birth, diagnosis and death or emigration, time since diagnosis, ICD-10 code, basis of diagnosis, vital status at end of follow-up, topography, stage and histology. The extent of the disease at time of diagnosis was reported as localized tumor, tumor with regional spread, tumor with distant metastases or alternatively as stage according to the International Federation of Gynecology and Obstetrics (FIGO) classification. In this paper, we divided cases into four groups: localized tumors (stage I), tumors with regional spread (stages II and III), tumors with distant metastases (stage IV) and tumors with unknown extent of disease. Data on all-cause mortality for the Danish population originally stems from the databank of Statistics Denmark, but was here retrieved from the Human Mortality Database (http://www.mortality.org/). To secure more stable population mortality rates, five-years running means over years were used of the one-year age- and period-specific morality rates. Statistical analyses Relative survival is defined as the ratio between the observed survival within the patient group and the expected survival if the group has the mortality of the Danish population. Five-year relative survival was calculated using the SAS-macros described by Dickman et al. (9). Relative survival was age-standardized using a slightly modified version of the age distribution standard version used in EUROCARE (10). The modifications were more detailed age-groups above age 60 years and a weight of zero for ages 90 years and above at diagnosis. Modeling of the five-year relative survival was made using its counterpart in mortality, the excess mortality rate (ER), the difference between the observed and the expected all-cause mortality rates and analyzing it by Poisson regression models, here called excess mortality risk modeling. We used exact survival time, collapsed data and General Linear Models (GLM) with a Poisson error structure (9). Excess mortality risk models assume constant excess rates in intervals of follow-up. An ER can be expressed as the excess number of deaths per 100 person-years, excess to the expected number of deaths derived from population mortality rates. Modeling of the ER was explored using the explanatory factors: time since diagnosis, age at diagnosis, period of diagnosis, stage at diagnosis and histological subgroup. The first year is often the most heterogeneous year in survival analysis (9). Therefore, we subdivided the first year of follow-up using the intervals 02 months, 35 months and 612 months. The remaining follow-up time was categorized as follows: 12, 23, 34 and 45 years after diagnosis. Age groups were 1549, 5059, 6069, 7079 and 8089 years, period of diagnosis were 19781982, 19831987, 19881992, 1993 1997 and 19982002, and stages were localized, regional, distant metastases and unknown. The histological diagnosis was categorized as undifferentiated carcinomas, adenocarcinomas, serous tumors,

Survival of Danish ovarian cancer patients 1355 mucinous tumors, endometrioid tumors, non-epithelial tumors and clear cell neoplasms. Plots of the agespecific observed ER curves by follow-up intervals showed the overall pattern, the pattern by stage and by histological subgroup. The deviance with degrees of freedom for a model evaluated in a Chi-square distribution was used as goodness-of-fit of the model. Modeling of the ER in the first five years following diagnosis revealed that it was only possible to fit relatively simple models without interactions by restricting analyses to the follow-up interval 6 months to 5 years and ages 1579 years at diagnosis. In addition, it was only possible to fit a simple model with histological subgroups by including only some specific subgroups. Statistical modeling was carried out using the SAS version 9.1. Results Table I shows no major differences in the distribution of age and stage at diagnosis over time; although the percentage of distant metastasis decreased together with a slight increase in the other groups. However, the percentage of some histological subgroups decreased from 1978 through 2002 (undifferentiated carcinoma, adenocarcinoma), whereas others increased (serous, endometrioid, clear cell) or remained the same (mucinous, non-epithelial). In general, most women were diagnosed between 60 and 69 years, had tumors with regional spread and had serous tumors. The percentage of excluded patients due to incidental findings at autopsy or DCO notifications decreased over time, from around 3% at the start of the study period to B1% in the period 19982002. The age-standardized incidence rate (as a first cancer) decreased slightly over time from 18.3 per 100,000 in 19781982 to 16.9 per 100,000 in 19931997. The age-standardized mortality rate decreased somewhat more over time and was highest in the period 19781982 with 15.4 per 100,000 and lowest at the end of the study period (13.1 per 100,000). The mortality-incidence ratio is a coarse indicator of completeness and case fatality, and generally decreased over time from 80.1% in 19781982 to 74.2% in 19982002. A short length of survival (02 months after diagnosis) was strongly related to age, with an increasing percentage of older women having shorter survival (Table II). Similarly, stage at diagnosis influenced the survival time. Whereas only 2% of Table I. Characteristics of the study population of ovarian cancer patients without a previous or simultaneous other cancer, aged 1589 years at diagnosis in the period 19782002 in Denmark, and general measures for all ovarian cancer patients in the period. Period of diagnosis 19781982 (n2,601) 19831987 (n2,617) 19881992 (n2,553) 19931997 (n2,577) 19982002 (n2,687) Age at diagnosis (years) (%) 1549 16.2 16.7 20.4 17.5 16.3 5059 23.5 21.1 18.7 22.6 23.6 6069 27.6 29.6 28.2 26.6 25.3 7079 24.0 23.0 23.0 22.9 25.6 8089 8.7 9.6 9.8 10.3 9.2 Stage at diagnosis (%) Localized 17.2 19.5 20.8 20.4 18.9 Regional 42.3 45.2 40.7 45.1 47.6 Distant metastases 33.1 29.8 31.7 27.4 28.2 Unknown 7.4 5.5 6.9 7.2 5.2 Histological subgroup (%) Undifferentiated carcinoma 12.4 8.8 8.9 8.6 6.6 Adenocarcinoma 39.1 30.0 25.9 22.0 20.4 Serous 24.0 33.0 33.9 38.0 43.3 Mucinous 9.5 10.1 10.5 10.8 9.8 Endometrioid 6.6 9.7 11.2 10.4 9.4 Non-epithelial 5.9 5.2 6.0 5.8 5.9 Clear cell 2.6 3.4 3.6 4.4 4.5 Incidental finding at autopsy, and DCO$ (%) 2.7 3.3 1.6 1.0 0.8 Age-standardized incidence rate per 100,000%, 18.3 18.0 17.1 16.9 17.0 Age-standardized mortality rate per 100,000%, 15.4 14.0 14.2 13.8 13.1 Mortalityincidence ratio (%) 80.1 73.5 79.4 79.8 74.2 $Death certificate only (DCO). %All ages and age-standardized to the European standard population. Only first cancers. 19982001.

1356 C.G. Hannibal et al. women with localized disease died within the first three months, more than one-third of women with distant metastases died during the first three months following diagnosis. Finally, a slightly higher percentage of women diagnosed in the period 19781982 died within the first three months compared to women diagnosed in more recent periods. Figure 1 shows the age-standardized five-year relative cohort survival divided by stage and period in 19782002 in Denmark. In general, the relative survival of ovarian cancer patients has slightly increased in Denmark since 1978. For the entire period, the relative survival was lowest for patients diagnosed with tumors with distant metastases followed by patients with tumors with regional spread and unknown stage, whereas patients with localized tumors by far had the highest five-year relative survival. Figure 2 shows the overall pattern of age-specific observed ERs by follow-up interval for the ovarian cancer patients diagnosed during the years 1978 2002. For all age groups older than 49 years, the level of the ER was decreasing with longer time since diagnosis. In the shortest follow-up interval, 02 months after diagnosis, a large increase was seen in the ER with age, especially after 60 years. A similar, but less strong, pattern was seen for the two following periods (35 months and 612 months after diagnosis). For longer follow-up intervals, there was practically no association with age. When the age and follow-up specific ERs were stratified by stage, the patterns were fairly similar. For tumors with distant metastases and tumors with unknown stage, the largest increase in the ER was observed in the first follow-up interval, 02 months after diagnosis, especially among older women (Figure 3AD). In addition, a large increase in the ER in the first follow-up interval, 02 months after diagnosis, was observed among women 6070 years or older who had localized tumors and tumors with regional spread. Looking at the different histological subgroups, the largest increase in the ER was still seen in the first follow-up interval, 02 months after diagnosis, although the level of the rate by age differed between subgroups (data not shown). For undifferentiated carcinomas, adenocarcinomas and non-epithelial tumors, the ER in the first follow-up interval, 02 months after diagnosis, was significantly increased compared with the case for serous, mucinous and endometrioid tumors as well as clear cell neoplasms. Also, for undifferentiated carcinomas and adenocarcinomas, the large increase was seen over the whole age span, whereas for the other histological subgroups the increase was not apparent until after around 60 years. When comparing the age and follow-up specific ERs between different time periods, the patterns were fairly similar; however, the ER was, in general, slightly lower in the more recent periods, which was most evident for the shorter follow-up intervals (data not shown). Table III shows modeling of the ERs and relative excess mortality risks among patients aged 1579 years who had survived the first six months after diagnosis. The ER for the reference group (6069 Table II. Distribution of length of survival for Danish ovarian cancer patients diagnosed in the period 19782002 by age at diagnosis, period of diagnosis and stage at diagnosis. Length of survival p-value n 02 months (%) 35 months (%) 6 months (%) Age at diagnosis B0.0001 (years) 1549 2,269 3 3 94 5059 2,859 9 5 86 6069 3,576 15 7 78 7079 3,092 28 12 60 8089 1,239 47 13 40 Period of diagnosis B0.0001 19781982 2,601 22 11 67 19831987 2,617 19 8 72 19881992 2,553 18 7 75 19931997 2,577 15 7 78 19982002 2,687 15 5 80 Stage at diagnosis B0.0001 Localized 2,522 2 1 97 Regional 5,764 13 8 79 Distant metastases 3,910 34 12 54 Unknown 839 22 8 71

Survival of Danish ovarian cancer patients 1357 Figure 1. Age-standardized five-year relative survival for Danish ovarian cancer patients over calendar time according to stage. years and period of diagnosis 19831987) decreased with increasing follow-up time, from 37 excess deaths per 100 person-years 6 months to 1 year after diagnosis, to 10 excess deaths per 100 personyears 45 years following diagnosis (Table III). The relative excess mortality risk varied with age, as women aged 1549 years had the lowest risk with a statistically significant 46% lower excess mortality risk compared to women aged 6069 years after adjustment for follow-up time and period of diagnosis. In addition, compared to women diagnosed in the period 19831987, women diagnosed in the years 19982002 had a statistically significant 20% Figure 2. Age-specific observed excess mortality rates by followup interval for Danish ovarian cancer patients diagnosed in the period 19782002. lower excess mortality risk. After including stage in the model (using regional spread as reference), the ERs and relative risks changed slightly. Compared to women diagnosed with tumors with regional spread, women having tumors with distant metastases had a statistically significant 82% higher excess mortality risk, whereas women with unknown stage and localized tumors had a statistically significant lower excess mortality risk, 39% and 83%, respectively. In a model including histology with serous tumors as reference, we found that women with undifferentiated carcinoma tumors had a statistically significant 38% higher excess mortality risk and women with non-epithelial tumors had a borderline statistically significant 15% lower excess mortality risk, whereas women with mucinous tumors had the same risk, after adjustment for follow-up time, age, period and stage at diagnosis (data not shown). Discussion We found a substantial variation in ovarian cancer survival with a higher ER within the first year after diagnosis and among older patients, most evident in the first three months and after 60 years, compared to longer follow-up times and younger patients. The pattern remained fairly much the same when stratified by stage or by histological subgroup. Older age at diagnosis, more advanced stage and undifferentiated carcinomas were the most important factors in the analysis, and predicted poorer survival among Danish ovarian cancer patients diagnosed from 1978 through 2002. We found that the ER was highest during the first three months after diagnosis and among older patients. The pattern was evident even for localized and regional tumors. This might be explained by existing co-morbidity or because older women are less likely to tolerate treatment or do not want to be treated. Few population-based studies have reported survival rates of different histological subgroups over time. We found a better prognosis among women diagnosed with non-epithelial tumors compared to women diagnosed with serous tumors, but a much poorer prognosis among women diagnosed with undifferentiated carcinomas, whereas mucinous tumors had a similar risk to serous tumors. This is partly in line with a Norwegian study by Bjorge et al. (11) who also found a poorer prognosis for patients diagnosed with undifferentiated tumors compared to serous tumors, but a better prognosis among women diagnosed with mucinous tumors. The age-standardized five-year relative survival increased slightly for all stages of ovarian cancer

1358 C.G. Hannibal et al. Figure 3. (AD) Age-specific observed excess mortality rates by follow-up interval for Danish ovarian cancer patients diagnosed in the period 19782002 by stage at diagnosis. during the study period from 1978 through 2002. Women diagnosed with localized tumors had a much more favorable prognosis compared to women diagnosed with more advanced stages during the entire period. As shown in a recent EUROCARE-4 study, the five-year age-adjusted relative survival for ovarian cancer patients increased in Europe from 1990 through 1999; however, Denmark has had one of the poorest survivals compared to the other European countries in that period (12). Cancer survival is a complex indicator of a country s performance in managing cancer. Improvements in cancer survival may reflect earlier diagnosis or later death. Better survival might also be derived from inclusion of a number of patients identified at screening, where the tumor would not otherwise have progressed to give clinical signs. So far, no screening method for ovarian cancer exists, but increasing access to more specific diagnostic methods, such as ultrasound and the tumor marker CA125, may explain the trend towards earlier stages at diagnosis (8). In two ongoing studies examining screening for ovarian cancer, results have suggested that the tumor marker CA125 can lead to discover-

Survival of Danish ovarian cancer patients 1359 Table III. Models for excess mortality in the first five years among Danish ovarian cancer patients diagnosed in the period 19782002, aged 1579 years, among patients who survived the first half year following diagnosis. ER$, 95% CI p ER%, 95% CI p Follow-up time (years) 0.51 37 (3340) 39 (3542) 12 34 (3237) 40 (3744) 23 24 (2226) 31 (2834) 34 13 (1215) 19 (1722) 45 10 (811) 14 (1217) RER RER Age at diagnosis (years) B0.01 B0.01 1549 0.54 (0.500.59) 0.62 (0.570.67) 5059 0.85 (0.790.92) 0.88 (0.820.95) 6069 1 (ref.) 1 (ref.) 7079 1.18 (1.081.28) 1.26 (1.161.37) Period of diagnosis B0.01 B0.01 19781982 1.20 (0.500.59) 1.20 (1.101.32) 19831987 1 (ref.) 1 (ref.) 19881992 1.00 (0.911.10) 1.01 (0.921.10) 19931997 0.91 (0.831.00) 0.89 (0.810.98) 19982002 0.80 (0.730.87) 0.72 (0.650.79) Stage at diagnosis B0.01 Localized 0.17 (0.150.19) Regional 1 (ref.) Distant metastases 1.82 (1.711.94) Unknown 0.61 (0.530.70) ER, excess rate; RER, relative excess risk, relative to the reference category. $Excess deaths per 100 person-years for patients aged 6069 years at diagnosis in the years 19831987. %Excess deaths per 100 person-years for patients aged 6069 years with regional stage in the years 19831987. Adjusted for all other variables in the table except for stage at diagnosis. Adjusted for all other variables in the table. ing of more cases of ovarian cancer (13). Whether or not this will influence the mortality or just lead to the detection of more cases is a question. As shown in the paper by Kjaerbye-Thygesen et al. (8) as well as in the present paper, it is reassuring that the mortality rate of ovarian cancer has decreased in Denmark over the last 25 years; yet Denmark still has the highest ovarian cancer mortality rate observed worldwide (5). Kjaerbye-Thygesen et al. (8) showed that compared to the other Nordic countries, Denmark has had the highest ovarian cancer mortality rate during the entire period 19781999. This is surprising considering the similarities in ovarian cancer incidence (8) and health care systems in the Nordic countries in the same period (6). In general, it appears that Danish cancer patients overall had the same survival as patients in the other Nordic countries in the 1950 and 1960s, but did not experience the same level of improvement in prognosis during the later decades (14). In the present study, we found that 19.3% of women had localized tumors at time of diagnosis. This is a much smaller fraction compared to some of the other Nordic countries around the same time, since they found fractions of 2430% (15). The fact that more patients in Denmark seem to be diagnosed with ovarian cancer at more advanced stages might explain the higher ovarian cancer mortality observed in Denmark compared to the other Nordic countries. This is not only the case for ovarian cancer, but it appears that Danish cancer patients in general are diagnosed in more advanced stages of disease than patients in the other Nordic countries, taking into consideration that the classification of localized tumors is quite similar in the countries (14). Thus, a different distribution between countries of the diagnoses in later stages might explain the cancer survival differences within the Nordic countries, but to examine if the differences in survival are actually due to different diagnostic practices is difficult using cancer registry data only. In the present study, we included data from the Danish Cancer Registry and the NORDCAN database, which is high quality data as it is nationwide and population based. In addition, we used the unique nature of the Danish personal identification number, implying a virtually complete follow-up. In our study, we assessed excess mortality risk and relative survival, whereas most other studies have

1360 C.G. Hannibal et al. assessed only relative survival. The advantage of using excess mortality risk is that the estimates represent the specific ovarian cancer mortality after subtracting the general background mortality. In conclusion, the mortality of Danish ovarian cancer patients has slightly decreased in the period 19782002. Older age, more advanced stage and having undifferentiated carcinomas at time of diagnosis seem to be important factors in explaining the poorer survival among Danish ovarian cancer patients during the last 25 years. It is striking that Denmark still has the highest ovarian cancer mortality rate in the world and thus a much higher ovarian cancer mortality rate than in the other Nordic countries. Explanations for these differences are still to be identified. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. References 1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin. 2005;55:74108. 2. Gatta G, Lasota MB, Verdecchia A. Survival of European women with gynaecological tumours, during the period 1978 1989. Eur J Cancer. 1998;34:221825. 3. Tingulstad S, Skjeldestad FE, Halvorsen TB, Hagen B. Survival and prognostic factors in patients with ovarian cancer. Obstet Gynecol. 2003;101:88591. 4. Balvert-Locht HR, Coebergh JW, Hop WC, Brolmann HA, Crommelin M, van Wijck DJ, et al. Improved prognosis of ovarian cancer in The Netherlands during the period 1975 1985: a registry-based study. Gynecol Oncol. 1991;42:38. 5. Ferlay J, Bray F, Pisani P, Parkin DM. GLOBOCAN 2002: cancer incidence, mortality and prevalence worldwide. IARC Cancer Base No. 5, Version 2.0. Lyon: IARC Press; 2004. 6. Engholm G, Ferlay J, Christensen N, Bray F, Klint Å, Ólafsdóttir E, et al. NORDCAN: cancer incidence, mortality and prevalence in the Nordic countries, Version 3.2. Association of Nordic Cancer Registries. Danish Cancer Society (http://www.ancr.nu); 2008. 7. Bray F, Sankila R, Ferlay J, Parkin DM. Estimates of cancer incidence and mortality in Europe in 1995. Eur J Cancer. 2002;38:99166. 8. Kjaerbye-Thygesen A, Huusom LD, Frederiksen K, Kjaer SK. Trends in the incidence and mortality of ovarian cancer in Denmark 19782002. Comparison with other Nordic countries. Acta Obstet Gynecol Scand. 2005;84:100612. 9. Dickman PW, Sloggett A, Hills M, Hakulinen T. Regression models for relative survival. Stat Med. 2004;23:5164. 10. Corazziari I, Quinn M, Capocaccia R. Standard cancer patient population for age standardising survival ratios. Eur J Cancer. 2004;40:230716. 11. Bjorge T, Engeland A, Hansen S, Trope CG. Prognosis of patients with ovarian cancer and borderline tumours diagnosed in Norway between 1954 and 1993. Int J Cancer. 1998;75:66370. 12. Berrino F, De AR, Sant M, Rosso S, Bielska-Lasota M, Coebergh JW, et al. Survival for eight major cancers and all cancers combined for European adults diagnosed in 199599: results of the EUROCARE-4 study. Lancet Oncol. 2007;8:77383. 13. Rosenthal AN, Menon U, Jacobs IJ. Screening for ovarian cancer. Clin Obstet Gynecol. 2006;49:43347. 14. Engeland A, Haldorsen T, Dickman PW, Hakulinen T, Moller TR, Storm HH, et al. Relative survival of cancer patients a comparison between Denmark and the other Nordic countries. Acta Oncol. 1998;37:4959. 15. Bjorge T, Engeland A, Hansen S, Trope CG. Trends in the incidence of ovarian cancer and borderline tumours in Norway, 19541993. Int J Cancer. 1997;71:7806.