M. Arbyn, A.O. Raifu, J. Antoine

TRENDS OF CERVICAL CANCER

MORTALITY IN EUROPE

Section Epidemiology Juliette Wytsmanstreet, 14 1050 Brussels | Belgium www.iph.fgov.be

(VERSION 1)

Epidemiology | 07 2009 | Brussels, Belgium No deposit: D/2009/2505/41

TRENDS OF CERVICAL CANCER MORTALITY IN EUROPE M. Arbyn1 A.O. Raifu1 J. Antoine1 1 Unit of Cancer Epidemiology / Belgian Cancer Centre Department of Epidemiology Scientific Institute of Public Health, Brussels, Belgium The project is financially supported by

1. DG SANCO of the European Commission, through the EUNICE and EUROCHIP-3 Networks

2. DG Research of the European Commission, through the CCPRB and EUROCOURSE Networks

3. Belgian National Cancer Plan

Science at the service of Public health, Food chain safety and Environment.

Acknowledgments Funding was received from: (1) the DG SANCO of the European Commission (Grand-Duchy of Luxembourg), trough EUNICE (European Network for Information in Cancer Epidemiology), via IARC (Lyon, France) and, through the EUROCHIP-3 Network (Istituto Nazionale dei Tumori, Milan, Italy) and through the European Network for Information on Cancer (IARC, Lyon, France); (2) 6th and 7th Framework Programme of DG Research of the European Commission through the CCPRB Network of Excellence (Cancer Control using population-based Cancer Registries and Biobanking (University of Lund, Sweden) and the EUROCOURSE Network (Optimisation of the Use of Registries for Scientific Excellence in research (Comprehensive Cancer Centre South, Eindhoven, the Netherlands); (3) Belgian National Cancer Plan, through the National Health Insurance Institute (Brussels, Belgium).

© Scientific Institute of Public Health, Brussels 2008 This report may not be reproduced, published or distributed without the consent of the ISP | WIV.

1. TABLE OF CONTENTS 1. TABLE OF CONTENTS..............................................................................................................6

2. EXECUTIVE SUMMARY...........................................................................................................7 2.1. BACKGROUND..........................................................................................................................7 2.2. MATERIAL AND METHODS ........................................................................................................7 2.3. RESULTS...................................................................................................................................7 2.4. DISCUSSION..............................................................................................................................7

3. INTRODUCTION.........................................................................................................................9

4. MATERIAL AND METHODS..................................................................................................11 4.1. SOURCE OF DATA ...................................................................................................................11

4.1.1. Mortality data ...................................................................................................................11 4.1.2. Population data ................................................................................................................11 4.1.3. Geographical identification of states................................................................................11 4.1.4. Missing data .....................................................................................................................12 4.1.5. Codification of deaths from uterine cancer ......................................................................13 4.1.6. Age groups........................................................................................................................20

4.2. REALLOCATION OF DEATHS CAUSED BY CANCERS OF THE UTERUS.........................................20 4.2.1. Cancer of the uterus .........................................................................................................20 4.2.2. Reallocation of causes of death ........................................................................................20 4.2.3. Imputation.........................................................................................................................25

4.3. TREND ANALYSES ..................................................................................................................26 4.3.1. Mortality from uterus cancer among women younger than 45 years...............................26 4.3.2. Corrected number of deaths from cervix uteri cancer, mortality rates ............................26 4.3.3. Join Point regression........................................................................................................27 4.3.4. Standardised cohort mortality ratio .................................................................................27

4.4. STATISTICAL SOFTWARE ........................................................................................................28 5. RESULTS ....................................................................................................................................29

5.1. MORTALITY FROM UTERUS CANCER AMONG WOMEN BETWEEN 20-44 YEARS ........................30 5.2. STANDARDISED COHORT MORTALITY RATIO FOR UTERUS CANCER AMONG WOMEN AGED

20-44 YEARS..........................................................................................................................32 5.3. CRUDE RATES OF MORTALITY FROM CERVIX CANCER (CERTIFIED AND CORRECTED FOR

NOS) AND UTERUS CANCER ..................................................................................................34 5.4. AGE-STANDARDISED RATE OF MORTALITY FROM CERVIX CANCER (CERTIFIED AND

CORRECTED FOR NOS) AND UTERUS CANCER........................................................................36 5.5. CORRECTED AGE-SPECIFIC MORTALITY FROM CERVICAL CANCER BY PERIOD ........................38 5.6. CORRECTED AGE-SPECIFIC RATES OF MORTALITY FROM CERVICAL CANCER BY BIRTH

COHORT .................................................................................................................................40 5.7. STANDARDISED COHORT MORTALITY RATIO (CERVICAL CANCER, CORRECTED, ALL AGES) ....42 5.8. JOINPOINT REGRESSION OF STANDARDISED CORRECTED CERVICAL CANCER MORTALITY.......44

6. DISCUSSION ..............................................................................................................................45

7. ABBREVIATIONS .....................................................................................................................48

8. FILES USED ...............................................................................................................................49

9. REFERENCES............................................................................................................................50

7

2. EXECUTIVE SUMMARY

2.1. Background Cervical cancer mortality can be avoided to a large extent by screening and treatment of screen-detected cervical lesions. However, in 2004, approximately 27,000, in the European continent, and 16,000 women, in the the European Union, died from this cancer. Straightforeward analysis of trends in the cervical cancer mortality is hampered by inaccuracies in death cause certification. Adjustments are needed to correct for deaths from uterus cancer not otherwise specified. In the current report, we analysed cervical cancer mortality trends in all countries of the the European Continent but we focussed mainly on the 27 current Member States of the European Union. 2.2. Material and methods Data on number of deaths from uterine cancers and overall female population from all European countries were extracted from the WHO mortality data base. Three different reallocation rules were applied to correct cervical cancer mortality for inaccuracies in certification of death cause of not otherwise specified uterine cancer. Since the large majority of uterine cancers before the age of 45 years is from cervical origin, we analysed mortality from cancer of the uterus in the age group 20-44 years. This method does not require any reallocation. Joinpoint regression was used to study annual variation of corrected cervical cancer mortality in all European countries. 2.3. Results Corrected age-standardised cervical cancer mortality rates decreased significantly over the last decades in the 15 old Member States of the European Union. Member States in Eastern Europe and also the Baltic states showed mortality rates that decreased at lower intensity (Czech Republic, Poland), remained constant at a high rate (Estonia, Slovakia) or even increased (Bulgaria, Latvia, Lithuania, Romania). The standardised cohort mortality ratio indicated that mortality did not decrease further among women born after 1940. 2.4. Discussion Remarkable contrasts were observed in cervical cancer mortality, in particular, between the old and new EU Member States, which might probably be explained by differences in coverage and quality of cytology-based screening in the past. Cohorts

8

born after 1940 are a higher risk of cervical cancer because of increased exposure to the sexually transmitted high-risk human papillomavirus types, the main etiological risk factor. There is an elevated burden of cervical cancer in Central and Eastern Europe. Moreover, mortality rates tend to rise or remain stable in the most affected countries such as Romania, Bulgaria and the three Baltic states, whereas in most other European countries trends are decreasing. The east-west contrast might increase in the future, unless adequate preventive measures are adopted. Public health authorities should set up well-organised cervical cancer prevention programmes without delay as recommended by the European Council according to the European Guidelines for Quality Assurance in Cervical Cancer Screening. Carefully organised HPV vaccination might further corroborate the impact of screening.

9

3. INTRODUCTION

According to recent estimates for the year 2004, approximately 34,300 women in the European Union developed cervical cancer and about 16,300 died from the disease (4). For whole European continent, estimates (which are less precise) are of the order of 52,000 cases and 27,000 deaths (5). The main etiologic factor for cervical cancer is persistent infection with sexually transmittable high-risk human papillomaviruses (6). By well organised screening and treatment of screen-detected high-grade cervical intraepithelial neoplasia (CIN) invasive cancer can be avoided (7). Therefore, trends in incidence of cervical cancer largely reflect coverage and quality of screening, as well as changes in exposure to risk factors which are mainly related to sexual habits of successive cohorts (8,9). Mortality trends are determined by the incidence and by the case fatality rate. Survival (the complement of case fatality) is influenced by stage and age of diagnosis, access to specialised care and effectiveness of available cancer management procedures (5,8-11). Screening plays a role in detecting invasive cancer at an early curable stage (12). The study of incidence trends would be more pertinent to assess the impact of cervical cancer screening. However, incidence data reported by cancer registries are less comprehensive than mortality statistics, which are compiled from nearly all European countries since several decades by the World Health Organisation. Furthermore, cancer incidence statistics from early periods in certain registries are inflated by inclusion of pre-invasive lesions, and cancer registries often do not separate micro-invasive (easily curable) from fully invasive cancer cases (resulting in substantial mortality) (13). Trend analyses of cervical cancer mortality often are hampered by inaccuracies in certification of cause of death, since, in many countries, a substantial fraction of uterine cancer deaths are coded as cancer from the uterus not otherwise specified (NOS) where it is not determined whether the cancer originated form the cervix or the corpus uteri (14,15). Moreover, in the 8th International Codification of Diseases, cancer of the corpus uteri or of the uterus NOS were grouped in one 3-digit code. In a previous special issue dedicated to cervical cancer screening in Europe, Levi et al, analysed the trend of mortality from cervical cancer in Europe (16). No attempt was made to correct for inaccuracies in the certification of death by uterine cancers. As a proxy for cervical cancer mortality, cancer of all uterus cancers combined was studied among women younger than 45 years, since in this age group the large majority of uterine cancers originate from the cervix (17). However, this age group may not enable assessment of the full population impact of screening, as the majority of deaths from cervix cancer occur after the age of 45 years. In the current study an algorithm was developed to reallocate deaths from the uterus NOS or combined groups, building further on previously published methods (14,15). Finally, the trends of the corrected rates are tentatively explained as a result of secondary prevention taking into account changes in exposure to risk factors and impact of oncologic treatment on survival. This report has been prepared in the framework of the European Network for Information on Cancer Epidemiology (EUNICE), which was coordinated by the

10

International Agency for Research on Cancer (Lyon, France) and which was supported by the European Commission via DG SANCO (Luxembourg, Grand Duchy of Luxembourg). Therefore, the report focusses essentially on the 27 current Member States of the EU, although avaialble data from all countries from the European continent were processed; The trend analyses will be continued throughout other networks (EUROCHIP-3 [www.tumori.net/eurochip]), national and international collaborations (for instance cancer registries of Baltic States, Bulgaria, Croatia, and Romania; Belgian Cancer Plan) and is therefore necessarily incomplete. These collaborations should result in separate publications.

11

4. MATERIAL AND METHODS

4.1. Source of data 4.1.1. Mortality data Mortality data were downloaded from the WHO (World Health Organisation) Mortality Database available on the web at http://www.who.int/whosis/mort. This database contains aggregated data on the number of deaths arranged by country, sex, age group and cause of death. The WHO Mortality Database contains separate zipped files corresponding to the ICD (International Classification of Diseases) versions: icd7.zip, icd8.zip, icd9.zip and icd10.zip. The last downloading was done on 23 October 2007. The earliest data were from 1950, the most recent from 2006. The availability of mortality data per country and period is detailed in Table 1. We selected data from countries belonging to the European continent (with country codes between 4000 and 4400, 3080 for Cyprus and 3400 for Turkey) concerning mortality from cancer of the different parts of the uterus: cervix uteri cancer, corpus uteri cancer, cancer of the uterus not otherwise specified (NOS) and other uterine cancers. For Romania, the number of deaths for the period 1959-68 was aggregated over 10-year age groups, whereas since 1969, 5-year age groups were used. We used the two earliest years (1969-70) for which 5-year age details were available and computed the average proportion within each pair of subsequent age groups and applied this proportion on the corresponding 10-year age groups for the period 1959-68. 4.1.2. Population data Population data files were downloaded alongside with the mortality data from the WHO website and data from females in the European states were selected and merged with the mortality files. The availability of population data per state and period is detailed in Table 1. 4.1.3. Geographical identification of states Several historical changes occurred in the geographical delineation of countries and states. Czechoslovakia was split in the Czech Republic and Slovakia in 1993. Several states were split from the Yugoslavian Republic at different time points (Slovenia and Croatia in 1991, the Former Yugoslavian Republic of Macedonia in 1993 and Bosnia & Herzegovina in 1995) leaving Serbia & Montenegroa. From 1973 to 1989, data from East and West Germany were combined and from 1990 onwards data for the whole of Germany were available. For the United Kingdom (UK), data are available for different subparts (England & Wales, Northern-Ireland and Scotland) and for the whole of the UK as well.

a In 2006, Montenegro, was split from Serbia & Montenegro, but no data is as yet available for this new state.

12

Henceforth, we will use the term areas to describe countries or states. 4.1.4. Missing data Small gaps in the data Seven areas including Armenia, Belarus, Estonia, Georgia, Lithuania, Moldova, and Ukraine lacked the population data for the year 1984. The population data of the year 1970 for Spain and of 2000 for the UK was also missing. Recent population data was lacking for Moldova (2006) and Slovenia (2005) but were available for mortality (see Table 1). We estimated the missing population in these areas through interpolation. The interpolation method is the process of estimating the missing observation within a data set. Data from one missing year x were estimated from the average computed over the years x-2, x-1, x+1 and x+2. When data is missing in two years (x1 and x2) consecutively, the missing data for year x1 was estimated by computing the average over the years x1-2, x1-1, x2+1 and x2+2. Data for the second missing year were estimated from the average computed over x1-1, x1, x2+1 and x2+2. Excluded areas because of missing data No data were available for population and mortality for the areas: Andorra and Liechtenstein. For Cyprus (1964-74), Monaco (1986-87), San Marino (1995-2000), only population data are present for a restricted period. For Turkey, no population figures are available and mortality data are present for only three years (1983, 1984 and 1987). All these six areas were excluded from the current trend analysis. Table 1. Availability of data for population and deaths: period (range of years) and lacking years for each area of the European continent.

Area Population Deaths Range of available years Missing Range of available years Missing Albania 1987-2004 - 1987-1989, 1992-2004 1990-1991 Andora - - - - Armenia 1981-1983, 1985-2003 1984 1981-1982, 1985-2003 1983-1984 Austria 1955-2004 - 1955-2004 - Belarus 1981-1983, 1985-2003 1984 1981-1982, 1985-2003 1983-1984 Belgium 1954-1997 - 1954-1997 - Bosnia & Herzegovina 1985-1991 - 1985-1991 - Bulgaria 1964-2004 - 1964-2004 - Croatia 1985-2005 - 1985-2005 - Cyprus 1964-1974 - - - Czechoslovakia 1953-1991 1953-1991 - Czech Republic 1986-2005 - 1986-2005 - Denmark 1951-2001 - 1951-2001 - Estonia 1981-1983, 1985-2004 1984 1981-1982, 1985-2004 1983-1984 Finland 1952-2005 - 1952-2005 - France 1950-2004 - 1950-2004 - Georgia 1981-1983, 1985-2001 1984 1981-1982, 1985-2001 1983-1984 Germany 1990-2004 - 1990-2004 -

East Germany 1973-1990 - 1973-1990 - West Berlin 1955-1970 - 1955-1960, 1962-1970 1961

West Germany 1952-1990 - 1952-1990 -

13

Area Population Deaths Range of available years Missing Range of available years Missing Greece 1961-2004 - 1961-2004 - Hungary 1955-2005 - 1955-2005 - Iceland 1952-2005 - 1952-2005 - Ireland 1950-2005 - 1950-2005 - Italy 1951-2002 - 1951-2002 - Latvia 1981-2004 - 1981-2004 - Liechtenstein - - - - Lithuania 1981-1983, 1985-2004 1984 1981-1982, 1985-2004 1983-1984 Luxembourg 1967-2004 - 1967-2004 - Macedonia 1991-2003 - 1991-2003 - Malta 1965-2004 - 1965, 1967-2004 1966 Moldova 1981-1983, 1985-2005 1984, 2006 1981-1982, 1985-2006 1983-1984 Monaco 1986-1987 - - - Netherlands 1950-2004 - 1950-2004 - Norway 1951-2004 - 1951-2004 - Poland 1959-2005 - 1959-1996, 1999-2005 1997-1998 Portugal 1955-2003 - 1955-2003 - Romania 1959-2004 - 1959-1978, 1980-2004 1979 Russian Federation 1980-2005 - 1980-2004 - San Marino 1995-2000 - - - Serbia & Montenegro 1997-2002 - 1997-2002 - Slovakia 1992-2005 - 1992-2005 - Slovenia 1985-2004 2005 1985-2005 - Spain 1951-1969, 1971-2004 1970 1951-2004 - Sweden 1952-2004 - 1952-2004 - Switzerland 1951-2004 - 1951-2004 - Turkey - 1983-1987 1983-1984, 1987 1985-1986 Ukraine 1981-1983, 1985-2005 1984 1981-1982, 1985-2005 1983-1984 United Kingdom 1950-1999, 2001-2004 2000 1950-1999, 2001-2004 2000

England & Wales 1950-2004 - 1950-2004 - Northern Ireland 1950-2004 - 1950-2004 -

Scotland 1950-2004 - 1950-2004 - Yugoslavia 1960-1990 - 1960-1990 -

4.1.5. Codification of deaths from uterine cancer The codes used to identify cancer of the different parts of the uterus in the subsequent editions of the International Classification of Diseases, Injuries, and Causes of Death (ICD) are shown in Table 2. Table 2. ICD codes used to identify cancer of the different parts of the uterus

Cancer of the ICD edition Cervix Corpus Uterus NOS Uterus other parts

Abbreviation CVX CRP NOS OTH 7 171 172 174 173 8 180 182.0 182.9 181 9 180 182 179 181 10 C53 C54 C55 C57/C58* *C57: malignant neoplasm of other and unspecified female genital organs C58: malignant neoplasm of placenta

14

In all ICD editions, separate codes were foreseen to identify cervical cancer (171 in the 7th, 180 in the 8th and 9th, and C53 in the 10th edition). Corpus uteri cancer and uterus NOS cancer were codified separately in all editions (172 [ICD-7], 182 [ICD-9] and C54 [ICD-10] for corpus cancer; 174 [ICD-7], 179 [ICD-9] and C55 [ICD-10] for uterus NOS cancer. However, in the 8th edition, 182 was used for both corpus and uterus NOS cancer. They could only be distinguished with the 4th digit (182.0 for corpus cancer and 182.9 for uterus NOS cancer), but distinction was in many countries not possible by lack of this 4th digit. Other rare cancers of the uterus such as placenta tumours (chorioblastoma) were coded with 173 in the 7th edition and with 181 in the 8th and 9th ICD edition. Table 3. Additional WHO codes used to identify uterus cancers.

Cancer of the ICD edition Cervix Uterus

other parts Corpus or

uterus NOS Corpus or uterus

NOS or other Abbreviation CVX OTH CRPNOS CRPNOSOTH

7 A052 - - A053 8 A055 - - A056 9 B120 B121 B122 - 10 1037 - - 1038 In the WHO Mortality Database, still other non-ICD codes were used (see Table 3). A052, A055, B120 and 1037 were used to identify cancer of the cervix uteri, respectively in the 7th, 8th, 9th and 10th ICD period; A053 (ICD7), A056 (ICD8), and 1038 (ICD10) were used for the group of corpus uteri, uterus NOS and other uterus cancers. In the 9th ICD period, B122 was the code for the combined group of cancers of corpus uteri and uterus NOS, whereas B121 was the code for other uterine cancers including placenta cancer. The periods, in which the respective ICD-codes were used, are shown for each area in Table 4. Table 4. ICD codes used for classification of death causes in the WHO Mortality Database, by ICD-edition and area.

ICD Edition Area 7 8 9 10 Albania - - 1987-2004 - Andora - - - - Armenia - - 1981-2003 - Austria 1955-1968 1969-1979 1980-2001 2002-2004 Belarus - - 1981-2001 2002-2003 Belgium 1952-1967 1968-1978 1979-1997 - Bosnia & Herzegovina

- - 1985-1991 -

Bulgaria 1964-1967 1968-1979 1980-2004 - Croatia - - 1985-1994 1995-2005 Cyprus - - - - Czechoslovakia 1953-1967 1968-1978 1979-1991 - Czech Republic 1986-1993 1994-2005 Denmark 1951-1968 1969-1993 - 1994-2001 Estonia - - 1981-1996 1997-2004

15

ICD Edition Area 7 8 9 10 Finland 1952-1968 1969-1986 1987-1995 1996-2005 France 1950-1967 1968-1978 1979-1999 2000-2004 Georgia - - 1981-1997 1998-2001 Germany - - 1990-1997 1998-2004

East Germany 1973-1978 1980-1990 - West Berlin 1955-1967 1968-1970 - -

West Germany 1952-1967 1968-1978 1979-1990 - Greece 1961-1967 1968-1978 1979-2004 - Hungary 1955-1968 1969-1978 1979-1995 1996-2003 Iceland 1952-1970 1971-1980 1981-1995 1996-2005 Ireland 1950-1967 1968-1978 1979-2005 - Italy 1951-1967 1968-1978 1979-2002 - Latvia - - 1981-1995 1996-2004 Liechtenstein - - - - Lithuania - - 1981-1997 1998-2004 Luxembourg 1967-1970 1971-1978 1979-1997 1998-2004 Macedonia - - 1991-2003 - Malta 1965-1967 1968-1978 1979-1994 1995-2004 Moldova - - 1981-1995 1996-2006 Monaco - - - - Netherlands 1950-1968 1969-1978 1979-1995 1996-2004 Norway 1951-1968 1969-1985 1986-1995 1996-2004 Poland 1959-1968 1969-1979 1980-1996 1999-2005 Portugal 1955-1970 1971-1979 1980-2001 2002-2003 Romania 1959-1968 1969-1978 1980-1998 1999-2004 Russian Federation - - 1980-1998 1999-2005 San Marino - - - - Serbia & Montenegro

- - - 1997-2002

Slovakia - - 1992-1993 1994-2005 Slovenia - - 1985-1996 1997-2005 Spain 1951-1967 1968-1979 1980-1998 1999-2004 Sweden 1952-1968 1969-1986 1987-1996 1997-2004 Switzerland 1951-1968 1969-1994 - 1995-2004 Turkey - - - - Ukraine - - 1981-2004 2005 United Kingdom 1950-1967 1968-1978 1979-1999 2001-2004

England & Wales 1950-1967 1968-1978 1979-2000 2001-2004 Northern Ireland 1950-1967 1968-1978 1979-2000 2001-2004

Scotland 1950-1967 1968-1978 1979-1999 2000-2004 Yugoslavia 1960-1967 1968-1978 1979-1990 -

We used the following own abbreviated labels to recodify causes of deaths throughout all the periods: CVX (cervix uteri cancer), CRP (corpus uteri cancer), NOS (cancer uterus NOS), OTH (cancer of other parts of the uterus), CRPNOS (cancer of corpus uteri or uterus NOS), CRPNOSOTH (cancer of corpus uteri, uterus NOS or other parts of the uterus). The abbreviation UT is used for all cancers of the uterus together. The availability of mortality data for cancer from the different parts of the uterus is further detailed in Table 5, by area and period.

16

Table 5. Availability of data on mortality from cervix uteri cancer (CVX), corpus uteri cancer (CRP), uterus cancer not otherwise specificied (NOS), other uterine cancer (OTH), combination of CRP and NOS (CRPNOS) or combination of CRP, NOS and OTH (CRPNOSOTH) per area and period.

CVX CRP NOS OTH CRPNOS CRPNOSOTH Areas

Available Missing Available Missing Available Missing Available Missing Available Missing Available Missing Albania 1987-1989,

1992-2003 1990-1991 1987-1989,

1992-2003 1990-1991 1987-1989,

1992-2003 1990-1991 1987-1989,

1992-2003 1990-1991 1987-1989,

1992-2003 1990-1991 - -

Armenia 1981-1982, 1985-2003

1983-1984 - - - - - - 1981-1982, 1985-2003

1983-1984 - -

Austria 1955-2004 - 1969-2004 1955-1968 1969-2004 1955-1968 1969-2002 1955-1968, 2003-2004

1969-2004 1955-1968 1955-1979 1980-2004

Belarus 1981-1982, 1985-2003

1983-1984 - - - - - - 1981-1982, 1985-2003

1983-1984 - -

Belgium 1954-1997 - 1955-1976, 1979-1984,

1986, 1992-1997

1954, 1977-1978, 1985, 1987-1991

1955-1976, 1979-1984,

1986, 1992-1997

1954, 1977-1978, 1985, 1987-1991

1955-1997 1954 1968-1997 1954-1967 1954-1978 1979-1997

Bosnia & Herzegovina 1985-1991 - 1985-1991 - 1985-1991 - 1985-1991 - 1985-1991 - - - Bulgaria 1964-2004 - 1966-1967,

1980-1982, 1984-2004

1964-1965, 1968-1979,

1983

1966-1967, 1980-1982, 1984-2004

1964-1965, 1968-1979,

1983

1966-2004 - 1968-2004 1964-1967 1964-1979 1980-2004

Croatia 1985-2004 - 1985-2004 - 1985-2004 - 1985-1994, 1996

1995, 1997-2004

1985-2004 - - -

Czechoslovakia 1953-1991 - 1955-1967, 1979-1991

1953-1954, 1958-1978

1955-1967, 1979-1991

1953-1954, 1958-1978

1955-1991 1953-1954 1968-1991 1953-1967 1953-1978 1979-1991

Czech Republic 1986-2004 - 1986-2004 - 1986-2004 - 1986-1994, 1996, 2004

1995, 1997-2003

1986-2004 - - -

Denmark 1951-2001 - 1955-1981, 1988-2001

1951-1954, 1982-1987

1955-1981, 1988-2001

1951-1954, 1982-1987

1955-1993 1951-1954, 1994-2001

1969-2001 1951-1968 1951-1993 1994-2001

Estonia 1981-1982, 1985-2004

1983-1984 1994-2004 1981-1993 1994-2004 1981-1993 1994-1996 1981-1993, 1995-2004

1981-1982, 1985-2004

1983-1984 - -

Finland 1952-2004 - 1955-1983, 1985-1987, 1989-2004

1952-1954, 1984, 1988

1955-1983, 1985-1987, 1989-2004

1952-1954, 1984, 1988

1955-1995, 2000

1952-1954, 1996-1999, 2001-2004

1969-2004 1952-1968 1952-1986 1987-2004

France 1950, 1952-2003

1951 1955-2003 1950-1954 1955-2003 1950-1954 1955-2003 1950-1954 1968-2003 1950-1967 1950, 1952-1978

1951, 1979-2003

Georgia 1981-1982, 1985-1992, 1994-2001

1983-1984, 1993

1998-2001 1981-1997 1998-2001 1981-1997 - - 1981-1982, 1985-1992, 1994-2001

1983-1984, 1993

- -

17

CVX CRP NOS OTH CRPNOS CRPNOSOTH Areas

Available Missing Available Missing Available Missing Available Missing Available Missing Available Missing Germany+++ 1973-2004 - 1973-2004 - 1973-2004 - 1973-2002,

2004 1973-1967,

2003 1973-2004 - 1973-1978 1979-2004

Germany 1990-2004 - 1990-2004 - 1990-2004 - 1990-2002, 2004

2003 1990-2004 - - -

East Germany 1973-1978, 1980-1990

1979 1983-1990 1973-1982 1983-1990 1973-1982 1976, 1980-1990

1973-1975, 1977-1979

1976, 1980-1990

1973-1975, 1977-1979

1973-1978 1979-1990

West Germany 1952-1990 - 1969-1990 1952-1968 1969-1990 1952-1968 1968-1990 1952-1967 1968-1990 1952-1967 1952-1979 1980-1990 West Berlin 1955-1960,

1962-1970 1961 1968-1970 1955-1967 1968-1970 1955-1967 1968-1970 1955-1967 1968-1970 1955-1967 1955-1960,

1962-1970 1961

Greece 1961-2004 - 1966-2004 1961-1965 1966-2004 1961-1965 1966-2004 1961-1965 1968-2004 1961-1967 1961-1978 1979-2004 Hungary 1955-2003 - 1970-2003 1955-1969 1970-2003 1955-1969 1970-1995,

1997, 2001 1955-1969, 1996, 1998-

2000

1970-2003 1955-1969 1955-1978 1979-2003

Iceland 1952-2004 - 1971-2004 1952-1970 1971-2004 1952-1970 1971-1995 1952-1970, 1996-2004

1971-2004 1952-1970 1952-1980 1981-2004

Ireland 1950-2005 - 1955-1973, 1975-2005

1974 1955-1973, 1975-2005

1974 1955-2005 1950-1954 1968-2005 1950-1967 1950-1978 1979-2005

Italy 1951-2002 - 1955-2002 1951-1954 1955-2002 1951-1954 1956-2002 1951-1955 1968-2002 1951-1967 1951-1978 1979-2002 Latvia 1981-2004 - 1996-2004 1981-1995 1996-2004 1981-1995 - - 1981-2004 - - - Lithuania 1981-1982,

1985-2004 1983-1984 1993-2004 1981-1992 1993-2004 1981-1992 1993-1998,

2004 1981-1992, 1999-2003

1981-1982, 1985-2004

1983-1984 - -

Luxembourg 1967-2004 - 1979-2004 1967-1978 1979-2004 1967-1978 1979-1997 1967-1978, 1998-2004

1979-2004 1967-1978 1967-1978 1979-2004

Macedonia 1991-2003 - 1991-2003 - 1991-2003 - 1991-2003 - 1991-2003 - - - Malta 1965, 1967-

2004 1966 1978-2004 1965-1977 1978-2004 1965-1977 1978-1994 1965-1977,

1995-2004 1978-2004 1965-1977 1965, 1967-

1978 1966, 1979-

2004 Moldova 1981-1982,

1985-2004 1983-1984 1996-2004 1981-1995 1996-2004 1981-1995 1991-1996,

1998, 2002 1981-1990, 1997, 1999, 2000-2001, 2003-2004

1981-1982, 1985-2004

1983-1984 - -

Netherlands 1950-2004 - 1955-1969, 1972-2004

1950-1954, 1970-1971

1955-1969, 1972-2004

1950-1954, 1970-1971

1955-1997, 2000, 2002

1950-1954, 1998-1999, 2001, 2003-

2004

1969-2004 1950-1968 1950-1978 1979-2004

Norway 1951-2004 - 1955-1978, 1982-2004

1951-1954, 1979-1981

1955-1978, 1982-2004

1951-1954, 1979-1981

1955-1995, 2003-2004

1951-1954, 1996-2002

1969-2004 1951-1968 1951-1985 1986-2004

18

CVX CRP NOS OTH CRPNOS CRPNOSOTH Areas

Available Missing Available Missing Available Missing Available Missing Available Missing Available Missing Poland 1959-1996,

1999-2004 1997-1998 1961-1968,

1999-2004 1959-1960, 1969-1998

1961-1968, 1999-2004

1959-1960, 1969-1998

1961-1970, 1980-1989, 2001-2002

1959-1960, 1971-1979, 1990-2000, 2003-2004

1969-1970, 1980-1996, 1999-2004

1959-1968, 1971-1979, 1997-1998

1959-1979 1980-2004

Portugal 1955-2003 - 1984-2003 1955-1983 1984-2003 1955-1983 1980-2003 1955-1979 1980-2003 1955-1979 1955-1979 1980-2003 Romania 1959-1978,

1980-2004 1979 1970, 1999-

2004 1959-1969, 1971-1998

1970, 1999-2004

1959-1969, 1971-1998

1969-1978, 1980-2001,

2003

1959-1968, 1979, 2002,

2004

1969-1978, 1980-2004

1959-1968 1959-1978 1979-2004

Russian Federation 1980-2004 - 1999-2004 1980-1998 1999-2004 1980-1998 - - 1980-2004 - - - Serbia & Montenegro 1997-2002 - 1997-2002 - 1997-2002 - - - 1997-2002 - - - Slovakia 1992-2002 - 1992-2002 - 1992-2002 - 1992-1994,

1997-1998, 2001-2002

1995-1996, 1999-2000

1992-2002 - - -

Slovenia 1985-2004 - 1985-2004 - 1985-2004 - 1985-1996, 1999

1997-1998, 2000-2004

1985-2004 - - -

Spain 1951-2004 - 1955-1967, 1975-2004

1951-1954, 1968-1974

1955-1967, 1975-2004

1951-1954, 1968-1974

1955-1999, 2001, 2003-

2004

1951-1954, 2000, 2002

1968-2004 1951-1967 1951-1979 1980-2004

Sweden 1952-2002 - 1955-1969, 1987-2002

1952-1954, 1970-1986

1955-1969, 1987-2002

1952-1954, 1970-1986

1955-1996, 2000-2001

1952-1954, 1997-1999,

2002

1969-2002 1952-1968 1952-1986 1987-2002

Switzerland 1951-2004 - 1955-2004 1951-1954 1955-1994 1951-1954, 1995-2004

1955-1994 1951-1954, 1995-2004

1969-2004 1951-1968 1955-1994 1951-1954, 1995-2004

Ukraine 1981-1982, 1985-2004

1983-1984 - - - - - - 1981-1982, 1985-2004

1983-1984 - -

United Kingdom 1950-1999, 2001-2004

2000 1955, 1958-1962, 1967, 1979-1999, 2001-2004

1950-1954, 1956-1957, 1963-1966, 1968, 1969-1978, 2000

1955, 1958-1962, 1967, 1979-1999, 2001-2004

1950-1954, 1956-1957, 1963-1966, 1968, 1969-1978, 2000

1955, 1958-1962, 1967-1999, 2001-

2004

1950-1954, 1956-1957, 1963-1966,

2000

1968-1999, 2001-2004

1950-1967, 2000

1950-1978 1979-2004

England & Wales 1950-2004 - 1955-2004 1950-1954 1955-2004 1950-1954 1955-2004 1950-1954 1968-2004 1950-1967 1950-1978 1979-2004 Northern Ireland 1950-2004 - 1955-1967,

1979-2004 1950-1954, 1968-1978

1955-1967, 1979-2004

1950-1954, 1968-1978

1955-2000 1950-1954, 2001-2004

1968-2004 1950-1967 1950-1978 1979-2004

Scotland 1950-2004 - 1955-2004 1950-1954 1955-2004 1950-1954 1955-1999, 2002

1950-1954, 2000-2001, 2003-2004

1968-2004 1950-1967 1950-1978 1979-2004

19

CVX CRP NOS OTH CRPNOS CRPNOSOTH Areas

Available Missing Available Missing Available Missing Available Missing Available Missing Available Missing Yugoslavia 1961-1990 1960 1960-1967,

1970, 1979-1990

1968-1969, 1971-1978

1960-1967, 1970, 1979-

1990

1968-1969, 1971-1978

1960-1990 - 1968-1990 1960-1967 1961-1978 1960, 1979-1990

+++ Unified Germany

20

4.1.6. Age groups Eighteen age groups were considered: 17 groups with equal width of 5 years (0-4, 5-9, …, 80-84) and an 18th open group of women aged 85 or older. The world reference population was used to adjust for differences in age composition between areas and periods (18). 4.2. Reallocation of deaths caused by cancers of the uterus 4.2.1. Cancer of the uterus To compute the number of deaths from uterus cancer (UT), the sum was made of:

1. UT = CVX + CRP + NOS + OTH, or 2. UT = CVX + CRPNOS + OTH, or 3. UT = CVX + CRPNOSOTH, or. 4. UT = CVX + CRPNOS

The fourth option was used, when no data was available for OTH (separately or included as a group). No effort was done to adjust for this lack, since OTH cancers were very rare (<5 deaths/year) in all areas and periods. 4.2.2. Reallocation of causes of death Because the proportion of cervical cancer deaths that is included in the composed groups NOS, CRPNOS or CRPNOSOTH is not constant over time (ICD-editions), age and area, different reallocation procedures are required to study trends of cervical cancer mortality in the whole period for which data are available. Three different reallocation rules could be applied according to the proportion of NOS (< or ≥25%) or availability of CRPNOS(OTH). The three possible periods that can be distinguished according to this criterion are listed for each area in Table 6. Table 6. Range of years with proportion of NOS < 25%, proportion of NOS ≥ 25% or where the group of CRPNOS or CRPNOSOTH were used.

NOS Area <25% ≥ 25%

CRPNOS/CRPNOSOTH but no NOS

Albania - 1987-1989, 1992-2004 - Armenia - - 1981-1982, 1985-2003 Austria - 1969-2004 1955-1968 Belarus - - 1981-1982, 1985-2003 Belgium - 1955-1976, 1979-1984, 1986,

1992-1997 1954

Bosnia & Herzegovina

- 1985-1991 -

Bulgaria - 1966-1967, 1980-1982, 1984-2004

1964-1965, 1968-1979, 1983

Croatia - 1985-2005 - Czechoslovakia 1979, 1981-1991† 1955-1967, 1980† 1953-1954, 1968-1978 Czech Republic 1986-2005 - -

21

NOS Area <25% ≥ 25%

CRPNOS/CRPNOSOTH but no NOS

Denmark 1956, 1962-1964, 1966-1981, 1988-2001

1955, 1957-1961, 1965 1951-1954, 1982-1987

Estonia 1994, 1998-1999, 2003-2004 1995-1997, 2000-2002 1981-1982,1985-1993 Finland 1955-1958, 1963-1964, 1966,

1968-1983, 1985-1987, 1989-2005†

1959-1962, 1965, 1967† 1952-1954, 1984, 1988†

France - 1955-2004 1950, 1952-1954 Georgia 1998-2001 - 1981-1982, 1985-1992,

1994-2004 Germany+++ 1979 1973-1978, 1980-2004 -

East Germany 1983-1988, 1989-1990 - 1973-1978, 1980-1982 West Germany - 1952-1990 1952-1968

West Berlin 1968 1969-1970 1955-1960, 1962-1970 Greece - 1966-2004 1961-1965 Hungary 1980-1990, 1991-2005 1970-1979 1955-1969 Iceland 1973-1974, 1976, 1978 -1984,

1986-1988, 1990-1992, 1994-1995, 1999-2001

1972, 1975, 1977, 1985, 1989, 1993, 1996-1998, 2002-2005

1952-1970

Ireland 1967-1973, 1975-1977, 1984-1985, 1987-2005

1955-1966, 1978-1983, 1986 1950-1954, 1974

Italy - 1955-2002 1951-1954 Latvia 1996 1997-2004 1981-1992 Lithuania 1993-2004 - 1981-1982, 1985-1992 Luxembourg 1979-1997 1998-2005 1967-1978 Macedonia - 1991-2003 - Malta 1988, 1997, 2000, 2002, 2004 1978-1987, 1989-1996,

1998-1999, 2001, 2003 1965, 1967-1977

Moldova 1996-1998, 1999-2006 - 1981-1982, 1985-1995 Netherlands 1955-1962, 1972-2004 1963-1969 1950-1954, 1970-1971 Norway 1955-2004 - 1951-1954, 1979-1981 Poland 1999-2005 1961-1968 1959-1960, 1969-1979,

1980-1996 Portugal - 1984-2003 1955-1979, 1980-1983 Romania 1999-2004 - 1959-1968, 1969,

1971-1978, 1980-1998 Russian Federation 1999-2005 - 1980-1998 Serbia & Montenegro 2000 1997-1999, 2001-2002 - Slovakia 1992-2005 - - Slovenia 1993, 1995, 1998-1999, 2005 1985-1992, 1994, 1996-1997,

2000 -

Spain 2000-2004 1955-1967, 1975-1999 1951-1954, 1968-1974 Sweden 1960,1962-1969, 1987,1989-1991 1955-1959, 1961, 1988,

1992-2004 1952-1954, 1970-1986

Switzerland 19691994 1955-1968 1951-1954, 1995-2004 Ukraine - - 1981-1982, 1985-2005 UK 1955, 1958-1962, 1967,1979-1999,

2001-2004 - 1950-1954, 1956-1957,

1963-1966, 1968-1978 England & Wales 1955-2004 - 1950-1954 Northern Ireland 1996, 2000 1955-1967, 1979-1995,

1997-1999, 2001-2004 1950-1954, 1968-1978

Scotland 1967, 1969-2004 1955-1966 1950-1954, 1968 Yugoslavia - 1960-1967, 1970, 1979-1990 1968-1969, 1971-1978

+++ Unified Germany †: small gaps where %NOS was >25% or CRPNOS or CRPNOSOTH was used. For these cor CVX was estimated by interpolation.

22

Reallocation rule 1 For areas with a low proportion of NOS (NOS/UT < 25%), considered over all age groups combined, over a period of at least 10 years, and for a period where NOS was available, reallocation rule 1 was applied. First, within each age group i, the proportion of cervix and corpus cancer deaths among the deaths with code NOS, was estimated using equation 1 and 2:

)( ii

i CRPCVXCVXpCVX

+= Equation 1

)( ii

ii CRPCVX

CRPpCRP

+= Equation 2

Then, the proportion (pCVXi) and (pCRPi) were multiplied with the number of NOSi to obtain the estimated number of NOS cancers that were probably from cervical or corpus origin, respectively. The corrected number of cervix (corCVX) or corpus uteri (corCRP) cancer was obtained using equations (3) and (4).

)*( iiii pCVXNOSCVXcorCVX += Equation 3

)*( iiii pCRPNOSCRPcorCRP += Equation 4

The underlying assumption for reallocation rule 1 is that the death cause of NOS was applied at random. If this assumption would not correspond with the truth, the bias would be limited since the rule is only applied when pNOS<25%. Interpolation was applied when in an area with good data (pNOS<25%) over at least 10 years, a single or two consecutive years were found where pNOS ≥ 25% (Czechoslovakia: 1980; Denmark: 1965; Finland: 1965, 1967) or when no NOS was used during a very small period (Finland: 1984, 1988; Denmark: 1951-1954 and 1982-1987). Interpolation consisted in taking the average of corCVX for previous year and the year following the affected year or years. Reallocation rule 2 Allocation rule 2 was applied for areas where reallocation rule 1 was not applicable for all periods because pNOS >25% or because NOS was not available as a separate group but included in CRPNOS or CRPNOSOTH. The proportion of corrected cervical cancer (pcorCVX) was first computed from total uterus cancer corCVXi (see Equation 3) and UTi (see Section 4.2.1) as

ij ij ijpcorCVX corCVX UT= , for the corresponding age group i and year j of areas where realocation rule 1 was applied (source period). We then applied the imputation method to obtain the pcorCVXij for years where rule 1 was not applicable (target period). The imputation method was based on assumption that data from periods where rule 1 was not applicable were missing at random (MAR) (19-21). In general, a range of the most approximate 10 years was used as source period with independent variable (interaction between agei and yearj) where allocation rule 1 was applied to a target period preceding this source period (see Figure 1a). In this source period pcorCVXij is estimated from interaction between agei and yearj of the target period and the regression coefficients computed from the source period. The regression equation is presented as:

23

( )0 1ˆ ˆˆ *ij i jpcorCVX b b age year= + Equation 5

In case the estimated pcorCVXij was negative or greater than unity, we constrained them to be zero or 1 respectively. If the preceding source period contained only 5 or less years, a source range of only 6 years was used (see Figure 1b). If the target period was preceded and followed by source periods with <25% NOS, imputation was done using 3 years before and 3 years after the target period (see Figure 1c).

Figure 1. Principles of imputation used in reallocation rule 2

The estimated proportion of cervix uteri cancer deaths among all uterus cancer deaths ( ijVXCpcor ˆ i) was applied on the total number of uterus cancers deaths to compute the corrected number of cervical cancer deaths in the affected areas (see equation 6).

ˆ*ij ij ijcorCVX UT pcorCVX= Equation 6

Areas, where reallocation rule 1 or rules 1 and 2 were applied, were considered as potential reference areas, which can be used for reallocation in areas with more than 25% of NOS or areas with combined codes (see reallocation rule 3). The list of areas and periods where reallocation rules 1 and 2 were applied is presented in Table 7. Table 7. List of countries and periods by method of reallocation for death causes NOS or CRPNOS(OTH).

Reallocation rules Area 1 2 (*) 3

Albania 1987-2004 Armenia 1981-2003 Austria 1955-2004 Belarus 1981-2003 Belgium 1954-1997 Bosnia & Herzegovina 1985-1991

3 years

6 years

10 years

3 years

a

b

c

Source period Period with <25% NOS Period with >=25% NOS or CRPNOS

<=5 years

24

Reallocation rules Area 1 2 (*) 3

Bulgaria 1964-2004 Croatia 1985-2005 Czechoslovakia 1979-1991 1953-1978 (10) Czech Republic 1986-2005 Denmark 1962-1981,

1988-2001 1951-1961 (10), 1982-1987 (3+3)

Estonia 1981-2004 Finland 1963-2005 1952-1962 (10) France 1950-2004 Georgia 1981-2001 Germany+++ Σ

East Germany 1973-1990 West Germany 1952-1990

West Berlin 1955-1970 Greece 1961-2004 Hungary 1980-2005 1955-1979 (10) Iceland 1952-2005 Ireland 1967-1977,

1984-2005 1950-1966 (10), 1978-1983 (3+3)

Italy 1951-2002 Latvia 1981-2004 Lithuania 1993-2004 1981-1992 Luxembourg 1967-2004 Macedonia 1991-2003 Malta 1965-2004 Moldova 1981-2006 Netherlands 1955-1962,

1972-2004 1950-1954 (6),

1963-1971 (3+3)

Norway 1955-2005 1951-1954 (6) Poland 1959-2005 Portugal 1955-2003 Romania 1959-2004 Russian Federation 1980-2005 Slovakia 1992-2005 Slovenia 1985-2005 Spain 1951-2004 Sweden 1952-2004 Switzerland 1951-2004 United Kingdom Σ Σ Σ

England & Wales 1955-2004 1950-1954 (6) Northern Ireland 1950-2004

Scotland 1969-2004 1950-1968 (10) Ukraine 1981-2005 Yugoslavia 1960-1990

(*) number of years in the column (rule 1) used as source for imputation of lacking or inaccurate data in column (rule 2). +++Unified Germany

25

Reallocation rule 3 pcorCVX of reference areas were used as template to estimate the number of cervical cancer deaths from the total number of uterus cancer deaths in non-reference areas. corCVXia = UTia * pcorCVXaref, where the index i refers to age group, a for area and ref for the respective reference country. The list of reference areas used to correct the respective non-reference areas is shown in Table 8. Table 8. List of reference areas used to correct data from areas where >25% of uterine cancer deaths were of unspecified origin (NOS) or were included in mixed code groups.

Reference area Areas with >25% NOS or mixed codes (CRPNOS, CRPNOSOT) Finland Iceland, Sweden Hungary Albania, Armenia, Belarus, Bosnia and Herzegovina, Bulgaria,

Croatia, Georgia, Macedonia, Moldova, Poland, Romania, Russian Federation, Slovenia, Ukraine, Yugoslavia

Lithuania Estonia, Latvia The Netherlands Austria, Belgium, France, Germany (East Germany, West

Germany, West Berlin), Greece, Italy, Luxembourg, Malta, Portugal, Spain, Switzerland

England & Wales Ireland, Northern Ireland, Scotland Germany, United Kingdom For Germany, data were added from East- and West-Germany from 1973 to 1989 and data from the unified Germany were used thereafter. Data for corCVXi from England & Wales, Northern Ireland were summed to obtain those for the whole UK. 4.2.3. Imputation Imputation is the process of filling in missing observations with plausible values in order to have a complete data set. This method runs a regression by best-subset regression (19,20). It involves a number of auxiliary variables that are statistically related to the variable in which there are missing observations (21). The dependent variable is the variable whose missing values are to be imputed and the independent variables are the variables on which the imputations are to be made. We applied this method when estimating the number of corCVX for periods with great proportion of uterine cancer deaths that was NOS (pNOS ≥ 25%) or when only combined death causes CRPNOS or CRPNOSOTH were available.

26

4.3. Trend analyses

4.3.1. Mortality from uterus cancer among women younger than 45 years In spite of the common anatomical localization, cervix and corpus uteri cancer are completely different neoplastic diseases from pathological, etiological and epidemiological viewpoints. Corpus uteri cancer most often is diagnosed after the age of 50, whereas before the age of 45 years most uterine cancers arise from the cervix (5,16,17). We therefore analysed uterus cancer mortality in the age group 20 to 44 years as a proxy for the burden of cervix cancer which does not require any reallocation from mixed groups of death causes. Within, the targeted age-group (20-44 years), rates were standardized using the 5-year age group distribution of the world reference population. 4.3.2. Corrected number of deaths from cervix uteri cancer, mortality rates Subsequently, we will present trends in the corrected absolute number of deaths from cervix uteri cancer, the crude, standardised and cumulative cause-specific mortality rates by calendar yeara,b,c after application of the reallocation rules described earlier. Cohort-specific mortality rates are studied as well. Cohorts include women who are born in the same period, hence, are ageing together and are exposed to risks isolated in time. They constitute the diagonals in a table with period- and age-specific rates respectively in rows and columns (25). Since age groups and periods all span five years, the corresponding birth-cohorts are necessarily ten years large. Successive cohorts overlap partly and are usually indicated by their central year (10,25,26). For instance: women between 55 and 59 years old in the period 1975-79 will belong in the period 1985-89 to the age-group 65-69 and were between 35 and 39 years old in the period 1955-59. They all belong to the 1920 cohort (C1920), which means that they were born between 1915 and 1924 (see Lexis-diagram in Figure 2). Cohort effects The cohort-effect represents the relative risk of a certain cohort to die from breast cancer in comparison with the mean mortality rate of all generations together. This cohort-effect is calculated by an indirect standardisation method (27-29). It consists of the ratio of the number of observed deaths in a given cohort k over the number of expected deaths, if the average age-specific mortality rates are applied to the respective age segments of the population in cohort k.

a For definitions, methods of calculation of parameters and their 95% confidence intervals we refer to Jensen(22). b For the computation of 95% CIs around the SMR (standardised mortality ratios), we used the method proposed by Clayton (23): SMRlow/up=SMR:/*error factor. Error factor=exp(1.96*sqrt(1/O), where O is the number of observed deaths. c Cumulative mortality (CM) was computed as the complement of cumulated survival using the formula: CM=1-Π(1-e-ai*ΔT), where ai is the age-specific mortality rate over ΔT (=5 year age interval) (24).

27

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

1895 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990

Calendar time

Age

gro

up

Figure 2. A cohort can be presented as an oblique parallelogram in a Lexis-diagram, where calendar time and age constitute respectively the abscisse and the ordinate. Women in the age-category between 55 and 59 years in the period 1975-79 were in the period 1955-59 between 35 and 39 years old and will be between 1985 and 1990 in the group of 65-69 years. They belong to the cohort, born between 1915 and 1924, indicated as cohort C1920.

4.3.3. Join Point regression Mortality trends are often not linear over very long periods. In order to test for changes in slopes, we used the Join Point regression model, using a software developed recently by statisticians at the National Cancer Institute (30). The Join Point procedure allows the identification of periods with a distinct slope that can be separated by a number of breakpoints or "join points" (31). Modelling follows the principle of minimisation of the weighted sum of squared errors and the choice of the number of join points is based on permutation tests (31,32). 4.3.4. Standardised cohort mortality ratio The cohort-effect, or standardised cohort mortality ratio, represents the relative risk of a certain cohort of dying from cervical cancer compared to the mean mortality rate of all generations together. This cohort-effect is calculated by an indirect standardisation method (27,29). It consists of the ratio of the number of observed deaths in a given cohort, k, over the number of expected deaths if the average age-specific mortality rates are applied to the respective age segments of the population in cohort k.

28

4.4. Statistical software All statistical processing and analyses were carried out with the software package STATA, version 10.0 (Stata Corp, College Station, Texas, USA) at the exception of the Join Point regression.

29

5. RESULTS

Below, we present plots for all countries of the European continent with available data: 1. Mortality from uterus cancer among women between 20-44 years 2. Standardised cohort mortality ratio for uterus cancer among women aged 20-44 years 3. Crude rates of mortality from cervix cancer (certified and corrected for NOS) and uterus

cancer 4. Age-standardised rate of mortality from cervix cancer (certified and corrected for NOS)

and uterus cancer 5. Corrected age-specific mortality from cervical cancer by period 6. Corrected age-specific rates of mortality from cervical cancer by birth cohort 7. Standardised cohort mortality ratio (cervical cancer, corrected, all ages) 8. Joinpoint regression of standardised corrected cervical cancer mortality The first two series of plots involve mortality from cancer from all parts of the uterus, restricted to women between 20 and 44 years of age, where the large majority of cancers is of cervical origin. The next series of plots is based of the corrected number of cervical cancers where part of uterus NOS cancers are realocated to the cervix uteri accoring to the methods described in the material and methods section. Plots are ranked by alphatetic order of the country or area. For a detailed description of the trends, we refer to separate reports and papers derived from the report (33,34).

30

5.1. Mortality from uterus cancer among women between 20-44 years

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Albania

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Armenia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Austria

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belarus

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belgium0

510

15

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bosnia&Hezergovina

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bulgaria

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Croatia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czech Republic

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czechoslovakia0

510

15

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Denmark

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Estonia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Year

Finland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

France

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Georgia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Germany

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

East Germany

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Germany

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Berlin

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Greece

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Hungary

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Iceland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ireland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Italy

Figure 3. Age-standardised mortality from uterus cancer among women aged 20-44, by 5 year period.

31

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Latvia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Lithuania

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Luxembourg

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Macedonia0

510

15

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Malta

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Moldova

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Netherlands

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Norway

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Poland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Portugal

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Romania

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Russian Federation

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Serbia & Montenegro

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovakia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovenia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Spain

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Sweden

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Switzerland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

United Kingdom

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

England & Wales

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Northern Ireland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Scotland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ukraine

05

1015

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Yugoslavia

Figure 4. Age-standardised mortality from uterus cancer among women aged 20-44, by 5 year period.

32

5.2. Standardised cohort mortality ratio for uterus cancer among women aged 20-44 years

0

12

3

SC

MR

1910 1930 1950 1970

Birth cohort

Albania

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Armenia

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Austria

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Belarus

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Belgium0

12

3

SC

MR

1910 1930 1950 1970

Birth cohort

Bosnia & Hezergovina

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Bulgaria

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Croatia

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Czech Republic

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Czechoslovakia0

12

3

SC

MR

1910 1930 1950 1970

Birth cohort

Denmark

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Estonia

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Finland

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

France

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Georgia

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Germany

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

East Germany

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

West Germany

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

West Berlin

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Greece

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Hungary

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Iceland

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Ireland

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Italy

Figure 5. Standardised cohort mortality ratio for uterus cancer among women aged 20-44, by birth cohort.

33

0

12

3

SC

MR

1910 1930 1950 1970

Birth cohort

Latvia

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Lithuania

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Luxembourg

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Macedonia0

12

3

SC

MR

1910 1930 1950 1970

Birth cohort

Malta

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Moldova

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Netherlands

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Norway

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Poland

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Portugal

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Romania

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Russian Federation

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Serbia & Montenegro

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Slovakia

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Slovenia

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Spain

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Sweden

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Switzerland

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

United Kingdom

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

England & Wales

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Northern Ireland

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Scotland

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Ukraine

01

23

SC

MR

1910 1930 1950 1970

Birth cohort

Yugoslavia

Figure 6. Standardised cohort mortality ratio for uterus cancer among women aged 20-44, by birth cohort.

34

5.3. Crude rates of mortality from cervix cancer (certified and corrected for NOS) and uterus cancer

0

1020

3040

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Albania

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Armenia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Austria

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belarus

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belgium0

1020

3040

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bosnia & Herzegovina

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bulgaria

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Croatia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czech Republic

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czechoslovakia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Denmark

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Estonia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Finland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

France

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Georgia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Germany

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

East Germany

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Germany

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Berlin

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Greece

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Hungary

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Iceland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ireland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Italy

Uterus Cervix, corrected Cervix, certified

Figure 7. Crude rates of mortality from cervix cancer (certified & corrected for NOS) and uterus cancer.

35

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Latvia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Lithuania

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Luxembourg

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Macedonia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Malta

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Moldova

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Netherlands

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Norway

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Poland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Portugal

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Romania

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Russian Federation

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Serbia & Montenegro

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovakia0

1020

3040

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovenia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Spain

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Sweden

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Switzerland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

United Kingdom

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

England & Wales

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Northern Ireland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Scotland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ukraine

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Yugoslavia

Uterus Cervix, corrected Cervix, certified

Figure 7. Crude rates of mortality from cervix cancer (certified & corrected for NOS) and uterus cancer.

36

5.4. Age-standardised rate of mortality from cervix cancer (certified and corrected for

NOS) and uterus cancer 0

1020

30

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Albania

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Armenia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Austria

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belarus

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belgium0

1020

30

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bosnia & Herzegovina

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bulgaria

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Croatia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czech Republic

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czechoslovakia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Denmark

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Estonia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Year

Finland

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

France

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Georgia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Germany

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

East Germany

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Germany

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Berlin

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Greece

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Hungary

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Iceland

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ireland

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Italy

Uterus Cervix, corrected Cervix, certified

Figure 8. Age-standardised rate of mortality from cancer of the uterus, cervix (certified) and cervix (corrected).

37

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Latvia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Lithuania

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Luxembourg

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Macedonia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Malta

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Moldova

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Netherlands

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Norway

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Poland

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Portugal

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Romania

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Russian Federation

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Serbia & Montenegro

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovakia0

1020

30

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovenia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Spain

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Sweden

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Switzerland

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

United Kingdom

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

England & Wales

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Northern Ireland

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Scotland

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ukraine

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Yugoslavia

Uterus Cervix, corrected Cervix, certified

Figure 8. Age-standardised rate of mortality from cancer of the uterus, cervix (certified) and cervix (corrected).

38

5.5. Corrected age-specific mortality from cervical cancer by period

02

46

810

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Albania

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Armenia

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Austria

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belarus0

2040

6080

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Belgium

05

1015

2025

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bosnia & Herzegovina

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Bulgaria

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Croatia

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czech Republic

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Czechoslovakia

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Denmark

020

4060

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Estonia

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Finland

020

4060

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

France

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Georgia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Germany

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

East Germany

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Germany

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

West Berlin

05

1015

2025

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Greece

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Hungary

030

6090

120

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Iceland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ireland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Italy

80-84 70-74 60-64 50-54 40-44 30-34 20-24

Figure 9. Corrected age-specific rate of mortality from cervix uteri cancer by period

39

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Latvia

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Lithuania

015

3045

60

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Luxembourg

05

1015

2025

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Macedonia0

1020

3040

50

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Malta

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Moldova

020

4060

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Netherlands

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Norway

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Poland

020

4060

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Portugal

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Romania

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Russian Federation

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Serbia & Montenegro

010

2030

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovakia0

1020

3040

50

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Slovenia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Spain

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Sweden

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Switzerland

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

United Kingdom0

1020

3040

50

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

England & Wales

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Northern Ireland

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Scotland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Ukraine

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1950 1970 1990 2010

Period

Yugoslavia

80-84 70-74 60-64 50-54 40-44 30-34 20-24

Figure 9. Corrected age-specific rate of mortality from cervix uteri cancer by period

40

5.6. Corrected age-specific rates of mortality from cervical cancer by birth cohort

02

46

810

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Albania

010

2030

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Armenia

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Austria

010

2030

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Belarus0

2040

6080

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Belgium0

510

1520

25

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Bosnia & Herzegovina

010

2030

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Bulgaria

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Croatia

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Czech Republic

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Czechoslovakia

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Denmark

020

4060

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Estonia

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Finland

020

4060

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

France

010

2030

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Georgia

010

2030

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Germany

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

East Germany

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

West Germany

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

West Berlin

05

1015

2025

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Greece

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Hungary

030

6090

120

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Iceland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Ireland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Italy

80-84 70-74 60-64 50-54 40-44 30-34 20-24

Figure 10. Corrected age-specific rate of mortality from cervix uteri cancer by cohort.

41

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Latvia

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Lithuania

015

3045

60

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Luxembourg

05

1015

2025

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Macedonia

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Malta

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Moldova

020

4060

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Netherlands

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Norway

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Poland

020

4060

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Portugal

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Romania

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Russian Federation

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Serbia & Montenegro

010

2030

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Slovakia0

1020

3040

50

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Slovenia

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Spain

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Sweden

020

4060

80

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Switzerland

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

United Kingdom

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

England & Wales

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Northern Ireland

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Scotland

010

2030

40

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Ukraine

010

2030

4050

Dea

ths

/ 100

000

/ ye

ar

1870 1890 1910 1930 1950 1970 1990

Birth cohort

Yugoslavia

80-84 70-74 60-64 50-54 40-44 30-34 20-24

Figure 10. Corrected age-specific rate of mortality from cervix uteri cancer by cohort.

42

5.7. Standardised cohort mortality ratio (cervical cancer, corrected, all ages)

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Albania

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Armenia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Austria

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Belarus

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Belgium0

12

3

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Bosnia & Herzegovina

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Bulgaria

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Croatia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Czech Republic

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Czechoslovakia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Denmark

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Estonia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Finland

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

France

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Georgia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Germany

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

East Germany

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

West Germany

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

West Berlin

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Greece

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Hungary

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Iceland

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Ireland

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Italy

Figure 11. Standardized cohort mortality ratio with 95% confidence intervals

43

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Latvia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Lithuania

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Luxembourg

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Macedonia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Malta

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Moldova

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Netherlands

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Norway

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Poland

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Portugal

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Romania

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Russian Federation

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Serbia & Montenegro

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Slovakia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Slovenia

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Spain

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Sweden

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Switzerland

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

United Kingdom0

12

3

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

England & Wales

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Northern Ireland

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Scotland

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Ukraine

01

23

SC

MR

1870 1890 1910 1930 1950 1970

Birth cohort

Yugoslavia

Figure 11. Standardized cohort mortality ratio with 95% confidence intervals

44

5.8. Joinpoint regression of standardised corrected cervical cancer mortality

05

1015

Dea

ths

/ 100

000

/ ye

1970 1980 1990 2000

Year

Austria

05

1015

Dea

ths

/ 100

000

/ ye

1970 1980 1990 2000

Year

Belgium

05

1015

Dea

ths

/ 100

000

/ ye

1970 1980 1990 2000

Year

Bulgaria

05

1015

Dea

ths

/ 100

000

/ ye

1970 1980 1990 2000

Year

Czech Republic0

510

15

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Denmark

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Estonia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Finland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

France

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Germany

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Greece

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Hungary

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Ireland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Italy

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Latvia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Lithuania

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Luxembourg

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Netherlands

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Poland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Portugal

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Romania

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Slovakia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Slovenia

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Spain

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Sweden

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

England & Wales

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Nothern Ireland

05

1015

Dea

ths

/ 100

000

/ ye

ar

1970 1980 1990 2000

Year

Scotland

Figure 12. Corrected and age-standardized mortality rate: observed rates (dots), rates fitted by joint regression (lines).

45

6. DISCUSSION

The current discussion will be focused on the 27 EU Member States. A discussion on the other European countries can be considered further upon special requests. The current trend analyses confirms previous reports revealing the large contrasts in the burden of cervical mortality between the old and new Member States of the European Union (4,5). Moreover, our study indicates that these contrasts will even rise in the future since mortality rates continue to decrease in the western part of Europe, whereas in Eastern Europe and in the Baltic states they decrease at lower intensity (Czech Republic, Poland), remain constant at a high rate (Estonia, Slovakia) or even increase (Bulgaria, Latvia, Lithuania, Romania). Quality of data An important question is whether the applied correction for certification inaccuracies allows the study of the true rates of cervical cancer mortality. If the assumption of random allocation (applied in rule 1) is incorrect, the error would be limited since the rule is only applied when the proportion of NOS is rather small. For Finland, we compared our corrected cervical cancer mortality rate with those adjusted by linkage between the death cause register and the cancer registry. Both corrected rates overlapped well, indicating that - at least in this example - reallocation rule 2 provided satisfactory results (35).

010

020

030

040

0

Dea

ths

1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

Year

Uterus, RegistryUterus, WHOCervix, RegistryCorcervix, WHO

Figure 13. Trend of the number of deaths from cancer of the cervix and uterus in Finland (1953-2005). Red: sum of number of deaths from all uterine cancers (as reported by the Finnish Cancer Registry); Blue: idem (as registered in the WHO mortality database); green: corrected number of cervical cancer deaths obtained by linkage between the Finnish mortality and cancer registries; purlple: corrected number of cervical cancer deaths obtained by applying allocation rules 1 and 2.

Less evidence of reliable correction can be found for reallocation rule 3. The assumption that the Lithuanian proportions are applicable to those of Estonia and Latvia look plausible given the common background risk and history of preventive health care. However, the application of proportions from the Netherlands to adjust data from different countries, such as Austria, France, Germany, Greece, Italy, Portugal or Spain, could be considered problematic. With alternative assumptions, it might be possible to obtain different patterns in the adjusted trends. Moreover, it is unclear whether cause-specific registration of deaths

46

was accurate with respect to the uterus as a primary site; or whether other errors may have affected data quality such as duplicates of some disease groups due to the use of various coding rules). Concerning Spain, LLorca et al concluded that the rate of cervical cancer mortality, based on certified cervix cancer deaths, was increasing (36). This conclusion was considered as possibly spurious (37) since the proportion of uterus NOS cancer deaths progressively decreased (NOSj/UTj: 86% in the 1950s, 26% near the end of the 1990s). In a later study, cervical cancer mortality was corrected by considering fixed proportions of NOS as being of cervical origin (38). The conclusion was that mortality was increasing among younger women. We found a nearly horizontal recent trend among young women in the Spanish data. We believe that corrections need to be age- and period-specific. Nevertheless, we are aware that such adjustments using a non-representative template country could also yield incorrect results. In order to find more reliable solutions to correct for NOS and CRPNOS cancer deaths, we propose further research, involving linkages between mortality and cancer registries (39-41). These same procedures are required for the producing current regular cancer statistics. Cohort effects Strong cohort effects could be discerned and some were common to nearly all European countries. The continuous decrease in cohorts born in the first decades of the 1900s, observed for Finland, France and many other countries but not for Romania, may be due to poorly understood etiological (co-) factors, linked to improved social conditions and access to health care (42). Women born between 1920 and 1940 showed a progressively lower risk of dying from cervical cancer, whereas women born thereafter tended to have increasing risk. This cohort effect is most plausibly explained by changes in sexual behaviour resulting in higher rates of HPV infection in younger cohorts as shown from studies using Finnish serum biobanks (43,44). Available data on HPV prevalence from other countries concern recent periods (45) but historical data are lacking. Therefore it is impossible to use them to interpret trends. At most it can be noticed that Denmark had high mortality in the seventies and also has current high prevalence of HPV suggesting that the background risk without screening is higher. The increased frequency of smoking and oral contraception, both established risk factors for cervical cancer, may have contributed also to the recent rise of the SCMR. It is also possible that some other factors such as early diagnosis of invasive cancer among younger women due to increased access to gynaecological care may be responsible for cohort effects observed in the deaths rates. In the future, the cohort effects possibly will be influenced by prophylactic HPV vaccination and further by screening practices in vaccinated cohorts. Screening effects In another paper, included in this issue of the EJC, we showed that substantial reductions in incidence and mortality, observed in several countries correlated with the level of implementation of organised screening (46). Opportunistic screening also resulted in a reduction of cervical cancer incidence and mortality in several other West-European countries (9,16). Difference in coverage and quality of screening, most plausibly explain the large differences between old and new Member States. The declining trend of cervical cancer mortality was initiated before screening became commonly practiced. The fact that increased coverage in the target population did not result in further decrease in cervical cancer mortality, has been sometimes suggested as evidence for a failure of screening (47,48). However, this viewpoint ignores the strong recent cohort effects, which we have illustrated for three countries. It seems that screening has counter-balanced the effect of increased exposure to etiologic factors in younger

47

cohorts, by limiting the upward tendency of the SCMR. In countries without established screening programs, the cohort effect was steeper (see Figure 3, for Romania; and the small differences between grey and black bars in several black and grey bars) than in countries with well organised screening. The contrast between the 1970-74 and 2000-04 periods underestimate the effect of screening in Finland, where organised screening was established already in the 1960s, and where age-standardised corrected mortality rates dropped by 80% over the last 45 years (49). It was estimated from an age-period-cohort model, that without screening standardised cervical cancer mortality, in 2003-07 in Finland, would have been 6.5/105/year whereas observed rates were 0.7/105/year (50). The greatest contrasts over the studied 35-year span were observed in Austria and Luxembourg (ratios of 0.22 and 0.18, respectively). We cannot ascertain that these decreases should be explained exclusively as the effects of intensive opportunistic screening existing these countries (46). Because of the particularly high rate of total uterus mortality, in both countries in the earliest periods, we cannot exclude that poor quality of historical data has driven these negative slopes. Improved survival A recent trend study of the 5-year survival from cervical cancer revealed a slow but steady improvement of about 2% per year among cancer patients diagnosed in the period 1983-94 in Europe(51). No improvement was noted in the areas where survival was lowest (Central/Eastern Europe and UK). Reduction of the case fatality can be expected by down staging through expansion of screening and by improved treatment. Unfortunately, there is no systematic data currently available on the quality of cervical cancer treatment in Europe. Behind age-standardised trends, complex changes over time, age and birth cohort can be hidden which require more detailed analyses. We are currently performing age-period-cohort modelling of European mortality data and comparing incidence and mortality trends with the purpose of disentangling the separate effects of screening and exposure to risk factors. These studies provide indirect evidence of effectiveness of preventive measures. Ideally the evaluation of performance of secondary prevention should come from linkages of individual screening histories with cancer and mortality registries, as recently described in the 2nd edition of the European Guidelines for Quality Assurance in Cervical Cancer Screening(52).

48

7. ABBREVIATIONS ACP model: age-cohort-period model A-effect: age-effect AIC: Akaike's Information Criterion ASMR: age-standardised mortality rate ASR: age-standardised rate C-effect: cohort effect CI: 95% confidence interval CM: cumulative mortality EU: European Union EUNICE: European Network for Information on Cancer Epidemiology IARC: International Agency for Research on Cancer HPV: human papillomavirus ICD: International Classification of Diseases IPH: Scientific Institute of Public Health NCR: National Cancer Register NIS: Nationaal Instituut voor Statistieken NOS: not otherwise specified P-effect: period effect PYLL: potential life years lost RR: relative risk SCMR: standardised cohort mortality ratio SIPH: Scientific Institute of Public Health SMR: standardised mortality ratio UK: United Kingdom WHO: World Health Organisation WIV: Wetenschappelijk Instituut Volksgezondheid WSR: world standardised rate (based on direct standardisation using the world standard population)

49

8. FILES USED

Raw data files downloaded from the WHO website according to the icd edition 7, 8, 9 or 10 with corresponding population data: W:\Amidu\WHO_Mortality06062006\firstupdates\morticd7.dta, morticd8.dta, morticd9.dta, morticd10.dta and Pop_1.dta. The latest update was done on 23th October 2007. W:\Amidu\WHO_Mortality06062006\thirdupdates_15102007\Morticd9.csv, W:\Amidu\WHO_Mortality06062006\thirdupdates_15102007\Morticd10.csv, W:\Amidu\WHO_Mortality06062006\thirdupdates_15102007\POP.csv These files were then transformed in a Stata format and managed to produce the reference file W:\Amidu\WHO_Mortality06062006\thirdupdates_15102007\morteurop.dta. This reference file was then split into country files: W:\Amidu\WHO_Mortality06062006\europ_country\country_mort_data\country.dta These files were first managed to explore and complete the missingness of data and deal with NOS data: W:\Amidu\WHO_Mortality06062006\europ_country\age_0-85+\data_management\cr_country_mort.do Data were then analysed and different graphs were drawn: W:\Amidu\WHO_Mortality06062006\europ_country\age_0-85+\data_analysis\country_mort.do For the joinpoint regression analysis, the data was first prepared W:\Amidu\WHO_Mortality06062006\joinpoint_reg\country_jpreg.do and a txt file was produced with the data needed in the software W:\Amidu\WHO_Mortality06062006\joinpoint_reg\country.txt The running of this software produced a txt file with the results of the joinpoint regression W:\Amidu\WHO_Mortality06062006\joinpoint_reg\countryjp.txt Eight graphs were produced per country: W:\Amidu\WHO_Mortality06062006\graphs\country_asmr.gph W:\Amidu\WHO_Mortality06062006\graphs\country_cmr1.gph W:\Amidu\WHO_Mortality06062006\graphs\country_cmr.gph W:\Amidu\WHO_Mortality06062006\graphs\country_cohort1.gph W:\Amidu\WHO_Mortality06062006\graphs\country_cohort2.gph W:\Amidu\WHO_Mortality06062006\graphs\country_cumr.gph W:\Amidu\WHO_Mortality06062006\graphs\country_exp.gph W:\Amidu\WHO_Mortality06062006\graphs\country_period.gph These graphs were then combined in eight graphs: W:\Amidu\WHO_Mortality06062006\graphs\asmr1-3.gph W:\Amidu\WHO_Mortality06062006\graphs\cmr1-3.gph W:\Amidu\WHO_Mortality06062006\graphs\cohort1-3.gph W:\Amidu\WHO_Mortality06062006\graphs\cohort11-13.gph W:\Amidu\WHO_Mortality06062006\graphs\cumr1-3.gph W:\Amidu\WHO_Mortality06062006\graphs\period1-3.gph W:\Amidu\WHO_Mortality06062006\graphs\scmr1-3.gph W:\Amidu\WHO_Mortality06062006\graphs\jpr.gph

50

9. REFERENCES

(1) Council of the European Union. Council Recommendation of 2 December 2003 on Cancer Screening. Off J Eur Union 2003; 878: 34-8.

(2) Arbyn M, Primic Zakelj M, Raifu AO, Grce M, Paraskevaidis E, Diakomanolis E et al. The burden of cervical cancer in South-East Europe at the beginning of the 21st century. Coll Antropol 2007; 31: 7-10.

(3) European Commission. European Guidelines for Quality Assurance in Cervical Cancer Screening. 2nd ed. Arbyn M, Anttila A, Jordan J, Ronco G, Schenck U, Segnan N, Wiener H, Daniel J, von Karsa L, editors. Luxembourg: Office for Official Publications of the European Communities, 2008. pp. 1-291. Accessible at: http://bookshop.europa.eu/eubookshop/publicationDetails.action?pubuid=547021&offset=0 and http://bookshop.europa.eu/eubookshop/download.action?fileName=ND7007117ENC_002.pdf&eubphfUid=10027853&catalogNbr=ND-70-07-117-EN-C

(4) Arbyn M, Autier P, Ferlay J. Burden of cervical cancer in the 27 member states of the European Union: estimates for 2004. Ann Oncol 2007; 18: 1425-7.

(5) Arbyn M, Raifu AO, Autier P, Ferlay J. Burden of cervical cancer in Europe: estimates for 2004. Ann Oncol 2007; 18: 1708-15.

(6) IARC Monograph Working Group. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol 90: Human Papillomaviruses. Cogliano V, Baan R, Straif K, et al, editors. Lyon: IARCPress, 2007. pp. 1-670.

(7) IARC. Cervix Cancer Screening. IARC Handbooks of Cancer Prevention. Vol. 10. Lyon: IARCPress, 2005. pp. 1-302.

(8) Laara E, Day NE, Hakama M. Trends in mortality from cervical cancer in the Nordic countries: association with organised screening programmes. Lancet 1987; 1: 1247-9.

(9) Bray F, Loos AH, McCarron P, Weiderpass E, Arbyn M, Moller H et al. Trends in cervical squamous cell carcinoma incidence in 13 European countries: changing risk and the effects of screening. Cancer Epidemiol Biomarkers Prev 2005; 14: 677-86.

(10) Arbyn M, Van Oyen H, Sartor F, Tibaldi F, Molenberghs G. Description of the influence of age, period and cohort effects on cervical cancer mortality by loglinear Poisson models (Belgium, 1955-94). Arch Public Health 2002; 60: 73-100.

(11) Karim-Kos HE, de Vries E, Soerjomataram I, Lemmens V, Siesling S, Coebergh JW. Recent trends of cancer in Europe: A combined approach of incidence, survival and mortality for 17 cancer sites since the 1990s. Eur J Cancer 2008; 44: 1345-89.

(12) Adami H-O, Ponten J, Sparen P, Bergstrom R. Survival trend after invasive cervical cancer diagnosis in sweden before and after cytologic screening. 1960-1984. Cancer 1994; 73: 140-7.

51

(13) Anttila A, Laara E. Cervix cancer: Geographical correlations. In: Sankila R, Demaret E, Hakama M et al, editors. Evaluation and Monitoring of Screening Programmes. Brussels, Luxemburg: Europe Against Cancer Programme, 2000: 77-97.

(14) Arbyn M, Geys H. Trend of cervical cancer mortality in Belgium (1954-94): tentative solution for the certification problem of not specified uterine cancer. Int J Cancer 2002; 102: 649-54.

(15) Loos AH, Bray F, McCarron P, Weiderpass E, Hakama M, Parkin DM. Sheep and goats: separating cervix and corpus uteri from imprecisely coded uterine cancer deaths, for studies of geographical and temporal variations in mortality. Eur J Cancer 2004; 40: 2794-803.

(16) Levi F, Lucchini F, Negri E, Franceschi S, La Vecchia C. Cervical cancer mortality among young women in Europe: patterns and trends. Eur J Cancer 2000; 36: 2266-71.

(17) Jenssen OM, Estève J, Moller H, Renard H. Cancer in the European Community and its member states. Eur J Cancer 1990; 26: 1167-256.

(18) Waterhouse JAH, Muir CS, Shanmugaratnam K. Cancer Incidence in Five Continents. 4 ed. Lyon: International Agency for Research on Cancer, 1976.

(19) Goldstein R. sed10: Patterns of missing data. Stata Technical Bulletin 1996; 32: 12-3.

(20) Little RJA, Rubin DB. Statistical Analysis with Missing Data. Second Edition ed. New York: Wiley, 2002. pp. 1-408.

(21) Durrant GB. Imputation Methods for Handling Item-Nonresponse in the Social Sciences: A Methodological Review. University of Southampton: ESRC National Centre for Research Methods and Southampton Statistical Sciences Research Institute (S3RI), 2005.

(22) Jenssen OM, Parkin DM, MacLennan R, Muir CS, Skeet RG. Cancer Registration Principles and Methods. 95 ed. International agency for Research on Cancer, 1991. pp. 1-288.

(23) Clayton D, Hills M. Statistical Models in Epidemiology. Oxford: Oxford University Press, 1995.

(24) Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic Research: Principles and Quantitative Methods. New York: Van Nostrand Reinhold, 1982. pp. 1-529.

(25) Kupper LL, Janis JM, Karmous A, Greenberg BG. Statistical age-period cohort analysis : a review and critique. J Chron Dis 1985; 38: 811-30.

(26) Moolgavkar SH. Analysis of vital statistics data. In: Rothman KJ, Greenland S, editors. Modern Epidemiology. Philadelphia: Lippincott-Raven Publishers, 1998: 481-97.

(27) Beral V. Cancer of the cervix: a sexually transmitted infection? Lancet 1974; may 25: 1037-40.

52

(28) Osmond C, Gardner MJ, Acheson DA. Analysis of trends in cancer mortality in England and Wales during 1951-80 separating changes associated with period of birth and period of death. BMJ 1982; 284: 1005-8.

(29) Osmond C, Gardner MJ. Age, Period and Cohort Models Applied to Cancer Mortality Rates. Stat Med 1982; 1: 245-59.

(30) National Cancer Institute. Join Point Regression Programme.Version 3.2. (Software produced by the Statistical Research and Application Branch). Silver Spring, USA, 2008. Acessible at: http://srab.cancer.gov/joinpoint/

(31) Kim H-J, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med 2000; 19: 335-51.

(32) Lerman PM. Fitting segmented regression models by grid search. Appl Stat 1980; 29: 77-84.

(33) Arbyn M, Raifu AO, Bray F, Weiderpass E, Anttila A. Trends of cervical cancer mortality in the member states of the European Union. Eur J Cancer 2009; in-press.

(34) Arbyn M, Antoine J, Valerianova Z, Mägi M, Stengrevics A, Smailyte G et al. Trends in cervical cancer incidence and mortality in the Baltic Countries, Bulgaria and Romania. Tumori 2009; in-vited paper.

(35) Arbyn, M., Raifu, A. O., Antoine, J. Trend of cervical cancer mortality in Europe. IPH/EPI-REPORTS 99, 1-50. 2008. Brussels, Scientific Institute of Public Health.

(36) Llorca J, Prieto MD, Delgado-Rodr¡guez M. Increase in cervical cancer mortality in Spain, 1951-1991. J Epidemiol Community Health 1999; 53: 408-11.

(37) Bosch FX. Trends in cervical cancer mortality. J Epidemiol Community Health 1999; 53: 392.

(38) Llorca J, Rodriguez-Cundin P, Dierssen-Sotos T, Prieto-Salceda D. Cervical cancer mortality is increasing in Spanish women younger than 50. Cancer Letters 2006; 240: 36-40.

(39) Zakelj MP, Kirn VP, Skrlec F, Selb J. Can we rely on cancer mortality data? Checking the validity of cervical cancer mortality data for Slovenia. Radiol Oncol 2002; 35: 243-7.

(40) Suzan, F., Séblain, C., Boussac-Rarebska, M., Poillot, M. L., Arveux, P., Laurent, F., and Carré, N. Survey on the distribution of "uterus cancer, cervix corpus cancer and not otherwise specified cancer" among death certificates for cervix cancer and corpus uteri cancer, France. Bull.Epidemiol.Hebdom. 9, 79-81. 2007.

(41) Capocaccia R, Martina L, Inghelmann R, Crocetti E, De L, V, Falcini F et al. A method to estimate mortality trends when death certificates are imprecisely coded: an application to cervical cancer in Italy. Int J Cancer 2009; 124: 1200-5.

(42) Parkin DM, Bray FI, Devesa SS. Cancer burden in the year 2000. The global picture. Eur J Cancer 2001; 37: 4-66.

53

(43) Laukkanen P, Koskela P, Pukkala E, Dillner J, Laara E, Knekt P et al. Time trends in incidence and prevalence of human papillomavirus type 6, 11 and 16 infections in Finland. J Gen Virol 2003; 84: 2105-9.

(44) Lehtinen M, Kaasila M, Pasanen K, Patama T, Palmroth J, Laukkanen P et al. Seroprevalence atlas of infections with oncogenic and non-oncogenic human papillomaviruses in Finland in the 1980s and 1990s. Int J Cancer 2006; 119: 2612-9.

(45) De Vuyst H, Clifford G, Li N, Franceschi S. HPV infection in Europe. Eur J Cancer 2009; in-press.

(46) Arbyn M, Rebolj M, de Kok IM, Becker N, O'Reilly M, Andrae B. Challenges for organising cervical screening programmes in the 15 old member states of the European Union. Eur J Cancer 2009; in -press.

(47) Giard RWM, Coebergh JWW, Wijnen JA. Causes of low effectiveness of population screening for cervival cancer. Ned Tijdschr Geneeskd 1996; 141: 317-21.

(48) Raffle AE. Deaths from cervical cancer began falling before screening programmes were established. BMJ 1997; 315: 953-4.

(49) Anttila A, Nieminen P. Cervical cancer screening programme in Finland with an example on implementing alternative screening methods. Coll Antropol 2007; 31 Suppl 2:17-22.: 17-22.

(50) Hristova L, Hakama M. Effect of screening for cancer in the Nordic countries on deaths, cost and quality of life up to the year 2017. Acta Oncol 1997; 36 Suppl 9: 1-60.

(51) Bielska-Lasota M, Inghelmann R, Poll-Franse L, Capocaccia R. Trends in cervical cancer survival in Europe, 1983-1994: a population-based study. Gynecol Oncol 2007; 105: 609-19.

(52) Ronco G, Anttila A. Chapter 7: Summary table of key performance indicators. In: Arbyn M, Anttila A, Jordan J et al, editors. European Guidelines for Quality Assurance in Cervical Cancer Screening, 2nd edition. Luxembourg: Office for Official Publications of the European Communities, 2008: 231-42.

54