Area Abbonati
LoginPassword
doi   10.1700/386.4536




Life tables for world-wide comparison of relative survival for cancer (CONCORD study)

Paolo Baili, Andrea Micheli, Roberta De Angelis, Hannah K Weir, Silvia Francisci, Mariano Santaquilani, Timo Hakulinen, Manuela Quaresma, Michel P Coleman, CONCORD Working Group
1Descriptive Epidemiology and Health Planning Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy; 2Istituto Superiore di Sanità, National Center of Epidemiology, Surveillance and Health Promotion, Cancer Epidemiology Unit, Rome, Italy; 3Division of Cancer Prevention and Control, Centers for Disease Control and Prevention, Atlanta, Georgia, USA; 4Finnish Cancer Registry, Helsinki, Finland; 5Cancer Research UK Cancer Survival Group, Non-Communicable Disease Epidemiology Unit, London School of Hygiene and Tropical Medicine, London, UK; 6Members of the CONCORD Working Group (see p 667)

Key words: CONCORD, life tables, mortality, relative survival.

abstract


Background. The CONCORD study compares population-based relative survival from cancer using data from cancer registries in five continents. To estimate relative survival, general mortality life tables are required. Available statistics are incomplete, so various approaches are used to construct complete life tables. This article outlines how the life tables were constructed for CONCORD; it compares life expectancy at birth between 101 populations covered by cancer registries in 31 countries and compares the impact of two approaches to the deployment of life tables in relative survival analysis.
Methods. The CONCORD approach, using specific mathematical methods, produced complete (single-year-of-age) life tables by sex, cancer registry area, calendar year (1990-1999) and race (only in the USA). In order to study the impact of different approaches, we compared relative survival in the USA using the US national life table, centered on the relevant census years, and the CONCORD approach. We estimated relative survival in each American participating cancer registry for patients diagnosed with breast (women), colorectal or prostate cancer during 1990-1994 and followed up to 1999.
Results. Average life expectancy at birth during 1990-1999 varied in CONCORD cancer registry areas from 64 to 78 years in males and from 71 to 84 years in females. It increased during the 1990s more in men than in women. In the USA, it was lower in blacks than in whites. Relative survival in American populations was lower with the CONCORD approach, which incorporates trends and geographic variation in background mortality, than with the USA census life tables.
Conclusions. International variation in background mortality by geographic area, calendar time, race, age and sex is wide. We suggest that in international comparisons of cancer relative survival, complete life tables that are specific for cancer registry area, calendar year and race should be used.

Introduction
International comparisons of population-based cancer relative survival provide valuable information about cancer outcomes and progress in cancer control for clinicians, patients and policy-makers1. Relative survival estimates are included among the cancer control indicators prepared by the European Cancer Health Indicators Project (www.tumori.net/eurochip), designed to improve European cancer information systems and to promote their use in reducing cancer inequalities.
In Europe, the EUROCARE-3 study2 provided relative survival estimates for cancer patients with data from population-based cancer registries in 22 countries. In the USA, the Surveillance, Epidemiology and End Results (SEER) Program of the US National Cancer Institute has regularly produced relative survival estimates for participating cancer registries3. Results from the EUROCARE project and the SEER Program have indicated that survival for most adult cancers is higher in the USA than in Europe4-7.
The CONCORD project, originally designed to provide a systematic comparison of cancer survival between Europe and the USA, has now provided comparable estimates of relative survival for 31 countries in all five continents, covering patients diagnosed in 1990-1994 for four cancers of substantial public health importance: breast cancer in women, and cancers of the colon, rectum and prostate8.
Relative survival in a group of cancer patients is defined as the ratio of the observed probability of survival and the expected probability of survival if the cancer patients had been subject only to the all-cause (general) mortality rates in the general population from which they were drawn9,10. General mortality varies by age (the risk of death in adults increases with age), by sex (men always have higher mortality), by geographic area (the risk of death differs between countries, but also by sub-populations within countries), by year (falling in recent decades in developed countries11) and by race* (e.g., between blacks and whites in the USA12). In order to control for background mortality variation in international cancer survival comparisons, the CONCORD protocol13 specified that life tables should incorporate death risks for each combination of these variables and should be reconstructed with mathematical methods when they were not directly available from official statistics.
Life tables have been deployed differently to estimate relative survival in Europe (EUROCARE) and in the USA (SEER Program), and this can affect the comparison of relative survival. In both approaches, the life tables used are specific for sex and single year of age. However, in the EUROCARE approach, life tables are also specific for each cancer registry area and each calendar year of the study period14, whereas in the USA, the SEER Program usually uses the overall US national, race-specific life table centered on the most recent census, for all cancer registry areas in the SEER Program and over a decade (SEER approach). However, the publicly available SEER*Stat software, which provides a convenient procedure for the analysis of SEER data and other cancer registry databases, also allows users to deploy any set of life tables for the calculation of relative survival. The choice of approach reflects the different purpose of the analyses: EUROCARE compares cancer survival in a set of countries with strikingly different economies and health systems15, whereas the SEER Program compares cancer survival in a set of populations within a single country.
For world-wide comparison of relative survival, in the CONCORD approach, we constructed complete life tables by sex and single calendar year (1990-1999) for all participating cancer registry populations. Race-specific life tables were also reconstructed where possible and where comparable information on race was available for the cancer patients: this only proved possible for the USA.
In this article, we outline (a) how the CONCORD life tables were constructed and document the differences in life expectancy at birth, by participating cancer registry area, calendar year and (in the USA) race. We also show (b) the impact of different approaches to the deployment of life tables for relative survival analysis; we used the US data to show the impact on relative survival estimates of using life tables with the CONCORD approach, by cancer registry area and calendar year, instead of the more usual SEER approach, with a national life table for a decade and for all cancer registry areas.
Material and methods
Data on general mortality by age, sex (and, in the USA, race) and calendar year or period were obtained for 101 registries in 31 countries (Table 1). The data varied considerably in detail by registry. In the USA, CONCORD cancer registries cover 16 states and six metropolitan areas, with populations ranging from 466,000 inhabitants in Wyoming to 31 million in California8. These areas were covered either by SEER registries or by registries participating in the US Center for Disease Control and Prevention (CDC) National Program of Cancer Registries (NPCR). Life tables for the USA are produced by the CDC National Center for Health Statistics (NCHS) and are directly available for the entire US population by age, sex, calendar year and race16. The SEER*Stat database includes data on the population (US state and country) and the number of deaths from all causes by 5-year age class, sex, calendar year and cancer registry area.
(a) The objective was to construct complete life tables containing the probability of death (all causes of death combined) in each country or region, for each calendar year between 1990 and 1999 and, in the USA, by race. We used several standard approaches to produce complete life tables estimating probabilities of death by single year of age by registry (Table 1).

Table 1 - Characteristics of general population mortality data submitted and methods used to estimate complete life tables by single calendar year in the CONCORD study

 

Cancer registry

Mortality data submitted

Complete life tables estimated for CONCORD

 

 

 

 

 

 

 

 

 

Width of age

Highest

Calendar years

Single calendar

Methodb

 

 

bands (yr)

age

or periodsa

yearsa

 

Africa

 

 

 

 

 

 

 Algeria

Setifc

5

100

2000

1990, … 1999

Elandt-Johnson; Fraction

 

 

 

 

 

 

 

America, Central & South

 

 

 

 

 

 Brazil

Both registriesd

10

80

1990, ... 1999

1990, ... 1999

Akima

 Cuba

Cuba (national)

5

85

1990, ... 1999

1990, ... 1999

Elandt-Johnson

 

 

 

 

 

 

 

America, North

 

 

 

 

 

 Canada

British Columbia

15 (for 0-14)

90

1990, ... 1999

1990, ... 1999

Elandt-Johnson

 

 

1 (for 15-44)

 

 

 

 

 

 

5 (for 45-90)

 

 

 

 

 

Manitoba

1

90

1990, ... 1999

1990, ... 1999

Elandt-Johnson

 

Nova Scotia

1 (for 0-70)

90

1990, ... 1999

1990, ... 1999

Elandt-Johnson

 

 

5 (for 70-90)

 

 

 

 

 

Ontario

15 (for 0-14)

90

1990, ... 1999

1990, ... 1999

Elandt-Johnson

 

 

1 (for 15-44)

 

 

 

 

 

 

5 (for 45-90)

 

 

 

 

 

Saskatchewan

15 (for 0-14)

90

1990, ... 1999

1990, ... 1999

Elandt-Johnson

 

 

1 (for 15-44)

 

 

 

 

 

 

5 (for 45-90)

 

 

 

 

 USA

All registriese

5

85

1990, ... 1999

1990, ... 1999

Elandt-Johnson

 

 

 

 

 

 

 

Asia

 

 

 

 

 

 

 Japan

Fukui

5

95

1990-94, 1995-99

1990, ... 1999

Elandt-Johnson; Fraction

 

Osaka

1

99

1990, ... 1999

1990, ... 1999

Exponential

 

Yamagata

1

90

1991, ... 1994

1991, ... 1994

Elandt-Johnson for adult ages

 

 

 

99

1990, 1995, ... 1999

 

No change

 

 

 

 

 

 

 

Europef

 

 

 

 

 

 

 Ireland

Ireland (national)

1

105

1985-87, 1990-92,

1990, … 1999

Fraction

 

 

 

 

1995-97, 2001-03

 

 

 The Netherlands

North Netherlands

5

95

1990, … 1999

1990, … 1999

Elandt-Johnson

 Switzerland

Three registriesf

1

99

1990, … 1999

 

No change

 UK

Two registriesf

1

99

1989-91, 1998-00

1990, … 1999

Fraction

 

 

 

 

 

 

 

Oceania

 

 

 

 

 

 

 Australia

All registriesg

1

99

1990, ... 1999

1990, ... 1999

Exponential

 

 

 

 

 

 

 

a“1991-1995” means that a single data set was provided for that period, whereas “1991, ... 1995” means data were provided for each year from 1991 to 1995 inclusive. bExponential, to obtain death probabilities from death rates; Elandt-Johnson, to obtain complete life tables from abridged (5-year) life tables; Elandt-Johnson for adult ages, to estimate death probabilities for ages 75+; Akima, to obtain complete life tables from abridged (10-year) life tables; Fraction, to estimate life tables for each calendar year. cLife tables for Algeria in 2000. dCampinas SP, Goiânia GO. eAtlanta GA, California, Los Angeles CA, San Francisco CA, Colorado, Connecticut, Florida, Hawaii, Idaho, Iowa, Louisiana, Michigan, Detroit MI, Nebraska, New Jersey, New Mexico, New York [State], New York City NY, Rhode Island, Seattle WA, Utah, Wyoming. Life tables reconstructed by race: whites, blacks and all races. Life tables for blacks were not reconstructed for Hawaii, Idaho, New Mexico, Utah, Wyoming. fInformation only for registries that were not included in EUROCARE-313. Switzerland: Graubunden-Glarus, St Gall-Appenzell, Valais; UK: England (national), Northern Ireland. gAustralian Capital Territory, New South Wales, Northern Territory, Queensland, South Australia, Tasmania, Victoria, Western Australia.


Equivalent details for most of the European participating cancer registries have already been published
14. We used the exponential method to obtain probabilities of death (qx) from death rates17; the fraction method to interpolate life tables for each calendar year14; the Elandt-Johnson method to obtain complete life tables from abridged life tables with 5-year age classes; the Elandt-Johnson method for adult ages to estimate the probability of death at ages 75 and over using the Gompertz distribution18,19; and the Akima method to obtain complete life tables from abridged life tables with 10-year age classes19,20. For the USA, the SEER*Stat database allowed us the construction of life tables for both SEER and NPCR registry areas, by calendar year and race21. Five-year abridged life tables were estimated from these data, and converted to complete life tables with the Elandt-Johnson method. We constructed complete life tables for whites and for all races combined for all 22 cancer registry areas and for blacks in 17 of the 22 areas (Table 1). We could not construct robust life tables for blacks in the states of Hawaii, Idaho, New Mexico, Utah and Wyoming, because the populations were too small.
(b) In the CONCORD study, anonymized individual tumor records on cancer patients diagnosed during 1990-1994 and followed up to 31 December 1999 were supplied to agreed standards by the cancer registries, then subjected to uniform quality control and analyzed centrally8. For the present paper, only the US individual data were used. The present analyses aimed to compare relative survival estimates with different life table approaches. The SEER*Stat software3 was used to compare relative survival estimated with the Hakulinen method10. We used (A) the US race-specific census life tables for 1990 and 2000 applied to the periods 1990-1995 and 1996-1999 respectively (thus following the SEER approach), and (B) the life tables constructed for CONCORD for the period 1990-1999, specific for cancer registry area, single calendar year and race (CONCORD approach) (Table 2). Thus for example, with the SEER approach, the general population probability of survival of a black male cancer patient resident in Los Angeles in 1995 will be the survival probability of a black male of the same age in the USA in 1990, whereas it will be the survival probability of a black male of the same age resident in Los Angeles in 1995 using the CONCORD approach. We compared raw (not age-standardized) estimates of relative survival obtained with the two sets of life tables. We present the absolute differences between survival estimates, e.g., 15% is shown as 5% (not 50%) higher than 10%. Logarithmic transformation was used to estimate normal confidence intervals of the relative survival estimates2.

Table 2 - Characteristics of life tables used for relative survival estimates in US cancer registries participating in the CONCORD study

Approach

Life table characteristics

 

 

 

 

 

 

 

Age

Sex

Area

Year(s)

Race

 

 

 

 

 

 

SEER

Single year of age

Male

USA (national)

1990 (for years 1990, … 1995)

White,

 

 

Female

 

2000 (for years 1996 ,… 1999)

Black,

 

 

 

 

 

All races*

 

 

 

 

 

 

CONCORD

Single year of age

Male

Registry area

1990, 1991, … 1999

White,

 

 

Female

 

 

Black,

 

 

 

 

 

All races*

 

 

 

 

 

 

*The “all-races” life tables were used for cancer patients who were not of white or black race; or of unknown race; or of black race in Hawaii, Idaho, New Mexico, Utah and Wyoming.



Results
Life expectancy at birth following the CONCORD
approach
Average life expectancy during the decade 1990-99 in the 101 cancer registry populations participating in the CONCORD study varied from 63.7 years for males in Estonia to 77.6 years for males in Japan, a range of 13.9 years (Table 3). For females, the range was from 70.9 years in Algeria to 83.7 years in Japan, a difference of 12.8 years. The median life expectancy among these populations was 74.2 years for men and 80.3 years for women. Life expectancy was below this median in all US black populations, in Cuba, Brazil and Algeria, and in all Eastern European areas (Table 3). The range of life expectancy in Canada and in US white populations was similar to that in northern and western Europe (Figure 1). Average life expectancy during the decade 1990-1999 was always lower in US blacks than in US whites (Figure 1).






Life expectancy increased in almost all countries and regions during the decade 1990-1999. Figure 2 shows average life expectancy during 1990-1994 (x-axis) against the average life expectancy during 1995-1999 (y-axis) for each population (specific by registry area and race) and both sexes. Most points are above the bisection. Life expectancy increased more in men than in women, and more in populations where life expectancy was low during 1990-1994. The largest increases for life expectancy at birth between 1990-1994 and 1995-1999 were seen in the black population of New York City: +4.8 years in males and +2.3 years in females (Table 3).

Table 3 - Life tables constructed for CONCORD study: average life expectancy at birth (years) and difference (years) in life expectancy at birth between 1995-1999 and 1990-1994

Cancer registrya

Country

World region

Male

Female

 

 

 

 

 

 

 

 

 

 

1990-1999

Differenceb

1990-1999

Differenceb

 

 

 

 

 

 

 

Fukui

Japan

Asia

77.6

+0.6

83.7

+0.9

Yamagata

Japan

Asia

77.0

+0.9

83.1

+1.2

Iceland

Iceland

Northern Europe

76.6

+0.3

80.8

+0.6

Sweden

Sweden

Northern Europe

76.5

+1.3

81.7

+0.9

Australian Capital Territory

Australia

Oceania

76.5

+1.0

81.3

0.0

Macerata

Italy

Western Europe

76.4

+1.1

82.2

+0.8

Tuscany

Italy

Western Europe

76.4

+1.1

82.2

+0.8

Osaka

Japan

Asia

76.3

+1.1

82.5

+1.4

Geneva

Switzerland

Western Europe

76.1

+1.9

82.6

+0.8

Romagna

Italy

Western Europe

76.1

+1.2

82.3

+0.7

Ragusa

Italy

Western Europe

75.9

+0.7

80.4

+0.9

British Columbia

Canada

Canada

75.9

+1.0

81.6

+0.6

Utah (White)

USA

USA-White

75.9

+0.7

80.9

-0.1

Western Australia

Australia

Oceania

75.7

+1.0

81.4

+0.9

Victoria

Australia

Oceania

75.7

+1.4

81.2

+1.0

Navarra

Spain

Western Europe

75.6

+0.6

82.6

+0.9

Basel

Switzerland

Western Europe

75.6

+1.5

81.5

+1.0

South & West

UK

Northern Europe

75.5

+1.1

80.4

+0.8

East Anglia & Oxford

UK

Northern Europe

75.5

+1.0

80.0

+0.7

Ontario

Canada

Canada

75.5

+1.1

81.0

+0.5

South Australia

Australia

Oceania

75.5

+1.4

81.3

+1.1

Tarragona

Spain

Western Europe

75.3

+1.6

81.4

+1.1

Modena

Italy

Western Europe

75.2

+0.8

81.6

+0.8

Grisons-Glaris

Switzerland

Western Europe

75.2

+1.5

81.9

+1.1

St Gall

Switzerland

Western Europe

75.2

+1.3

81.6

+0.6

Saskatchewan

Canada

Canada

75.2

+0.3

81.4

-0.1

Connecticut (White)

USA

USA-White

75.2

+1.0

80.8

+0.2

Tyrol

Austria

Western Europe

75.1

+1.3

80.9

+1.1

Queensland

Australia

Oceania

75.1

+1.2

81.0

+0.9

New South Wales

Australia

Oceania

75.1

+1.6

81.0

+1.2

Amsterdam

Netherlands

Western Europe

75.1

+0.8

80.8

+0.3

South Thames

UK

Northern Europe

75.1

+1.1

80.0

+0.7

Seattle WA (White)

USA

USA-White

75.0

+1.1

80.5

+0.1

Latina

Italy

Western Europe

74.9

+1.1

81.3

+1.1

Parma

Italy

Western Europe

74.9

+1.2

81.8

+1.4

Idaho (White)

USA

USA-White

74.9

+0.8

80.5

+0.4

Torino

Italy

Western Europe

74.9

+1.5

81.2

+1.4

Iowa (White)

USA

USA-White

74.9

+0.8

80.9

0.0

Colorado (White)

USA

USA-White

74.9

+1.2

80.5

+0.1

North Netherlands

Netherlands

Western Europe

74.8

+0.6

80.7

+0.2

Sassari

Italy

Western Europe

74.8

+1.0

81.4

+0.9

Cote d’Or

France

Western Europe

74.7

+1.3

82.6

+1.0

Norway

Norway

Northern Europe

74.7

+1.2

80.6

+0.7

Nebraska (White)

USA

USA-White

74.7

+0.9

80.9

+0.1

Veneto

Italy

Western Europe

74.7

+1.3

81.8

+1.0

Manitoba

Canada

Canada

74.6

+0.6

80.3

-0.2

Isère

France

Western Europe

74.6

+1.2

82.4

+1.0

Nova Scotia

Canada

Canada

74.6

+0.8

80.6

+0.1

Hawaii (White)

USA

USA-White

74.5

+0.6

80.8

+1.1

Rhode Island (White)

USA

USA-White

74.5

+1.1

80.7

+0.7

New Jersey (White)

USA

USA-White

74.4

+1.3

80.1

+0.5

Malta

Malta

Western Europe

74.4

+0.9

79.1

+1.5

Valais

Switzerland

Western Europe

74.3

+1.8

81.7

+0.9

Varese

Italy

Western Europe

74.3

+1.8

81.6

+1.1

Tasmania

Australia

Oceania

74.3

+1.9

80.0

+1.0

Genova

Italy

Western Europe

74.3

+1.6

81.1

+1.3

Florida (White)

USA

USA-White

74.2

+1.2

80.9

+0.2

Murcia

Spain

Western Europe

74.2

+0.8

80.3

+0.7

England

UK

Northern Europe

74.2

+1.2

79.4

+0.8

Trent

UK

Northern Europe

74.2

+1.0

79.2

+0.8

Atlanta GA (White)

USA

USA-White

74.2

+1.6

80.3

+0.1

Michigan (White)

USA

USA-White

74.1

+1.0

79.7

+0.3

Ferrara

Italy

Western Europe

74.1

+0.9

81.0

+1.3

Wyoming (White)

USA

USA-White

74.0

+0.5

79.8

+0.2

West Midlands

UK

Northern Europe

74.0

+1.0

79.0

+0.9

Detroit MI (White)

USA

USA-White

74.0

+1.1

79.6

+0.4

California (White)

USA

USA-White

73.9

+1.7

80.0

+0.5

Eindhoven

Netherlands

Western Europe

73.9

+0.7

79.7

+0.6

Wales

UK

Northern Europe

73.9

+0.8

79.0

+0.5

San Francisco CA (White)

USA

USA-White

73.9

+2.6

80.5

+0.7

Granada

Spain

Western Europe

73.8

+0.7

80.5

+0.7

New Mexico (White)

USA

USA-White

73.7

+0.9

80.2

+0.4

Basque Country

Spain

Western Europe

73.6

+1.0

82.0

+0.8

New York State (White)

USA

USA-White

73.5

+2.1

79.8

+0.8

Mallorca

Spain

Western Europe

73.5

+1.2

81.2

+1.4

Northern Yorkshire

UK

Northern Europe

73.5

+1.1

78.5

+0.9

Los Angeles CA (White)

USA

USA-White

73.4

+2.1

79.9

+0.8

Calvados

France

Western Europe

73.2

+1.3

82.1

+1.0

Northern Ireland

UK

Northern Europe

73.2

+1.0

78.8

+0.6

Cuba

Cuba

South and Central America

73.1

+0.8

77.0

+1.1

Mersey and North Western

UK

Northern Europe

73.0

+0.9

78.2

+0.7

Bas-Rhin

France

Western Europe

73.0

+1.2

81.0

+0.8

Denmark

Denmark

Northern Europe

73.0

+1.0

78.2

+0.6

Ireland

Ireland

Northern Europe

72.9

+0.9

78.4

+0.8

Saarland

Germany

Western Europe

72.6

+1.2

78.9

+0.9

Finland

Finland

Northern Europe

72.5

+1.3

80.0

+1.0

Louisiana (White)

USA

USA-White

72.2

+1.1

78.8

+0.1

Scotland

UK

Northern Europe

72.0

+0.8

77.5

+0.7

Portugal

Portugal

Western Europe

71.3

+0.4

78.5

+0.5

New York City NY (White)

USA

USA-White

71.1

+3.6

79.0

+1.3

Slovenia

Slovenia

Western Europe

70.4

+1.6

78.1

+1.2

Colorado (Black)

USA

USA-Black

70.1

+2.5

76.3

-0.1

Cracow

Poland

Eastern Europe

69.7

+1.6

76.8

+0.8

Seattle WA (Black)

USA

USA-Black

69.6

+1.9

76.5

+0.5

West Bohemia

Czech Rep

Eastern Europe

69.6

+2.1

76.1

+1.5

Northern Territory

Australia

Oceania

69.0

+2.3

74.2

+2.3

Rhode Island (Black)

USA

USA-Black

69.0

+4.1

75.4

+1.2

Warsaw

Poland

Eastern Europe

68.9

+1.3

76.8

+0.6

Slovakia

Slovakia

Eastern Europe

68.1

+1.2

76.3

+0.7

Iowa (Black)

USA

USA-Black

67.7

+0.7

75.1

-0.4

Connecticut (Black)

USA

USA-Black

67.4

+3.1

76.2

+0.6

Nebraska (Black)

USA

USA-Black

67.1

+1.1

73.6

-1.4

Setifc

Algeria

Africa

67.1

-

70.9

-

California (Black)

USA

USA-Black

67.1

+2.6

74.8

+0.9

New York State (Black)

USA

USA-Black

67.1

+4.3

75.8

+2.0

Goiania

Brasil

South and Central America

67.0

+0.3

75.5

+0.6

Campinas

Brasil

South and Central America

66.6

-0.2

76.6

+1.0

New York City NY (Black)

USA

USA-Black

66.6

+4.8

76.0

+2.3

Florida (Black)

USA

USA-Black

66.3

+2.5

74.2

+1.2

Atlanta GA (Black)

USA

USA-Black

65.7

+2.4

74.7

+1.1

New Jersey (Black)

USA

USA-Black

65.7

+3.1

73.9

+1.1

Los Angeles CA (Black)

USA

USA-Black

65.7

+3.0

74.6

+1.0

San Francisco CA (Black)

USA

USA-Black

65.2

+2.7

74.6

+0.6

Michigan (Black)

USA

USA-Black

65.0

+1.9

73.6

+0.7

Louisiana (Black)

USA

USA-Black

64.8

+1.6

73.7

+0.3

Detroit MI (Black)

USA

USA-Black

64.0

+2.2

73.3

+0.5

Estonia

Estonia

Eastern Europe

63.7

+0.9

74.8

+1.3

 

 

 

 

 

 

 

aRanked by male life expectancy in 1990-1999. bDifferences between average life expectancy at birth in 1995-1999 and 1990-1994. cOne life table used for all calendar years.


Comparing CONCORD and SEER approaches
Life expectancy in the SEER approach was fixed for all 22 populations (71.8 years and 74.1 years in males respectively for 1990 and 2000; 78.8 years and 79.5 years in females respectively for 1990 and 2000) because it was derived from the overall US census national life table in 1990 and 2000. Average life expectancy in most of the US states and metropolitan cancer registry areas participating in CONCORD for the periods 1990-1995 and 1996-1999 was greater than or similar to the US census life expectancy respectively for 1990 and 2000 (Table 4). Louisiana, Detroit (MI), Atlanta (GA) and New York City (NY) were the only areas in which life expectancy was lower than the US census life expectancy both for males and females for both periods: major differences were in Louisiana. In contrast, the highest average life expectancy was always in Hawaii: 4.0 years higher for males and 3.2 years higher for females during 1990-1995 than the US census life expectancy for 1990, and 2.5 years higher for males and 3.0 years for females during 1996-1999 than the US census life expectancy for 2000. Life expectancy in the USA increased between 1990-1995 and 1996-1999 in all areas for males and in most areas for females (Table 4).

Table 4 - Life expectancy at birth in the USA, derived from CONCORD life tables and differences (in years) with life expectancy at birth derived from US Census life tables, all races

CONCORD US Area a

Male

Female

 

CONCORD

Difference

CONCORD

Difference

CONCORD

Difference

CONCORD

Difference

 

 

from US

 

from US

 

from US

 

from US

 

 

Censusb

 

Censusc

 

Censusd

 

Censuse

 

1990-1995

1990

1996-1999

2000

1990-1995

1990

1996-1999

2000

 

 

 

 

 

 

 

 

 

Louisiana

69.5

-2.3

70.8

-3.3

77.2

-1.6

77.4

-2.1

Detroit MI

71.0

-0.8

72.4

-1.7

77.8

-1.0

78.3

-1.2

Atlanta GA

71.1

-0.7

72.9

-1.2

78.5

-0.3

78.9

-0.6

New York City NY

68.8

-3.0

73.0

-1.1

77.7

-1.1

79.4

-0.1

Michigan

72.3

+0.5

73.5

-0.6

78.7

-0.1

79.0

-0.5

New Mexico

72.9

+1.1

74.0

-0.1

79.8

+1.0

80.3

+0.8

Florida

72.7

+0.9

74.3

+0.2

79.9

+1.1

80.3

+0.8

Wyoming

73.5

+1.7

74.3

+0.2

79.6

+0.8

79.6

+0.1

New York State

71.6

-0.2

74.4

+0.3

78.8

0.0

79.9

+0.4

New Jersey

72.7

+0.9

74.5

+0.4

79.0

+0.2

79.8

+0.3

Los Angeles CA

72.3

+0.5

74.7

+0.6

79.4

+0.6

80.2

+0.7

Nebraska

74.1

+2.3

74.9

+0.8

80.6

+1.8

80.6

+1.1

Rhode Island

73.7

+1.9

75.1

+1.0

80.2

+1.4

80.9

+1.4

California

73.1

+1.3

75.1

+1.0

79.8

+1.0

80.3

+0.8

Idaho

74.5

+2.7

75.3

+1.2

80.3

+1.5

80.8

+1.3

Iowa

74.4

+2.6

75.3

+1.2

80.8

+2.0

80.9

+1.4

Connecticut

74.0

+2.2

75.3

+1.2

80.3

+1.5

80.7

+1.2

San Francisco CA

72.5

+0.7

75.4

+1.3

80.1

+1.3

80.9

+1.4

Colorado

74.2

+2.4

75.6

+1.5

80.4

+1.6

80.4

+0.9

Seattle WA

74.5

+2.7

75.7

+1.6

80.4

+1.6

80.6

+1.1

Utah

75.5

+3.7

76.2

+2.1

80.8

+2.0

80.9

+1.4

Hawaii

75.8

+4.0

76.6

+2.5

82.0

+3.2

82.5

+3.0

 

 

 

 

 

 

 

 

 

aRanked by average male life expectancy during 1996-99 column. bUS Census life expectancy at birth in males for 1990 was 71.8 years. cUS Census life expectancy at birth in males for 2000 was 74.1 years. dUS Census life expectancy at birth in females for 1990 was 78.8 years. eUS Census life expectancy at birth in females for 2000 was 79.5 years.


The variation in all-cause mortality by time and place affects the international comparability of relative survival estimates. In the USA, relative survival estimates derived using the CONCORD approach to the deployment of life tables were systematically lower than those derived with the SEER approach to life tables (taken as the reference values) (Table 5). The only exception was Louisiana. The differences in colorectal cancer survival were generally greater for men. The largest differences were seen in Hawaii, where general population mortality is considerably lower than the US national average. For Hawaii, the absolute difference in relative survival was -1.6% for breast cancer in women, -3.3% and -2.1% for males and females, respectively, for colorectal cancer, and -5.7% for prostate cancer.
For each cancer, the CONCORD estimate of relative survival at five years in all registries combined fell below the 95% confidence interval for the SEER estimate (Table 5).



Discussion
The CONCORD study was originally designed to compare relative survival from cancer in the USA (including cancer registries participating in either the SEER or NPCR Programs) and Europe, using a single methodological approach. Relative survival is the ratio between the observed probability of survival in the cancer patients and the expected probability of survival based on general population mortality. A key difference exists, however, between the USA (SEER) and Europe (EUROCARE) in how life tables are used in the estimation of relative survival. To avoid problems of this type, the CONCORD study constructed complete life tables that were specific for sex, cancer registry area, calendar year and, where possible, race. These life tables reflect the wide variations in mortality between and within the participating 101 populations and the systematic increases in life expectancy with time during the decade 1990-1999.
We used this approach also in the USA because we expected for most areas that US relative survival estimates based on life tables by cancer registry area and calendar year (CONCORD approach) would be lower than those based on US census national life tables (i.e., US life table in 1990 for the years 1990-1995 and US life tables in 2000 for 1996-1999). This is because most CONCORD cancer registry areas in 1990-1995 and 1996-1999 had higher life expectancy than the USA in 1990 and 2000, that is higher expected survival than for the USA as a whole in census years. Relative survival under this approach thus reflects the survival of the cancer patients relative to that of the general population of that sex and race and in that place and time.
Relative survival estimates were, in fact, lower with the CONCORD approach than with the SEER approach. The single exception was Louisiana, where average life expectancy was lower than that of the US national population. Differences in relative survival between the SEER and CONCORD approaches were greater for men than for women, because general population mortality in participating areas of the USA during the decade 1990-1999 varied more widely from the US national mortality in males than it did for females. Differences in the survival estimates between the two approaches were always greatest in Hawaii, because general population mortality differed more widely from the US national average mortality in that state than elsewhere.
The analysis of the role of life tables in relative survival for different cancer sites is a complex matter. Under the hypothesis of invariability of expected survival ES1 and ES2, differences between two relative survival estimates obtained with different life tables




will increase with increasing prognosis (that is, with increasing observed survival, OS). In our database, the patient age distribution was similar for prostate and colorectal cancer in men but not for breast and colorectal cancer in women. This means that prostate cancer patients and male colorectal cancer patients had similar expected survivals (ES1 estimated with US census life tables and ES2 estimated with CONCORD life tables). Consequently, we expected larger differences




in relative survival using different life tables for prostate cancer than for colorectal cancer in men, because prostate cancer has a better prognosis (as can be seen in Table 5).
Possible limits of the CONCORD approach regard principally the instability of estimated life table for small populations, which can affect the standard errors of the relative survival estimates. To avoid these problems, three-year moving averages of expected survival can be used instead of life tables for single calendar years. The CONCORD life tables do not include socio-economic status among the sources of variation in background mortality, because few countries have adequate population mortality data for this purpose22.
In conclusion, for international comparisons of cancer survival, it is important to use standard methods in order to provide policy-makers with robust and comparable information for cancer control. After analyzing secular trends and global variations in all-cause mortality and variations by race in some countries, we suggest that relative survival estimates for international comparison should, if possible, be based on stable, complete (single year-of-age) life tables that capture background mortality in the area (country, region, city) where the cancer patients live by sex, single calendar year and race.
CONCORD Working Group
Africa Algeria: M Hamdi Chérif (Sétif Cancer Registry, Sétif).
America, Central and South Brazil: N Mahayri, DC Moreira Filho (Cancer Registry of Campinas, Campinas SP); MP Curado (Cancer Registry of Goiânia, Goiás GO); S Koifman* (National School of Public Health, Rio de Janeiro); G Azevedo e Silva* (Rio de Janeiro State University); Cuba: L Fernandes Garrote (National Cancer Registry, Havana).
America, North Canada: AJ Coldman, M McBride (British Columbia Cancer Registry, Vancouver); A Demers*, E Kliewer, D Turner (Manitoba Cancer Registry, Winnipeg); R Dewar (Nova Scotia Cancer Registry, Halifax); E Holowaty, L Marrett, D Mishri (Ontario Cancer Registry, Toronto); J Tonita (Saskatchewan Cancer Registry, Saskatchewan); J Hatcher, Y Mao (Health Canada, Ottawa); A-M Ugnat, C Waters (Case Surveillance Division, Ottawa); USA: JL Young* (Metro Atlanta Cancer Registry, Atlanta GA); HK Weir* (Centers for Disease Control and Prevention, Atlanta GA); WE Wright, D Yin (California State Cancer Registry, Sacramento CA); D Deapen (Los Angeles County Cancer Surveillance Program, Los Angeles CA); D West (Greater Bay Area Cancer Registry, San Francisco CA); K Bol, R Bott, J Finch (Colorado Central Cancer Registry, Denver CO); A Polednak (Connecticut Tumor Registry, Hartford CT); J MacKinnon (Florida Tumor Registry, Miami FL); MT Goodman (Hawaii Tumor Registry, Honolulu HI); S Carson, C Johnson (Cancer Data Registry of Idaho, Boise ID); CF Lynch (State Health Registry of Iowa, Iowa City IO); V Chen (Louisiana Tumor Registry, New Orleans LA); G Copeland (Michigan Tumor Registry, Lansing MI); J Graff (Metropolitan Detroit Cancer Surveillance System, Detroit MI); V Filos, S Frederick (Nebraska Cancer Registry, Lincoln NE); B Kohler (New Jersey State Cancer Registry, Trenton NJ); C Wiggins (New Mexico Tumor Registry, Albuquerque NM); CC McLaughlin, MJ Schymura (New York State Cancer Registry, Albany NY); J Fulton, D Rousseau (Rhode Island Cancer Registry, Providence RI); M Potts, S Schwartz (Cancer Surveillance System of Western Washington, Seattle WA): R Dibble, N Stroup (Utah Cancer Registry, Salt Lake City UT); J Brockhouse, J Grandpré (Wyoming Cancer Surveillance Program, Cheyenne WY).
Asia Japan: M Fujita (Fukui Cancer Registry, Fukui); W Ajiki, H Tsukuma* (Osaka Cancer Registry, Osaka); T Matsuda (Yamagata Cancer Registry, Yamagata).
Europe Austria: W Oberaigner (Tyrol Cancer Registry, Innsbruck); Czech Republic: J Holub (West Bohemia Cancer Registry, Prague); Denmark: HH Storm (Danish Cancer Society, Department of Cancer Prevention and Documentation, Copenhagen); Estonia: T Aareleid, M Rahu (Estonian Cancer Registry, Tallinn); Finland: T Hakulinen* (Finnish Cancer Registry, Helsinki); France: G Hédelin, M Velten (Bas-Rhin Cancer Registry, Strasbourg); G Launoy, J Macé-Lesech (Calvados Digestive Cancer Registry, Caen); G Chaplain (Côte d'Or Gynaecologic Registry, Dijon), J Faivre (Côte d'Or Digestive Tract Registry, Dijon); M Colonna (Isère Cancer Registry, Meylan); Germany: H Ziegler (Saarland Cancer Registry, Saarbrucken); Iceland: L Tryggvadóttir (Icelandic Cancer Registry, Reykjavik); Ireland: H Comber (National Cancer Registry of Ireland, Cork); Italy: S Ferretti (Ferrara Cancer Registry, Ferrara); M Vercelli (Liguria Cancer Registry, Genova); F Albertoni, E Conti, F Pannozzo (Latina Cancer Registry, Rome); F Pannelli, S Vitarelli (Macerata Cancer Registry, Camerino); C Allemani, P Baili, F Berrino*, L Ciccolallo, G Gatta*, A Micheli*, M Sant* (National Cancer Institute, Milan); ME Artioli, M Federico, M Ponz de Leon (Modena Cancer Registry, Modena); V De Lisi, L Serventi (Parma Cancer Registry, Parma); L Gafà, R Tumino (Ragusa Cancer Registry, Ragusa); F Falcini (Romagna Cancer Registry, Forlì); R Capocaccia*, R De Angelis, S Francisci, M Santaquilani, A Verdecchia* (Istituto Superiore di Sanità, Rome); M Budroni, R Cesaraccio (Sassari Cancer Registry, Sassari); S Patriarca, R Zanetti (Torino Cancer Registry, Torino); E Crocetti, E Paci (Tuscany Cancer Registry, Firenze); P Contiero, P Crosignani, G Tagliabue (Lombardy Cancer Registry, Varese); S Guzzinati, P Zambon (Venetian Cancer Registry, Padova); Malta: M Dalmas (Malta National Cancer Registry, Valletta); Netherlands: O Visser (Amsterdam Cancer Registry, Amsterdam); R Otter, M Schaapveld (North Netherlands Cancer Registry, Groningen); JW Coebergh (South Netherlands Cancer Registry, Eindhoven); Norway: A Andersen, F Langmark (Cancer Registry of Norway, Oslo); Poland: J Rachtan (Krakow Cancer Registry, Krakow); M Bielska-Lasota, Z Wronkoski, M Zwierko (Warsaw Cancer Registry, Warsaw); Portugal: AM da Costa Miranda (Southern Portugal Cancer Registry, Lisbon); Slovakia: M Ob sˇitniková, I Plesˇko (National Cancer Registry of Slovakia, Bratislava); Slovenia: V Pompe-Kirn, M Primic-Zˇakelj (Cancer Registry of Slovenia, Ljubljana); Spain: I Izarzugaza (Basque Country Cancer Registry, Vitoria-Gasteiz); C Martinez Garcia, MJ Sánchez-Pérez (Granada Cancer Registry, Granada); I Garau (Mallorca Cancer Registry, Palma de Mallorca); MD Chirlaque, C Navarro Sanchez (Murcia Cancer Registry, Murcia); E Ardanaz, C Moreno (Navarra Cancer Registry, Pamplona); J Galceran (Tarragona Cancer Registry, Reus); Sweden: TA Alvegård, L Barlow (Cancer Registry of Sweden, Stockholm); Switzerland: G Jundt (Basel Cancer Registry, Basel); J-M Lutz*, M Ussel (Geneva Cancer Registry, Geneva); H Frick (Graubunden-Glarus Cancer Registry, Chur); S Ess (St Gallen-Appenzell Cancer Registry, St Gallen); I Konzelmann (Valais Cancer Registry, Sion); UK - England: T Davies, S Godward, J Rashbass (Eastern Cancer Registration and Information Centre, Cambridge); N Cooper, MJ Quinn (National Cancer Registry, Office for National Statistics, London); MP Coleman*, M Quaresma, B Rachet (London School of Hygiene and Tropical Medicine, London); L Shack, EMI Williams (Merseyside and Cheshire Cancer Registry, Liverpool): M Roche (Oxford Cancer Intelligence Unit, Oxford); H Møller (South Thames Cancer Registry, London); J Smith, J Verne (South West Cancer Intelligence Service, Bristol); H Botha, D Meechan (Trent Cancer Registry, Sheffield); G Lawrence (West Midlands Cancer Intelligence Unit, Birmingham); D Forman (Northern and Yorkshire Cancer Registry and Information Service, Leeds); UK - Northern Ireland: A Gavin (Northern Ireland Cancer Registry, Belfast); UK - Scotland: R Black, DH Brewster (Scottish Cancer Intelligence Unit, Edinburgh); UK - Wales: J Steward (Welsh Cancer Intelligence and Surveillance Unit, Cardiff).
Oceania Australia: JM Elwood* (Australian Capital Territory Cancer Registry, Canberra); F Sitas (New South Wales Central Cancer Registry, Sydney); J Condon (Northern Territory Cancer Registry, Casuarina); J Aitken (Queensland Cancer Registry, Brisbane); David Roder (South Australian Cancer Registry, Adelaide); A Venn (Tasmanian Cancer Registry, Hobart); G Giles* (Victorian Cancer Registry, Carlton); T Threlfall (Western Australian Cancer Registry, East Perth).

References
 1. Cooper Y. Cancer. Commons Hansard [Parliamentary Record] 2001372 (20 July 2001)(372):c718W-c719W. Available at: http://www.publications.parliament.uk/pa/cm200102/cmhansrd/vo010720/text/10720w 66.htm#10720w66.html_sbhd1 (accessed April 14, 2008)
 2. Capocaccia R, Gatta G, Roazzi P, Carrani E, Santaquilani M, De Angelis R, Tavilla A, and the EUROCARE Working Group. The EUROCARE-3 database: methodology of data collection, standardisation, quality control and statistical analysis. Ann Oncol 14 (Suppl. 5): v14-v27, 2003.
 3. National Cancer Institute SEER*Stat software [computer program]. Version 6.4.4. Bethesda, MD: National Cancer Institute; 2008.
 4. Sant M, Capocaccia R, Coleman MP, Berrino F, Gatta G, Micheli A, Verdecchia A, Faivre J, Hakulinen T, Coebergh JW, Martinez-Garcia C, Forman D, Zappone A, EUROCARE Working Group. Cancer survival increases in Europe, but international differences remain wide. Eur J Cancer, 37: 1659-1667, 2001.
 5. Gatta G, Capocaccia R, Coleman MP, Gloeckler Ries LA, Hakulinen T, Micheli A, Sant M, Verdecchia A, Berrino F. Toward a comparison of survival in American and European cancer patients. Cancer, 89: 893-900, 2000.
 6. Sant M, EUROCARE Working Group. Overview of EUROCARE-2 results on survival of cancer patients diagnosed 1985-89. In: Survival of cancer patients in Europe: the EUROCARE-2 study. (IARC Scientific Publications No. 151), Berrino F, Capocaccia R, Estève J, Gatta G, Hakulinen T, Micheli A, Sant M, Verdecchia A (Eds), Lyon, International Agency for Research on Cancer, pp 525-541, 1999.
 7. Coleman MP, Gatta G, Verdecchia A, Estève J, Sant M, Storm H, Allemani C, Ciccolallo L, Santaquilani M, Berrino F, EUROCARE Working Group. EUROCARE-3 summary: Cancer survival in Europe at the end of the 20th century. Ann Oncol, 14 (Suppl. 5): v128-v149, 2003.
 8. Coleman MP, Quaresma M, Berrino F, Lutz JM, De Angelis R, Capocaccia R, Baili P, Rachet B, Gatta G, Hakulinen T, Micheli A, Sant M, Weir HK, Elwood JM, Tsukuma H, Koifman S, E Silva GA, Francisci S, Santaquilani M, Verdecchia A, Storm HH, Young JL, CONCORD Working Group: Cancer survival in five continents: a worldwide population-based study (CONCORD). Lancet Oncol, 9:730-756, 2008.
 9. Verdecchia A, Capocaccia R, Santaquilani M, Hakulinen T. Methods of survival data analysis and presentation issues: EUROCARE 1985-89. In: Survival of cancer patients in Europe: the EUROCARE-2 study. (IARC Scientific Publications No. 151), Berrino F, Capocaccia R, Estève J, Gatta G, Hakulinen T, Micheli A, Sant M, Verdecchia A (Eds), Lyon, International Agency for Research on Cancer, pp 41-45, 1999.
10. Hakulinen T. Cancer survival corrected for heterogeneity in patient withdrawal. Biometrics, 38: 933-942, 1982.
11. WHO. Life tables. World Health Organisation 2007 [cited 2005 Jul 15]. Available at http://www.who.int/whosis/database/life_tables/life_tables.cfm (accessed April 14, 2008).
12. Pickle LW, Mungiole M, Jones GK, White AA. Atlas of United States mortality. Hyattsville MD, National Center for Health Statistics, 1996.
13. Coleman MP, Berrino F, Capocaccia R, Gatta G, Hakulinen T, Hatcher J, Lutz JM, Micheli A, Sant M, Verdecchia A, Weir H. The CONCORD protocol (rev. 2001). CONCORD Steering Committee, 15 June 2007. Available at http://www.lshtm. ac.uk/ncdeu/concordinvestigatorsmtg/CONCORD%20Protocol.pdf  (accessed April 14, 2008).
14. Micheli A, Baili P, Quinn M, Mugno E, Capocaccia R, Grosclaude P, The EUROCARE Working Group. Life expectancy and cancer survival in the EUROCARE-3 cancer registry areas. Ann Oncol, 14 (Suppl. 5): 28-40, 2003.
15. Micheli A, Coebergh JW, Mugno E, Massimiliani E, Sant M, Oberaigner W, Holub J, Storm HH, Forman D, Quinn M, Aareleid T, Sankila R, Hakulinen T, Faivre J, Ziegler H, Tryggvadòttir L, Zanetti R, Dalmas M, Visser O, Langmark F, Bielska-Lasota M, Wronkowski Z, Pinheiro PS, Brewster DH, Plesko I, Pompe-Kirn V, Martinez-Garcia C, Barlow L, Möller T, Lutz JM, André M, Steward JA: European health systems and cancer care. Ann Oncol, 14 (Suppl. 5):v41-v61, 2003.
16. National Center for Health Statistics. National vital statistics reports. Centers for Disease Control and Prevention, 2007. Available at: http://www.cdc.gov/nchs/products/ pubs/pubd/lftbls/lftbls.htm (accessed April 14, 2008).
17. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic Research. van Nostrand Reinhold Company, New York, 1982.
18. Elandt-Johnson RC, Johnson NL: Survival models and data analysis (Wiley series in probability and mathematical statistics). John Wiley & Sons, Indianapolis, 1980.
19. Baili P, Micheli A, Montanari A, Capocaccia R: Comparison of four methods for estimating complete life tables from abridged life tables using mortality data supplied to EUROCARE-3. Mathematical Population Studies, 12: 183-198, 2005.
20. Akima H: A method of univariate interpolation that has the accuracy of a third-degree polynomial. ACM Transactions on Mathematical Software, 17: 341-366, 1991.
21. SEER Program. SEER*Stat database: mortality, all causes of death, public-use with county, total US for expanded races (1990-2002). Underlying mortality data provided by NCHS. National Cancer Institute, 2007.
22. Coleman MP, Babb P, Damiecki P, Groscalude P, Honjo S, Jones J, Knerer G, Pitard A, Quinn M, Sloggett A, De Stavola B: Cancer survival trends in England and Wales 1971-1995: deprivation and NHS Region. (Studies on Medical and Population Subjects No. 61). Stationery Office, London, 1999.