---

WORLD HEALTH ORGANIZATION
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER
IARC
Workng Group Reports
Volume 3
ATTRIBUTABLE CAUSES OF CANCER
IN FRANCE IN THE YEAR 2000


---

Publshed by the Internatonal Agency for Research on Cancer,
150 cours Albert Thomas, 69372 Lyon Cedex 08, France
© Internatonal Agency for Research on Cancer, 2007
Dstrbuted by
WHO Press, World Health Organzaton, 20 Avenue Appa, 1211 Geneva 27, Swtzerland (tel: +41 22 791 3264;
fax: +41 22 791 4857; emal: bookorders@who.nt).
Publcatons of the World Health Organzaton enjoy copyrght protecton n accordance wth the provsons of
Protocol 2 of the Unversal Copyrght Conventon. All rghts reserved.
The desgnatons employed and the presentaton of the materal n ths publcaton do not mply the expresson
of any opnon whatsoever on the part of the Secretarat of the World Health Organzaton concernng the legal
status of any country, terrtory, cty, or area or of ts authortes, or concernng the delmtaton of ts fronters or
boundares.
The menton of specfc companes or of certan manufacturers’ products does not mply that they are endorsed
or recommended by the World Health Organzaton n preference to others of a smlar nature that are not
mentoned. Errors and omssons excepted, the names of propretary products are dstngushed by ntal
captal letters.
The authors alone are responsble for the vews expressed n ths publcaton.
The Internatonal Agency for Research on Cancer welcomes requests for permsson to reproduce or translate
ts publcatons, n part or n ful . Requests for permsson to reproduce or translate IARC publcatons – whether
for sale or for noncommercal dstrbuton – should be addressed to WHO Press, at the above address
(fax: +41 22 791 4806; emal: permssons@who.nt).
IARC Library Cataloguing in Publication Data
Attrbutable causes of cancer n France n the year 2000.

(IARC Workng Group Reports ; 3)
1. Neoplasms–etology 2. Rsk Factors 3. France
I. Internatonal Agency for Research on Cancer. II. Seres
ISBN 978 92 832 2443 4




(NLM Classfcaton: W1)


---

This report is co-authored by:
The International Agency for Research on Cancer (IARC)
P. Auter, P. Boffetta, M. Bonol, P. Boyle (Co-Char), J. Ferlay
The Académie Nationale de Médecine
A. Aurengo, R. Masse, G. de Thé
The Académie des Sciences
R. Moner, M. Tubana (Co-Char), A.J. Val eron
The Fédération Nationale des Centres de Lutte Contre le Cancer (FNCLCC)
C. Hl
In partnership with the Institut National du Cancer (INCa)
L. Borel a, D. Marannch
Institutions consulted:
Insttut de Vel e Santare (InVS)
Reviewers:
France,
For the entire report: J. Benchou (Unversté de Rouen), J. Estève (Hospces Cvls de Lyon).
For specific parts of the report: P. Bougnoux (Insttut Natonal de la Santé et de la Recherche Médcale
[INSERM]) for the sub-secton on nutrton, M. Goldberg (INSERM) for the secton on occupatonal exposure, G.
Orth (Insttut Pasteur) for the secton on nfectous agents, H. Rochefort (INSERM) and C. Sureau (Académe
Natonale de Médecne) for the secton on hormone replacement therapy and oral contraceptves, H. Sancho-
Garner (Unversté de Montpel er) for the secton on ultravolet lght, D. Zmrou-Naver (INSERM) for the sub-
secton on ar pol uton.
International,
For the entire report: J. Peto (London School of Hygene and Tropcal Medcne, London, Unted Kngdom),
J. Sematyck (Unversty of Montreal, Montreal, Canada).
For specific parts of the report: H. zur Hausen (Deutschen Krebsforschungszentrums [DKFZ], Hedelberg,
Germany) for the secton on nfectous agents.
We thank the following persons and institutions for their collaboration in the provision and
interpretation of data: B. Blondel (INSERM), E. Cards (IARC), D. Costaglola (INSERM), S. Francesch (IARC),
S. Gandn (Isttuto Europeo d Oncologa, Mlan, Italy), E. Imbernon (InVS), E. Jougla (CepDc-INSERM), G.
Launoy (Regstre des Cancers du Calvados), H. Lérdon (Insttut Natonal Etudes Démographques [INED]), P.
Mul e (Insttut Jules Bordet, Brussels, Belgum), A. Rannou (Insttut de radoprotecton et de sûreté nucléare -
IRSN), L. Toulemon (INED), F. de Vathare (INSERM), W. Zatonsk (Mare Sklodowska-Cure Memoral Cancer
Centre, Warsaw, Poland)
Correspondence:
Prof. Peter Boyle, Internatonal Agency for Research on Cancer, 150 cours Albert Thomas, 69372 Lyon, France.
Emal: drector@arc.fr
Prof. Maurce Tubana, Académe Natonale de Médecne, 16 rue Bonaparte, 75005 Pars, France.
Emal: maurce.tubana@bomedcale.unv-pars5.fr


---

TABLE OF CONTENTS
Introduction










1
Secton A1
Objectves and methodology






1
Secton A2
Temporal trends n cancer ncdence and mortalty n France


9
Risk factors selected for estimate calculations





29
Secton B1
Tobacco smokng






29
Secton B2
Alcohol drnkng






36
Secton B3
Infectous agents






42
Secton B4
Occupaton







47
Secton B5
Obesty and overweght






60
Secton B6
Physcal nactvty






65
Secton B7
Hormone replacement therapy and oral contraceptves


69
Secton B8
Ultravolet lght






85
Secton B9
Reproductve factors






88
Secton B10
Water, ar, sol and food pol utants




97
Synthesis of results








103
Secton C1
Attrbutable fractons: summary estmates




103
Secton C2
Interactons between cancer rsk factors




111
Risk factors for which no estimates were calculated




120
Secton D1
Ionzng radaton






120
Secton D2
Establshed rsk factors assocated wth cancer occurrence not relevant for France 130
Secton D3
Factors suspected, but not demonstrated, to be assocated wth cancer n humans 132
Discussion









147
Secton E1
Knowledge gaps n causaton of cancers: Progress made and further research needs 147
Secton E2
General dscusson






165
Secton E3
Recommendatons






171
v

---

Introduction
Introduction
Section A1: Objectives
and methodology
1. Background
Many factors, whether genetc, or related to lfestyle
over the meanng of the term “envronment”, whch
or the envronment, have been dentfed over the
s varably nterpreted to encompass qute dfferent
past 50 years as beng assocated wth cancer
types of factor rangng from pol utants to behavours.
occurrence.
Also, ths term (or ts equvalent) s gven dfferent
About 2 to 4% of all cancers seem to have a genetc
meanngs n dfferent languages. In ths report, we use
orgn, .e., gene defects known to be assocated wth
the term “envronment” as meanng “envronmental
these cancers can be transmtted from parents to
pol utants”, an expresson that ncludes pol utants of
ther offsprng. Moreover, genetc polymorphsms
water, ar, sol and food.
and epgenetc phenomena may enhance or reduce
The frst estmate of the relatve mportance of
the rsk assocated wth endogenous or exogenous
genetc and envronmental factors n the global burden
carcnogenc factors. Durng the past two decades,
of cancer was made by Rchard Doll and Rchard
t has been assumed that most cancers are due to
Peto (1981), based on US cancer mortalty data.
lfestyle or to envronmental rsk factors. Very many
Snce then, only a few studes have tred to estmate
epdemologcal studes have been reported, but they
the relatve mportance of cancer rsk factors (see
are often contradctory or of debatable value because
Secton E2, General Dscusson for a revew). In 1981,
of methodologcal problems or lack of suffcent
a number of rsk factors were stll unknown and good
statstcal power. Hence, ther results have to be
qualtatve and quanttatve nformaton on exposure
crtcal y revewed. In paral el, our understandng of
of populatons to rsk factors was rare. Many natons
carcnogeness has markedly progressed, but the
have now entered the era of “nformaton socetes.”
data are stll nsuffcent to ful y establsh the dfferent
In ths respect, n 2007, we have more nformaton on
steps of carcnogeness and the nteracton between
exposure patterns and thus should be able to estmate
the varous endogenous or exogenous factors. In
better the burden of cancer that can be attrbuted to
many felds, further research s clearly requred.
known causes, and to provde an evaluaton of ther
Nevertheless, the strategy of cancer preventon must
relatve mportance.
be based on the latest estmates of the relatve weght
At the begnnng of 2005, the IARC created a
of the varous lfestyle and envronmental rsk factors.
“thnk-tank” on ths topc, wth the am of developng
The am of ths report s to estmate the proportons
methods for frst obtanng estmates of the proportons
of cancer attrbutable to such rsk factors and also
of cancers attrbutable to known causes and second
to evaluate the weght of each factor n the burden
estmatng the number of cancers that could be
of cancer. Ths report dstngushes sold data from
avodable. In July 2005, a workshop at IARC brought
those whch are stll dubous or controversal; the
together cancer epdemologsts who concluded that
former may be consdered and taken nto account
studes on attrbutable causes of cancer should start
n decson-makng n cancer preventon and for
by examnng a few selected countres n the fve
prortzng publc health and research efforts.
contnents.
Dscussons about the roles of lfestyle and of the
In September 2005, the French Académe
envronment n cancer are often hndered by confuson
Natonale de Médecne and the French Académe
1

---

Attributable causes of cancer in France in the year 2000
des Scences proposed to IARC to col aborate on a
avoded or suppressed, a sutable approach conssts
study on attrbutable causes of cancer n France. The
of estmatng the avodable fracton of cancer, that
present report s the product of ths col aboraton.
s the fracton of cancer that would not occur f an
alternatve scenaro of attanable exposure level or
2. Objectives
exposure ntensty were consdered (Murray and
Lopez, 1999).
The purpose of ths report s to provde an assessment
Most estmates of AF n ths report are based
of the number of cancer cases and cancer deaths
on the scenaro of no exposure, as ths does not
n France n the year 2000 attrbutable to factors of
requre assumpton of mnmal levels of exposures
demonstrated carcnogencty or wth a demonstrated
to carcnogens that would represent realstc targets
assocaton wth carcnogenc processes.
for the French populaton. However, “total absence”
Ionzng radaton s a well establshed rsk factor for
s not a realstc alternatve scenaro for several rsk
cancer at many stes. There s farly good knowledge
factors, notably the number of chldren a women has
of the cancer rsk due to exposure to moderate and
(for breast and ovaran cancer). For such factors, t
hgh doses of onzng rradaton. However, the vast
was deemed best to choose an alternatve scenaro
majorty of exposure to onzng radaton n France
that was hstorcal y realstc, .e., exposure levels that
conssts of low and very low doses. The specfc
had exsted n France n the past.
effects of low-dose onzng radaton on cancer rsk
are stll controversal and dffcult to quantfy properly.
4. Incidence data
Therefore, t was decded not to present data on cancer
cases and deaths possbly attrbutable to radaton for
France does not have natonwde cancer regstraton
the whole country. Fol owng the same argument, no
that would al ow the montorng of cancer ncdence
estmate was made for resdental exposure to radon
at the natonal level. There are, however, regstres
decay products. Secton D1 on onzng radaton
operatng n several departments, some of whch focus
addresses ths ssue n more detal.
on specfc cancers. For the year 2000, estmates of
For a number of factors, the evdence of a role
cancer ncdence n France were obtaned from a
n human cancer s suggestve but not demonstrated;
study that estmated the natonwde burden of cancer
these factors are revewed n a separate secton of the
for the perod 1997–2000 (Remontet et al., 2002).
report (Secton D3), but no estmates of attrbutable
Ths report presented estmates of the ncdence of
fracton are provded for them.
cancer at the man stes for the perod 1978–2000,
usng ncdence data from departmental regstres and
3. Methodology
the natonal mortalty data for the perod 1978–1997.
Cancer ncdence n France n 2000 was derved by
Estmaton of attrbutable causes of cancers was
age–cohort model ng of () ncdence from cancer
performed by calculatng the proportons of specfc
regstres, ( ) mortalty n populatons covered by
cancers occurrng n France n 2000 attrbutable to
cancer regstres, and ( ) ncdence-to-mortalty ratos
specfc rsk factors. The proporton of cancers n the
n populatons covered by cancer regstres. Ths
total populaton that can be attrbuted to a rsk factor
model was appled to predcted natonal mortalty for
s cal ed the attributable fraction (AF) (Armtage and
the year 2000 so as to estmate the natonal cancer
Berry, 1987) and s expressed as a percentage.
ncdence n 2000.
For cancer rsk factors that can be avoded or
Some specfc cancer stes were not reported by
completely suppressed, at least n theory, the most
Remontet et al. (2002):
straghtforward way to estmate the attrbutable fracton
s to calculate the fracton of all cases (exposed and
(1) For sinonasal cancer ncdence (ICD 10: C30,
unexposed) that would not have occurred f exposure
C31), we calculated the rato of ncdence of snonasal
had not occurred (Rothman and Greenland, 1998).
to lung cancer n nne cancer regstres that record
For ths approach, the alternatve scenaro to current
snonasal cancers (Parkn et al., 2002: Bas-Rhn,
exposure s the absence of exposure.
Calvados, Doubs, Haut-Rhn, Hérault, Isère, Manche,
For cancer rsk factors that cannot be completely
Somme and Tarn) and appled that rato (0.019 for
2

---

Introduction
men and 0.033 for women) to lung cancer ncdence
5. Mortality data
n France, whch yelded estmates for snonasal
cancer ncdence for France of 453 cases for men and
Mortalty data were provded drectly by the Insttut
151 cases for women. Mortalty data were avalable
Natonal de la Santé et de la Recherche Médcale,
drectly from CepDc data: 99 deaths for men and 42
Centre d’Epdémologe sur les Causes Médcales
deaths for women.
de Décès (INSERM-CepDC) for the year 2000 by
fve-year age groups and by sex for each ICD 10
(2) For the ncdence of pharynx cancer (ICD 10:
code (Internatonal Classfcaton of Dsease, 10th
C09–14), we estmated the proporton of pharynx
revson).
cancer among oral cavty and pharynx cancers (ICD
Ffty-sx per cent of all uterus cancers were coded
10: C00–14) n French regstres (Parkn et al., 2002:
as “uterus not further specfed” (ICD 10 code C55).
Bas-Rhn, Calvados, Doubs, Isère, Somme and
Mortalty data for cancers of the cervx and corpus
Tarn). The proporton of pharynx cancer among oral
uter would be underestmated unless ths “not
cavty and pharynx cancers was 57% for men and
specfed” category s redstrbuted among the two
35% for women. We appled ths proporton to data
stes. Therefore, we estmated for each age group
reported by Remontet et al. (2002) for oral cavty and
the proporton of deaths due to cervx or corpus uter
pharynx combned, and obtaned fgures of 7396
cancer (ICD 10 codes C53 or C54). We appled these
cases of pharynx cancer for men and 833 cases for
proportons to the “not classfed” uterne cancer
women. Mortalty data were avalable drectly from
deaths and real ocated these to ether cervx uter
CepDc data: 2558 deaths for men and 312 deaths
cancer or corpus uter cancer.
for women.
6. Issues in the classification of diseases
(3) For colon cancer (ICD 10: C18), we estmated
and causes of death
the proporton of colon cancer among colorectal
cancers (ICD 10: C18–21) n French regstres
Remontet and co-workers (2002) compled cancer
(Parkn et al., 2002: Bas-Rhn, Calvados, Doubs,
ncdence and mortalty data usng the 9th revson of
Isère, Somme and Tarn). We estmated that colon
the Internatonal Classfcaton of Dsease (ICD 9), and
cancer represents 57% of colorectal cancers for men
estmated cancer ncdence n 2000 usng projectons
and 63% for women. We appled these proportons
of mortalty for 2000. INSERM mortalty data for
to data reported by Remontet et al. (2002) for colon
2000 were classfed usng the 10th revson of the
and rectum combned, and obtaned fgures of 11 132
ICD. Dfferences between the two ICD classfcatons
cases of colon cancer for men and 10 606 cases for
could have affected the mortalty estmates, notably
women. Mortalty data were avalable drectly from
for uterus and prostate cancer, multple myeloma and
CepDc data: 6092 deaths for men and 5719 deaths
leukaema. However, Pavl on and co-workers (2005)
for women.
estmated that dfferences n the two classfcaton
systems dd not nduce dscrepances greater than
(4) For adenocarcinoma of the oesophagus, we
10% n causes of deaths. Therefore, we dd not correct
had recourse to a European study that used data from
the ncdence data for 2000 compled by Remontet
the cancer regstres of Bas-Rhn and Calvados and
and co-workers to match the INSERM mortalty data
reported separately the ncdence of oesophageal
for 2000. Table A1.1 summarzes cancer ncdence
adenocarcnoma
(Botterweck
et
al.,
2000).
and mortalty n France n the year 2000 for males
Proportons of adenocarcnoma were estmated
and females.
as 17.6% of all oesophageal cancers n males, and
34.7% n females. We appled these proportons for
7. Risk factors for cancer in France
ncdence and mortalty data of oesophagus (ICD
10: C15), whch led to estmates of 711 cases for
Rsk factors consdered n ths report were those for
men and 322 for women. The correspondng fgures
whch there s evdence for a causal assocaton wth
for mortalty were 612 deaths for men and 241 for
cancer.
women.
3

---

Attributable causes of cancer in France in the year 2000
The frst type of rsk factor consdered comprses
as aflatoxns (nvolved n lver adenocarcnoma) (see
those agents classfed by the IARC as Group 1
Secton D2).
carcnogens, .e., agents for whch there s sufficient
evidence of carcinogenicity n humans. Exceptonal y,
8. Prevalence of exposures in France
an agent may be placed n ths category when evdence
of carcnogencty n humans s less than sufficient
The burden of cancer observed n the year 2000
but there s sufficient evidence of carcinogenicity
reflects past exposure to rsk factors. Usual y, exposure
n expermental anmals and strong evdence n
to a rsk factor s spread over many years, and cancer
exposed humans that the agent acts through a
may occur long after cessaton of the exposure (e.g.,
relevant mechansm of carcnogencty¹. Snce 1971,
lung cancer n ex-smokers, mesotheloma n retred
the Internatonal Agency for Research on Cancer has
shpbuldng workers). For most cancers and rsk
provded evaluatons of the carcnogenc potental of
factors, the average latency between frst exposure
substances based on epdemologcal and bologcal
and dagnoss s about 15 years. Hence, for evaluatng
evdence. The term “substance” encompasses sngle
the burden of cancer n 2000, we took nto account
physcal, chemcal, or bologcal agents, and mxtures
exposures that occurred n or around 1985.
of physcal chemcal, bologcal and physcal agents,
Data on prevalence of exposure to rsk factors
and also places or crcumstances concentratng stll
n France were assembled by scrutnzng many
unknown carcnogenc agents. Table A1.2 summarzes
dfferent sources, publcatons, reports and relevant
the lst of carcnogenc agents consdered n ths
nformaton publcly avalable on governmental
report.
organzaton web-stes.
The second type of rsk factor ncludes ndvdual
The most representatve exposure data for the
condtons known to be causal y assocated wth
populaton at rsk came from populaton surveys that
cancer occurrence. These factors are not evaluated
evaluated the prevalence of specfc exposures n
n IARC Monographs but some have been evaluated
France, and were conducted usng quota methods
by workng groups convened by the IARC. An IARC
on age, sex and socoeconomc characterstcs (e.g.,
workng group came to the concluson that there was
INSEE surveys). For most exposures, however,
suffcent evdence n humans for a cancer-preventve
prevalence surveys were not avalable for the year
effect of avodance of weght gan (IARC, 2002),
1985, but only for other years. In ths case, we
and thus ths report estmates AFs assocated wth
calculated a lnear nterpolaton of survey results
overweght and obesty. The same IARC workng
that used a smlar method for years before and after
group reported that there was suffcent evdence for
1985, wth weghtng for sample szes and, when
a protectve effect of physcal actvty on the rsk of
relevant, for age and sex dstrbuton. When smlar
breast cancer and colon cancer (IARC, 2002).
surveys before and after 1985 were not avalable,
Reproductve factors (e.g., number of chldren,
we selected the best avalable survey descrbng the
age at frst brth, duraton of breastfeedng) have
stuaton around 1985. When no survey was avalable,
never been evaluated by an IARC workng group.
we used proportons of exposed subjects reported n
However, a large body of evdence supports strong
observatonal studes conducted n France.
assocatons between reproductve factors and breast
Attrbutable fracton s very senstve to
and ovaran cancer (CGHFBC, 2001). We therefore
msclassfcaton of subjects who could have been
ncluded these factors n ths analyss.
exposed (even mnmal y) as unexposed subjects
A number of IARC Group 1 carcnogens
(Wacholder et al., 1994). For nstance, the error n
were not ncluded n ths report, ether because
an estmate of AF due to tobacco smokng s greater
exposure s very rare n France or because they
when occasonal smokers are categorzed as never-
are nsgnfcant. For nstance, parastc nfestaton
smokers than when they are ncluded n the ever-
wth Schistosoma haematobium (nvolved n bladder
smoker category. Therefore, the smplest and most
cancer) and Opisthorchis viverrini (nvolved n lver
robust method for estmatng the attrbutable rsk from
cholangocarcnoma), and ntake of nutrents such
several exposures s based on dvson of subjects nto
¹ http:/ monographs.iarc.fr
4

---

Introduction
two groups, a baselne consstng of those unexposed
for drect calculaton of the AF:
and a group ncludng everyone who was exposed.
9. Calculation of the attributable fraction

(AF)
Ths formula s vald when the rsk of cancer per
The AF can be calculated as a functon of the relatve
unt of exposure was estmated n a model usng log
rsk (RR) of cancer assocated wth exposure to
transformaton. Ths s the case for logstc regresson
a rsk factor and the prevalence of exposure (P) of
or Posson regresson, whch are models wdely used
a populaton to that rsk factor. Ths method was
n case–control and cohort studes respectvely. We
orgnal y descrbed by Levn (1953):
checked that the rsks per unt we used were all based

on models wth a log transformaton of the rsk.


It should be stressed that the dose–effect
relatonshp s n fact rarely lnear (or log-lnear) over
the whole range of exposures, but ths method s
The relatve rsks we used were based on
consdered to be the best approxmaton avalable n
estmates from the most recent meta-analyses or
ths respect.
from best estmates avalable n publshed lterature.
When a rsk factor was reported n the lterature
10. Sensitivity analysis
n multple exposure categores (.e., exposures
classfed n more than two categores), we used
For exposures havng a large mpact on cancer
Levn’s formula adapted by Hanley (2001). Because
burden, n order to check the robustness of AF wth
of the dstrbutve propertes of the AF, mult-level
respect to latency tme between exposure and cancer
exposures could be reduced to a smple dchotomous
occurrence, we took dfferent lag-tmes between frst
stuaton (.e., ever vs. never exposed) or to an average
exposure and cancer dagnoss (10 and 20 years)
exposure of the whole populaton at rsk when the
when prevalence data were avalable for these
relatve rsk was related to an exposure level greater
perods.
or lower than a pre-determned level. These ways of
When for a rsk factor, the alternatve hypothess
groupng or averagng strata of exposure do not affect
was not total absence of exposure, the senstvty
AF estmatons (Hanley, 2001).
analyss was performed takng dfferent alternatve
Data on exposure prevalence were sometmes
exposure scenaros.
avalable only as contnuous varables. For these
A more comprehensve descrpton of ths
contnuous-scale exposures, startng from relatve
senstvty analyss s presented n Secton C2.
rsks estmated for several exposure categores,
we derved the rsk of cancer per unt ncrease n
References
exposure (e.g., the ncrease n rsk of oesophagus
cancer per unt gram per day of alcohol consumpton).
Armtage P, Berry G. Statstcal Methods n Medcal
Assumng a log-lnear relatonshp between exposure
Research, second ed., London, Blackwell Scentfc
and rsk of cancer, we estmated the average rsk for
Publcatons, 1987.
the whole French populaton usng the average level
Botterweck AAM, Schouten LJ, Volovcs A, et al. Trends
of exposure of the whole populaton. Ths was done
n ncdence of adenocarcnoma of the oesophagus and
by applyng the fol owng formula:
gastrc carda n ten European countres. Int J Epdemol
2000;29:645–654.
CGHFBC. Collaboratve Group on Hormonal Factors
n Breast Cancer. Famlal breast cancer: collaboratve

reanalyss of ndvdual data from 52 epdemologcal studes
Because ths log-lnear relatonshp supposes that
ncludng 58 209 women wth breast cancer and 101 986
each ndvdual has experenced a smlar average
women wthout the dsease. Lancet 2001;358:1389–1399.
exposure, we can use the smplfed Levn’s formula
Doll R, Peto R. The causes of cancer: quanttatve
5

---

Document Outline

• Attributable Causes of Cancer in France in the year 2000
• Table of Contents
• Introduction
• Risk factors selected for estimate calculations
• Synthesis of results
• Risk factors for which no estimates were calculated
• Discussion

---