MIME-Version: 1.0
Content-Type: multipart/related; boundary="----=_NextPart_01C63A86.C02BEA50"
This document is a Single File Web Page, also known as a Web Archive file. If you are seeing this message, your browser or editor doesn't support Web Archive files. Please download a browser that supports Web Archive, such as Microsoft Internet Explorer.
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
This presentation contains content that your browser may not be able to =
show
properly. This presentation was optimized for more recent versions of Micro=
soft
Internet Explorer.
Str=
ategies
for data analysis: case-control studies Gilda Piag=
gio
UNDP/UNFPA/WHO/World Bank Special Programme of =
span>Research, Development and
Research Training in
Human Reproduction
World
Health Organization
Geneva,
23 February 2006
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0144.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
First select
Then m=
easure
e=
xposure<=
/div>
Exposed=
13;
Non-exposed=
13;
Cases=
13;
Controls=
13;
Design of case-control studies: retrospective<=
/span>
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0145.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
d
c
Non-exposed
b
a
Exposed
Controls
Cases
Fixed margins
a+c and b+d
Prevalence
of disease is fixed by design
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0146.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
In case-control studieswe can calculate:
•<=
span
style=3D'mso-spacerun:yes'> Prevalence of exposure in cases and in
controls
a/(a+c)and=
b/(b+d)
•The odds ra=
tio to
measure association between disease and exposure:
The odds of being exposed for a case is a/c
The odds of being exposed for a control is b/d
The odds ratio of exposed vs
non-exposed is
OR =3D (a/c)/(b/d) =3D (a x d)=
/(b x c)
We cannot calculate the
relative risk (RR)
Prevalence of exposure in cases and controls and odds of being exposed
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0147.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
•=
If exposure and disease are not associated, OR=3D1
=
<=
span
style=3D'font-family:Tahoma;font-size:92%'>•If exposure=
and
disease are positively associated, OR>1
•If exposure=
and
disease are negatively associated, OR<1
•=
The OR is a good estimation of=
the
RR if the disease is rare
Interpretation of the odds ratio
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0019.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
<=
span
style=3D'font-family:Tahoma;mso-ascii-font-family:Tahoma;mso-hansi-font-fa=
mily:
Tahoma;font-size:83%'>Strategy for data analysis for case-control <=
/span>studies
=
div>
•Describe study profile: number of cases=
and
controls, identified and analyzed
•Baseline characteristics of cases and
controls
•Crude ORs for different categories of u=
se and
risk factors
•ORs for different categories of use and=
risk
factors, adjusting for confounders
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0128.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
Lancet 1985; 326:9=
70-972
Cases: women 20-44 years at ini=
tial
diagnosis of breast cancer, between Dec 1 1980 and Dec 31 1=
982,
resident in 8 regions of the US, identified from
population-based cancer registries.
Controls: women 20-44 years sel=
ected
during same 25 months as the cases were diagnosed, residents of the 8 regi=
ons,
selected randomly by telephone calls to households.
Study
aim was to investigate relation between use of oral contraceptives (OC=
s) by
young women and their risk of breast cancer
Example Oral contracepti=
ves and
breast cancer
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0129.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
Identified:
2573<=
/div>
Included: 2088 (81=
.2%)
Not included: 485<=
/span>
6.4% refused to participate
=
;
3.7% died or were too ill
<=
/span>
8.7% miscellaneous reasons
Oral contraceptives and breast cancer Study profile: c=
ases
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0149.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
Sampled: 5698, =
3;
of which 2469 elig=
ible
Included: 2065 (83=
.6%)
Not included: 404<=
/span>
11.2% refused to participate=
3;
2.2% moved out
3.0% miscellaneous reasons
Oral contraceptives and breast cancer Study profile: c=
ontrols
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0150.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
•Cases(%)Controls(%)
•Age
•20-240.75.1
•25-296.08.2
•30-3418.320.8
•35-3933.528.6
•40-4441.437.3
•
•Family history of breast
cancer
•Yes29.1 =
18.7
•No42.3 =
51.4
•Unknown28.629.9
Oral contraceptives and breast cancer Baseline charact=
eristics
of cases and controls
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0151.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
•Cases(%)Controls(%)
•Age at first term pregnancy
•Nulliparous18.218.4
•<2019.122.1
•20-2223.124.7
•23-2622.321.6
•27-299.27.6
•>297.04.0
•Parous,
•unknown age1.11.6
•
•Benign breast disease surgery
•Yes4.52.3
•No87.291.7
•Unknown8.36.0
Oral contraceptives and breast cancer Baseline charact=
eristics
of cases and controls
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0152.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
403
387
Non-exposed
2065
2088
All
1662
1701
Exposed
Controls
Cases
OR=3D1.07
<=
/span>
Stratification
and confounding variables?
Oral contraceptives and breast cancer Results: =
crude ORs
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0153.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
•Age
at Cases(%)Controls(%=
)
•diagnosis
•or
OR
•selection
N% ever =
N% ever
•users users
•20-2415100.010669.8 <=
/span>-
•25-2912686.5 <=
/span>16987.60.91
•30-3438289.3 <=
/span>42988.11.13
•35-3970086.4 <=
/span>59085.11.11
•40-4486573.0 <=
/span>77172.61.02 <=
/span>
•Total208881.5 <=
/span>206580.51.07
•
•=
Oral contraceptives and breast cancer Results: =
crude ORs
------=_NextPart_01C63A86.C02BEA50
Content-Location: file:///C:/2409C636/Case-control_Piaggio_2006_files/slide0156.htm
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html; charset="windows-1252"
Strategies for data analysis: case-control studies - Gilda Piaggio=
title>
UNDP/UNFPA/WHO/WORLD BANK
HRP<=
/span>
We need to adjust for factors associate=
d with
the risk of breastcancer
or with the likelihood of diagnosis:
<=
span
style=3D'font-family:Tahoma;font-size:92%;color:#00279F;mso-color-index:3'=
>•=
Age at diagnosis or selection
<=
span
style=3D'font-family:Tahoma;font-size:92%;color:#00279F;mso-color-index:3'=
>•=
Family history of breast cancer
<=
span
style=3D'font-family:Tahoma;font-size:92%;color:#00279F;mso-color-index:3'=
>•=
Age at first term pregnancy
=
<=
span
style=3D'font-family:Tahoma;font-size:92%;color:#00279F;mso-color-index:3'=
>•=
History of surgery for benign breast
disease
<=
span
style=3D'font-family:Tahoma;font-size:92%;color:#00279F;mso-color-index:3'=
>•=
Frequency of breast examination