Lesson 14 Lesson 14 Lesson 14 Lesson 14

Confidence Intervals of Confidence Intervals of

Odds Ratio and Odds Ratio and
Relative Risk Relative Risk Relative Risk Relative Risk
Lesson 14 Outline Lesson 14 Outline Lesson 14 Outline Lesson 14 Outline
Lesson 14 covers Lesson 14 covers Lesson 14 covers Lesson 14 covers
Confidence Interval of an Odds Ratio Confidence Interval of an Odds Ratio
Reviewof Odds Ratio Reviewof Odds Ratio Review of Odds Ratio Review of Odds Ratio
Sampling distribution of OR on natural log scale Sampling distribution of OR on natural log scale
95%Confidence Interval of OR example 95%Confidence Interval of OR example 95% Confidence Interval of OR example 95% Confidence Interval of OR example
Confidence Interval of Relative Risk Confidence Interval of Relative Risk
Reviewof Relative Risk Reviewof Relative Risk Review of Relative Risk Review of Relative Risk
Sampling distribution of RR on natural log scale Sampling distribution of RR on natural log scale
95%Confidence Interval of RR example 95%Confidence Interval of RR example
2 PubH 6414 Lesson 13
95% Confidence Interval of RR example 95% Confidence Interval of RR example
Review: Odds Ratio and Relative Review: Odds Ratio and Relative
Ri k Ri k Risk Risk
Both Odds Ratio and Relative Risk are Both Odds Ratio and Relative Risk are Both Odds Ratio and Relative Risk are Both Odds Ratio and Relative Risk are
measures of association (or relationship) measures of association (or relationship)
between two nominal variables between two nominal variables between two nominal variables. between two nominal variables.
The Odds Ratio is typically estimated from The Odds Ratio is typically estimated from
data collected in a Case data collected in a Case Control study Control study data collected in a Case data collected in a Case--Control study. Control study.
The Relative Risk can be estimated from The Relative Risk can be estimated from
d t ll t d i ti t d d t ll t d i ti t d data collected in a prospective study. data collected in a prospective study.
3 PubH 6414 Lesson 13
Associations between 2 nominal Associations between 2 nominal
i bl i bl variable variable
If there is no association between the two If there is no association between the two
i bl th OR RR 1 i bl th OR RR 1 variables, the OR or RR = 1 variables, the OR or RR = 1
An OR or RR > 1.0 or < 1.0 indicates a An OR or RR > 1.0 or < 1.0 indicates a possible possible
t ti ti l l ti hi ( i ti ) b t t ti ti l l ti hi ( i ti ) b t statistical relationship (or association) between statistical relationship (or association) between
the two variables. the two variables.
Hypothesis tests for the OR and RR are not used Hypothesis tests for the OR and RR are not used Hypothesis tests for the OR and RR are not used Hypothesis tests for the OR and RR are not used
to determine statistical significance of the to determine statistical significance of the
association association association association
Instead, Confidence intervals of OR or RR are Instead, Confidence intervals of OR or RR are
constructed and used to determine whether or not constructed and used to determine whether or not
4 PubH 6414 Lesson 13
the association is statistically significant. the association is statistically significant.
Confidence Interval Confidence Interval
f Odd R ti f Odd R ti of Odds Ratio of Odds Ratio
Review: Odds Ratio Review: Odds Ratio Review: Odds Ratio Review: Odds Ratio
Odds of an event = Odds of an event = Odds of an event = Odds of an event =
probability that the event occurs probability that the event occurs
probability that the event does not occur probability that the event does not occur probability that the event does not occur probability that the event does not occur
Odds Ratio Odds Ratio
odds of event for group 1 odds of event for group 1 g p g p
odds of event for group 2 odds of event for group 2
6 PubH 6414 Lesson 13
Review: Disease OR for Exposed Review: Disease OR for Exposed
d t N t E d d t N t E d compared to Not Exposed compared to Not Exposed
Disease Disease No Disease No Disease Total Total Disease Disease No Disease No Disease Total Total
Exposed Group Exposed Group
aa bb a+b a+b
Not Exposed Not Exposed
Group Group
cc dd c+d c+d
Odds of Disease for Exposed = a/(a+b) b/(a+b) = a/b
Odds of Disease for Not Exposed = c/(c+d) d/(c+d) = c/d p ( ) ( )
OR = a/b c/d = ad / bc
7 PubH 6414 Lesson 13
Confidence Interval for an Confidence Interval for an
Odds Ratio Odds Ratio
The confidence interval for an Odds Ratio The confidence interval for an Odds Ratio The confidence interval for an Odds Ratio The confidence interval for an Odds Ratio
has the same general formula as the has the same general formula as the
Confidence Interval for a population mean Confidence Interval for a population mean p p p p
or population proportion or population proportion
Error Standard * t Coefficien Confidence Estimate Point
The difference is that the confidence The difference is that the confidence
interval for the Odds Ratio is calculated on interval for the Odds Ratio is calculated on
the natural log (LN) scale and then the natural log (LN) scale and then
t d b k t th i i l l t d b k t th i i l l
8 PubH 6414 Lesson 13
converted back to the original scale. converted back to the original scale.
Sampling Distribution of the Sampling Distribution of the
Odds Ratio Odds Ratio
The sampling distribution of the Odds Ratio is The sampling distribution of the Odds Ratio is The sampling distribution of the Odds Ratio is The sampling distribution of the Odds Ratio is
positively skewed. positively skewed.
However It is approximately normally However It is approximately normally However, It is approximately normally However, It is approximately normally
distributed on the natural log scale. distributed on the natural log scale.
Th fid i t l i l l t d th Th fid i t l i l l t d th The confidence interval is calculated on the The confidence interval is calculated on the
natural log scale (LN) natural log scale (LN)
After finding the limits on the LN scale, use the After finding the limits on the LN scale, use the
EXP function to find the limits on the original EXP function to find the limits on the original
9 PubH 6414 Lesson 13
scale. scale.
95% Confidence Interval of OR: 95% Confidence Interval of OR:
hh the steps the steps
1. 1. Calculate the Odds Ratio from the data Calculate the Odds Ratio from the data
2. 2. Find the Natural log of the OR using the LN Find the Natural log of the OR using the LN
function in Excel function in Excel
3. 3. The confidence coefficient is from the standard The confidence coefficient is from the standard
normal distribution: 1.96 for a 95% confidence normal distribution: 1.96 for a 95% confidence
interval interval interval interval
4. 4. Calculate the SE of LN(OR) Calculate the SE of LN(OR) see next slide see next slide
55 The lower and upper limits on the log scale = The lower and upper limits on the log scale = 5. 5. The lower and upper limits on the log scale = The lower and upper limits on the log scale =
LN(OR) LN(OR) 1.96* SE LN(OR) = (LL, UL) 1.96* SE LN(OR) = (LL, UL)
6. 6. Use the EXP function to find the CI limits on Use the EXP function to find the CI limits on
10 PubH 6414 Lesson 13
the original scale: EXP(LL), EXP(UL) the original scale: EXP(LL), EXP(UL)
SE of LN(OR) SE of LN(OR) SE of LN(OR) SE of LN(OR)
The SE of LN(OR) is calculated from the The SE of LN(OR) is calculated from the
cell counts in the 2 X 2 table: cell counts in the 2 X 2 table:
Cases Cases Controls Controls Cases Cases Controls Controls
Exposed Exposed
aa bb
Not exposed Not exposed
cc dd
SE of LN(OR) = SE of LN(OR) =
d
1
c
1
b
1
a
1

11 PubH 6414 Lesson 13
d c b a
95% Confidence Interval of Odds 95% Confidence Interval of Odds
R i R i Ratio Ratio
First calculate the upper and lower limits on First calculate the upper and lower limits on pp pp
the LN scale the LN scale
d c b a
OR
1 1 1 1
* 96 . 1 ) ln(
After calculating the upper and lower limits After calculating the upper and lower limits
d c b a
on the LN scale, use the exponential on the LN scale, use the exponential
function (EXP) to find the lower and upper function (EXP) to find the lower and upper
limits on the original scale limits on the original scale gg
If the 95% Confidence Interval does not If the 95% Confidence Interval does not
contain the value 1.0, the association is contain the value 1.0, the association is
statistically significant at alpha =0 05 statistically significant at alpha =0 05
12 PubH 6414 Lesson 13
statistically significant at alpha = 0.05 statistically significant at alpha = 0.05
Example: Preterm delivery Example: Preterm delivery Example: Preterm delivery Example: Preterm delivery
and Mothers SES level and Mothers SES level
Study background: A case Study background: A case--control study of control study of
th id i l f t d li th id i l f t d li the epidemiology of preterm delivery was the epidemiology of preterm delivery was
undertaken at Yale undertaken at Yale--New Haven Hospital in New Haven Hospital in
Connecticut during 1977 The table on the Connecticut during 1977 The table on the Connecticut during 1977. The table on the Connecticut during 1977. The table on the
following slide contains data for Mothers following slide contains data for Mothers
socioeconomic status (SES) for those with socioeconomic status (SES) for those with socioeconomic status (SES) for those with socioeconomic status (SES) for those with
(cases) and without (controls) preterm (cases) and without (controls) preterm
delivery. delivery.
13 PubH 6414 Lesson 13
yy
Preterm Delivery Preterm Deliveryyy
Socio Socio--
Economic Economic
Status Status
Cases Cases Controls Controls
Upper Upper 11 11 40 40
Upper Middle Upper Middle
Middle Middle
L Middl L Middl
14 14
33 33
59 59
45 45
64 64
91 91 Lower Middle Lower Middle
Lower Lower
Unknown Unknown
59 59
53 53
55
91 91
58 58
55 Unknown Unknown 55 55
Construct a 95% Confidence Interval for the odds ratio
14 PubH 6414 Lesson 13
for Preterm delivery comparing Lower to Upper SES
Data table for Confidence Interval of Data table for Confidence Interval of
Odds Ratio of preterm delivery Odds Ratio of preterm delivery
comparing lower to upper SES comparing lower to upper SES
Preterm Delivery Preterm Delivery
SES SES
Yes: Cases Yes: Cases No: Controls No: Controls
Lower Lower
53 53 58 58
Upper Upper
11 11 40 40
Upper Upper
11 11 40 40
15 PubH 6414 Lesson 13
1. Calculate the OR for Preterm Delivery: 1. Calculate the OR for Preterm Delivery:
L SES d t U SES L SES d t U SES Lower SES compared to Upper SES Lower SES compared to Upper SES
Preterm Delivery Preterm Delivery Preterm Delivery Preterm Delivery
SES SES
Yes: Cases Yes: Cases No: Controls No: Controls
Lower Lower
53 53 58 58
Upper Upper
11 11 40 40
Odds Ratio = (53/58) / (11/40) = 3.32
16 PubH 6414 Lesson 13
2. Calculate the LN (OR) and 2. Calculate the LN (OR) and
3 Fi d h fid ffi i 3 Fi d h fid ffi i 3. Find the confidence coefficient 3. Find the confidence coefficient
2 Calculate the LN(OR) 2 Calculate the LN(OR) 2. Calculate the LN(OR) 2. Calculate the LN(OR)
LN(3.32) = 1.20 LN(3.32) = 1.20
Thi i th i t ti t f th Thi i th i t ti t f th This is the point estimate for the This is the point estimate for the
confidence interval confidence interval
3. The confidence coefficient for the LN(OR) 3. The confidence coefficient for the LN(OR)
is from the standard normal distribution. is from the standard normal distribution.
For a 95% confidence interval, the For a 95% confidence interval, the
coefficient = 1.96 coefficient = 1.96
17 PubH 6414 Lesson 13
4. Calculate the SE of LN(OR) 4. Calculate the SE of LN(OR) 4. Calculate the SE of LN(OR) 4. Calculate the SE of LN(OR)
Preterm Delivery Preterm Delivery Preterm Delivery Preterm Delivery
SES SES
Yes: Cases Yes: Cases No: Controls No: Controls
LL
53 53 58 58
Lower Lower
53 53 58 58
Upper Upper
11 11 40 40
pp pp
1 1 1 1
39 . 0
40
1
11
1
58
1
53
1
ln(OR) SE
18 PubH 6414 Lesson 13
5. & 6. Find the Lower and Upper 5. & 6. Find the Lower and Upper
limits on the LN scale and the limits on the LN scale and the limits on the LN scale and the limits on the LN scale and the
original scale original scale
) 97 . 1 , 44 . 0 ( 39 . 0 * 96 . 1 20 . 1
5. The upper and lower limits of the 5. The upper and lower limits of the
confidence interval on the LN scale are confidence interval on the LN scale are confidence interval on the LN scale are confidence interval on the LN scale are
(0.44, 1.97) (0.44, 1.97)
6. Use the EXP function in Excel to find 6. Use the EXP function in Excel to find
the interval limits of the 95% CI of the the interval limits of the 95% CI of the
19 PubH 6414 Lesson 13
OR: OR: (1.55, 7.14) (1.55, 7.14)
Interpretation of the CI for Preterm Interpretation of the CI for Preterm
Delivery and SES Delivery and SES
State the conclusion in the original scale State the conclusion in the original scale gg
The odds ratio for pre The odds ratio for pre--term delivery for Lower SES term delivery for Lower SES
compared to the Upper SES is 3.32 indicating compared to the Upper SES is 3.32 indicating
increased odds of pre increased odds of pre--term delivery for mothers in term delivery for mothers in
Lower SES Lower SES
The 95% Confidence Interval of the Odds Ratio The 95% Confidence Interval of the Odds Ratio
(1.55 , 7.14) indicates that odds of pre (1.55 , 7.14) indicates that odds of pre--term delivery term delivery
i ifi tl i ifi tl hi h f th L SES hi h f th L SES are are significantly significantly higher for the Lower SES group higher for the Lower SES group
compared to the Upper SES group (at 0.05 compared to the Upper SES group (at 0.05
significance level) because the CI does not contain 1 significance level) because the CI does not contain 1
20 PubH 6414 Lesson 13
significance level) because the CI does not contain 1. significance level) because the CI does not contain 1.
Interpretation of 95% Interpretation of 95% te p etat o o 95% te p etat o o 95%
Confidence Interval of OR Confidence Interval of OR
An Odds Ratio = 1 indicates no association An Odds Ratio = 1 indicates no association
between the exposure and the disease. between the exposure and the disease.
If the 95% confidence interval for the OR does If the 95% confidence interval for the OR does
not contain 1.0 we can conclude that there is a not contain 1.0 we can conclude that there is a
statisticall significant* association bet een the statisticall significant* association bet een the statistically significant* association between the statistically significant* association between the
exposure and the disease. exposure and the disease.
* at the 0 05 significance level * at the 0 05 significance level at the 0.05 significance level at the 0.05 significance level
If the 95% confidence interval for the OR If the 95% confidence interval for the OR
contains 1.0, the association is contains 1.0, the association is not not significant at significant at
21 PubH 6414 Lesson 13
gg
the 0.05 level. the 0.05 level.
Confidence Interval Confidence Interval
f R l ti Ri k f R l ti Ri k of Relative Risk of Relative Risk
Review: Relative Risk Review: Relative Risk Review: Relative Risk Review: Relative Risk
Risk of an event = Risk of an event = Risk of an event = Risk of an event =
probability that the event occurs probability that the event occurs
Relative Risk Relative Risk
probability that event occurs for group 1 probability that event occurs for group 1
probability that event occurs for group 2 probability that event occurs for group 2 probability that event occurs for group 2 probability that event occurs for group 2
23 PubH 6414 Lesson 13
Review: RR Calculated from a Review: RR Calculated from a
table table
Disease Disease No Disease No Disease Total Total Disease Disease No Disease No Disease Total Total
Exposed Group Exposed Group
aa bb a+b a+b
Not Exposed Not Exposed
Group Group
cc dd c+d c+d
Risk of Disease for Exposed = a/(a+b)
Risk of Disease for Not exposed =c/(c+d) Risk of Disease for Not exposed = c/(c+d)
RR =a/(a+b) c/(c+d)
24 PubH 6414 Lesson 13
RR a/(a+b) c/(c+d)
Confidence Interval for Confidence Interval for
Relative Risk Relative Risk
Like the Odds Ratio, the sampling Like the Odds Ratio, the sampling
distribution of the Relative Risk is distribution of the Relative Risk is
positively skewed but is approximately positively skewed but is approximately
normally distributed on the natural log normally distributed on the natural log
scale scale scale. scale.
Constructing a Confidence Interval for the Constructing a Confidence Interval for the
Relative Risk is similar to constructing a CI Relative Risk is similar to constructing a CI Relative Risk is similar to constructing a CI Relative Risk is similar to constructing a CI
for the Odds Ratio except that there is a for the Odds Ratio except that there is a
different formula for the SE different formula for the SE
25 PubH 6414 Lesson 13
different formula for the SE. different formula for the SE.
95% Confidence Interval of RR 95% Confidence Interval of RR 95% Confidence Interval of RR 95% Confidence Interval of RR
1. Estimate the RR from the data 1. Estimate the RR from the data
2. Find the natural log of RR: LN(RR) 2. Find the natural log of RR: LN(RR)
3. The confidence coefficient is from the standard 3. The confidence coefficient is from the standard
normal distribution: 1.96 for a 95% confidence normal distribution: 1.96 for a 95% confidence
interval interval
4 C l l t th SE f LN(RR) 4 C l l t th SE f LN(RR) t lid t lid 4. Calculate the SE of LN(RR) 4. Calculate the SE of LN(RR) see next slide see next slide
5. Calculate the lower and upper limits on the LN 5. Calculate the lower and upper limits on the LN
scale: LN(RR) scale: LN(RR) 1 96* SE LN(RR) 1 96* SE LN(RR) scale: LN(RR) scale: LN(RR) 1.96 SE LN(RR) 1.96 SE LN(RR)
6. Use the EXP function to find the limits on the 6. Use the EXP function to find the limits on the
original scale: EXP(LL), EXP (UL) original scale: EXP(LL), EXP (UL)
26 PubH 6414 Lesson 13
g ( ), ( ) g ( ), ( )
SE of LN(RR) SE of LN(RR) SE of LN(RR) SE of LN(RR)
The SE of LN(RR) is calculated from the cell The SE of LN(RR) is calculated from the cell
t i th 2 X 2 t bl t i th 2 X 2 t bl counts in the 2 X 2 table: counts in the 2 X 2 table:
Disease Disease No Disease No Disease Total Total
Exposed Group Exposed Group
aa bb a+b a+b
Not Exposed Not Exposed
cc dd c+d c+d
Group Group
cc dd c d c d
d b
d) c(c
d
b) a(a
b
ln(RR) SE

27 PubH 6414 Lesson 13

This formula is algebraically equivalent to the formula in text on pg. 201
95% Confidence Interval of 95% Confidence Interval of
R l i Ri k R l i Ri k Relative Risk Relative Risk
First calculate the upper and lower limits on the First calculate the upper and lower limits on the pp pp
LN scale LN scale
* 96 1 ) l (
d b
RR
) ( ) (
* 96 . 1 ) ln(
d c c b a a
RR

After calculating the upper and lower limits on After calculating the upper and lower limits on
the LN scale, use the exponential function to find the LN scale, use the exponential function to find
the limits on the original scale the limits on the original scale the limits on the original scale the limits on the original scale
If the 95% Confidence Interval does not contain If the 95% Confidence Interval does not contain
the value 1.0, the association is statistically the value 1.0, the association is statistically
significant at alpha =0 05 significant at alpha =0 05
28 PubH 6414 Lesson 13
significant at alpha = 0.05 significant at alpha = 0.05
Example: 95% CI for RR of Example: 95% CI for RR of
Myocardial Infarction (MI) Myocardial Infarction (MI)
Background: Physicians enrolled in the Physicians Health Background: Physicians enrolled in the Physicians Health g y y g y y
Study were randomly assigned to take a daily aspirin or Study were randomly assigned to take a daily aspirin or
placebo. The table provides the number with MI in each placebo. The table provides the number with MI in each
group. group. g p g p
Group Yes No
Myocardi al
Infarcti on (MI)
Group Yes No
Aspirin 139 10898 11037
Placebo 239 10795 11034
378 21693 22071
Calculate the RR and 95% CI for the RR for MI (aspirin Calculate the RR and 95% CI for the RR for MI (aspirin
378 21693 22071
29 PubH 6414 Lesson 13
( p ( p
group compared to placebo group) group compared to placebo group)
1. 1. Calculate the RR for MI: Calculate the RR for MI:
aspirin group compared to aspirin group compared to
placebo placebo placebo placebo
Myocardi al
Infarcti on (MI)
Group Yes No
Aspirin 139 10898 11037
Placebo 239 10795 11034
Infarcti on (MI)
Placebo 239 10795 11034
378 21693 22071
RR = (139/11037) / (239/11034) = 0.581
30 PubH 6414 Lesson 13
2. Calculate the LN (RR) and 2. Calculate the LN (RR) and
3 Fi d h fid ffi i 3 Fi d h fid ffi i 3. Find the confidence coefficient 3. Find the confidence coefficient
2 Calculate the LN(RR) 2 Calculate the LN(RR) 2. Calculate the LN(RR) 2. Calculate the LN(RR)
LN(0.581) = LN(0.581) = --0.543 0.543
Thi i th i t ti t f th Thi i th i t ti t f th This is the point estimate for the This is the point estimate for the
confidence interval confidence interval
3. The confidence coefficient for the LN(RR) 3. The confidence coefficient for the LN(RR)
is from the standard normal distribution. is from the standard normal distribution.
For a 95% confidence interval, the For a 95% confidence interval, the
coefficient = 1.96 coefficient = 1.96
31 PubH 6414 Lesson 13
4. Calculate the SE LN(RR) 4. Calculate the SE LN(RR) 4. Calculate the SE LN(RR) 4. Calculate the SE LN(RR)
Group Yes No
Myocardi al
Infarcti on (MI)
Group Yes No
Aspirin 139 10898 11037
Placebo 239 10795 11034
378 21693 22071 378 21693 22071
10795 10898
106 . 0
11034 * 239
10795
11037 * 139
10898
ln(RR) SE
32 PubH 6414 Lesson 13
5. & 6. Find the Lower and 5. & 6. Find the Lower and
Upper limits on the LN scale and Upper limits on the LN scale and
the original scale the original scale the original scale the original scale
) 336 . 0 , 75 . 0 ( 106 . 0 * 96 . 1 543 . 0
5. The upper and lower limits of the 5. The upper and lower limits of the
confidence interval on the LN scale are confidence interval on the LN scale are
) (
confidence interval on the LN scale are confidence interval on the LN scale are
((--0.750, 0.750, --0.336) 0.336)
6. Use the EXP function in Excel to find 6. Use the EXP function in Excel to find
the interval limits of the 95% CI of the the interval limits of the 95% CI of the
33 PubH 6414 Lesson 13
RR: RR: (0.472, 0.715) (0.472, 0.715)
Interpretation of 95% CI of RR Interpretation of 95% CI of RR
for MI for MI
The Relative Risk estimate = 0.58 which The Relative Risk estimate = 0.58 which
indicates that physicians in the aspirin indicates that physicians in the aspirin
h d l i k f MI th h d l i k f MI th group had a lower risk of MI than group had a lower risk of MI than
physicians in the placebo group. physicians in the placebo group.
Th 95% fid i t l i di t th t Th 95% fid i t l i di t th t The 95% confidence interval indicates that The 95% confidence interval indicates that
the decreased risk related to daily aspirin the decreased risk related to daily aspirin
use is use is significant significant (at alpha =0 05) since (at alpha =0 05) since use is use is significant significant (at alpha = 0.05) since (at alpha = 0.05) since
the interval (0.472, 0.715) does not the interval (0.472, 0.715) does not
contain 1. contain 1.
34 PubH 6414 Lesson 13
contain 1. contain 1.
General Interpretation of 95% General Interpretation of 95% Ge e a te p etat o o 95% Ge e a te p etat o o 95%
Confidence Interval of RR Confidence Interval of RR
A Relative Risk = 1.0 indicates no association A Relative Risk = 1.0 indicates no association
between the exposure and the disease. between the exposure and the disease.
If the 95% confidence interval for the RR does If the 95% confidence interval for the RR does
not contain 1.0 we can conclude that there is a not contain 1.0 we can conclude that there is a
statisticall significant* association bet een the statisticall significant* association bet een the statistically significant* association between the statistically significant* association between the
exposure and the disease. exposure and the disease.
* at the 0 05 significance level * at the 0 05 significance level at the 0.05 significance level at the 0.05 significance level
If the 95% confidence interval for the RR If the 95% confidence interval for the RR
contains 1.0, the association is contains 1.0, the association is not not significant at significant at
35 PubH 6414 Lesson 13
gg
the 0.05 level. the 0.05 level.
Notes on the OR and RR CI: Notes on the OR and RR CI: Notes on the OR and RR CI: Notes on the OR and RR CI:
The RR estimate will not be exactly in the middle The RR estimate will not be exactly in the middle yy
of the confidence interval on the original scale of the confidence interval on the original scale
The LN(RR) is exactly in the middle of the interval on The LN(RR) is exactly in the middle of the interval on
the LN scale the LN scale the LN scale the LN scale
The OR estimate will not be exactly in the middle The OR estimate will not be exactly in the middle
of the confidence interval on the original scale of the confidence interval on the original scale of the confidence interval on the original scale of the confidence interval on the original scale
The LN(OR) is exactly in the middle of the interval on The LN(OR) is exactly in the middle of the interval on
the LN scale the LN scale
Ch k th t th RR OR ti t i t i d Ch k th t th RR OR ti t i t i d Check that the RR or OR estimate is contained Check that the RR or OR estimate is contained
in the final interval in the final interval its easy to forget the EXP its easy to forget the EXP
step step
36 PubH 6414 Lesson 13
step step
Readings and Assignments Readings and Assignments Readings and Assignments Readings and Assignments
Reading: Chapter 8 pgs 200 Reading: Chapter 8 pgs 200--201 201 Reading: Chapter 8 pgs 200 Reading: Chapter 8 pgs 200 201 201
Work through the Lesson 14 practice exercises Work through the Lesson 14 practice exercises
Excel Module 14 works through the examples in Excel Module 14 works through the examples in Excel Module 14 works through the examples in Excel Module 14 works through the examples in
this lesson for the confidence intervals of OR and this lesson for the confidence intervals of OR and
RR RR
Excel functions: LN and EXP Excel functions: LN and EXP
There is no CI function for the RR and OR There is no CI function for the RR and OR you just you just
work through the steps work through the steps
Complete Homework 10 Complete Homework 10
37 PubH 6414 Lesson 13