Measures of Association

Session five
Measure of association in
epidemiological studies
BY:
Sisay S. (BSc, MPH in Epidemiology)

July, 2022
7/11/2022 Sisay S (Bsc., MPH in Epidemiology) 1

• Learning objectives
• List common measures of association and measures
of public health impact
• Calculate and interpret risk ratio and odd ratio and

describe their use.
• Calculate and interpret attributable risk and risk

percent and describe their use.
7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 2

• Association
• Exposure Outcome
• Is there a relationship between the exposure and

the outcome of interest?
• What kind?
• How strong?

• Definition
• Association: is a statistical relationship between two
or more variables.
• Requiring comparing two group.

– Exposed Vs unexposed
– With outcome Vs without outcome
• It quantifies the relationship b/n exposure and
outcome variables

• Exposure
• It is an explanatory factor; any potential health
determinants; the independent variable
• Example
– Ingestion of contaminated foods
– Sharing needles, drinking alcohol
– Genetic factors
– Getting education about certain behavioural change

• Outcome
• Disease:
– Infectious disease like malaria on non-infectious disease
like DM, cancer
• Events:
– Car accidents, immigration, poverty etc…
• Death:

• Why do we need them?
• Move from descriptive to analytical epidemiology.
• Comparisons within and between populations.
• Risk factor identification.

• Epidemiological data often presented in the form of two
by two (contingency) table.
• A 2x2 table contains 4 cells = a, b, c and d

Disease
Yes No Total
Yes a b a+b
No c d c+d
Total a+c b+d a+b+c+d

Cont…
• a= Number of individuals who are exposed and have
the disease
• b= Number of individuals who are exposed but not

have the disease
• c=number of individual who are not exposed but have

the disease
• d=number of individuals who are not exposed and do

not have disease
• Cont…
• a+b = the total number of individuals exposed
• c+d = the total number of individuals who are not exposed
• a+c = the total number of individuals with disease
• b+d = the total number of individuals without disease
• a+b+c+d = total sample size of the study

• Strength of association is assessed by:
• Ratio Measures: relative risk
• Difference Measures: excess risk
• These measures may be applied at:

• Individual Level
• Population Level

• Ratio Measures includes:
• Relative risk
• Odd ratio (depending on the type of study design)
• Relative risk (risk ratio) (RR) for:

• Cohort study and experimental study
• Odd ratio (OR) for:

• Case control, cross sectional studies

• Relative risk/risk ratio (RR):
• It compare the occurrence of disease or risk factor
among two group using ratio.
• It is often measured / calculated from data presented in a
two by two table.
• It indicate the likelihood of developing disease in
exposed relative to non exposed.
• Compare the new occurrence (incidence) of disease
among exposed and non exposed.
• Indicate the strength of association b/n exposure and
outcome.
• Two-by-Two Table (Contingency Table)
• Relative risk = Incidence of disease among exposed

Incidence of disease among non exposed
• RR= Risk in exposed (a/(a+b))

Risk in unexposed(c/(c+d))
7/11/2022 14
Sisay S. (Bsc. MPH in Epidemiology)
• Cont…
• It indicates the likelihood of developing the disease

in the exposed group relative to those who are not
exposed.
• It is direct measure of risk.
• It usually used in cohort and experimental study

design

• Interpretation of RR
• The value of RR ranges between 0 and infinity.
• RR always a positive number.
• RR = 1
• No association b/n the risk factor (independent
variable) and the outcome (dependent variable).
• That means the risk acquiring the disease is the same

for those subjects with risk factor and without the
7/11/2022
risk factor Sisay S. (Bsc. MPH in Epidemiology) 16
• RR =/= 1,
• There is an association
• RR > 1 indicates that the risk of acquiring the disease is

greater among subject with risk factor than among
without the risk factor.
• RR < 1 indicate the risk of acquiring the disease is less

among with risk factor than among subject without the
risk factor
• Suggests that the factor exposure may be protective
7/11/2022
– i.e exposure associated with decreased risk
Sisay S. (Bsc. MPH in Epidemiology) 17
• RR come nearer to 1 (nullifying point) then the
association is said to be weak association.
• If RR is 3.0 or more association is high
• 1.5 to 2.9 association is moderate
No association
0 1
Preventive Risk

• Example 1:
• Table shows data from a cohort study of oral contraceptive
(OC) use and MI among women aged 16-49 years.
MI
Yes No Total
Yes 27 455 482
No 77 1831 1908
Total 104 2286 2390
• Calculate RR?
• Give interpretation for the result?

7/11/2022 19
• Solution
• RR= Ie/Io
• Ie = 27/482 = 0.056
• Io = 77/1908 = 0.040
• RR = 0.056/0.040 = 1.4
• Interpretation: Women who used oral contraceptive

had 1.4 times higher risk of developing MI when
compared from non oral contraceptive users.

• Example 2:
• A study conducted on postmenopausal hormone use
and coronary heart disease, after a total of 54, 308.7
person years follow up for a hormone users, 30
developed CHD and 60 develop CHD from 51, 477.5
person years follow upamong non hormone users.
• Calculate RR?
• Give interpretation?

• Solution
• RR = Ie/Io
• RR = IDe/Ido
• IDe = No. new cases among exposed/ Person years among exposed
group
= 30/54, 308.7 = 0.0006
• IDo = No. new cases among non exposed/ Person years among non
exposed group
= 60/51,477.5 = 0.0012
• RR = 0.0006/0.0012 = 0.5
• Interpretation: women who use postmenopausal hormone had 0.5
times less risk of developing CHD when compared from non users.

• Odd Ratio (OR)
• It is an indirect measures of a risk in disease of rare
occurrence
• It is usually used in case control and cross sectional study

design
• Cases and controls are predetermined and we are calculating to

determine whether cases or controls are more exposed to
postulated risk factors
• Indicates the likelihoods of having an exposure among cases

relative to controls
• Odd Ratio…
• Odds are calculated by dividing the probability of the
event by the probability that it will not occur.
• The probability of an event in a study population is

P(X=x) and, the probability that the event will not
occur is (1- P(X=x)).
• The odds of the event occurring in the study

population can, therefore, be expressed in the form
P(X=x)/ (1-P(X=x))
• Odd Ratio…

• Odd Ratio…

• Odd Ratio…
• Hence: OR = (a/c)/(b/d)
• Which can be simplified to

– OR = ad/bc

• Interpretation of OR
• The value of OR ranges between 0 and infinity.
• OR always a positive number.
• OR = 1
• No association b/n the risk factor and (independent
variable) and the outcome (dependent variable).
• That means the odds of acquiring the disease is the

same for those subjects with risk factor and without
7/11/2022
the risk factor Sisay S. (Bsc. MPH in Epidemiology) 28
• OR =/= 1,
• There is an association
• OR > 1 indicates that the odds of acquiring the disease
is greater among subject with risk factor than among
without the risk factor.
• OR < 1 indicate the odds of acquiring the disease is
less among with risk factor than among subject without
the risk factor
• Suggests that that the factor exposure may be
protective
7/11/2022
– i.e exposureSisay
associated with decreased risk
S. (Bsc. MPH in Epidemiology) 29
• OR come nearer to 1 (nullifying point) then the
association is said to be weak association.
• If OR is 3.0 or more association is high
• 1.5 to 2.9 association is moderate
• The further away from 1 the stronger the
association between exposure and disease.
No association
Preventive Risk
0 1
• Example 1: table shows data from case control study of
oral contraceptive (OC) use and mycocardial infraction
in pre-meneopausal female nurses.
Myocardial infarction
Current Yes No Total
OC
Yes 23 304 327
Use
No 133 2816 2949
Total 156 3120 3276
• Calculate OR?
• Solution:
• OR = ad/bc
= (23)*(2816)/ (133)*(304)
= 1.6
• Interpretation: the odds of developing MI among

oral contraceptive users is 1.6 times higher compared
from non-oral contraceptive users.

• Example 2: Researchers select 100 patients with colon cancer
and 300 patients without colon cancer. Of the 100 patients
with colon cancer, 20 smoke cigars while 30 of the 300
patients without colon cancer smoke cigars.
Colon cancer
Yes No Total
Smokers (exposed ) 20 30 50
Non smokers (non exposed) 80 270 350
Total 100 300 400
• Calculate OR?
• Solution:
• OR = ad/bc
= (20)*(270)/ (80)*(30)
= 2.25
• Interpretation: the odds of developing colon cancer

among cigarette smokers is 2.25 times higher
compared from non cigarette smokers.

• RR can be estimated by OR if the following
condition are fulfilled:
• The controls are representative of the general
population
• The selected cases are representative of all cases
• The disease is rare

• Difference Measures: excess risk
• Attributable risk (AR)/risk difference (RD)

• It measures absolute effect of the exposure or the excess risk of
disease in those exposed compared with those who are not exposed.
• The amount of disease that can be attributed to a certain exposure
• Quantifies the expected reduction in disease occurrence if harmful

exposure could be eliminated
• AR= incidence among exposed (Ie) – Incidence among non

exposed(Io)
• Measures public health impact of disease attributed by the certain

exposure in the exposed group
• AR…

• AR…
• It important for
• Policy decision
• Funding decision (E.g. prevention program)

• Example:
• Table shows data from a cohort study of oral contraceptive (OC)
use and MI among women aged 16-49 years.
MI
Yes No Total
Yes 27 455 482
No 77 1831 1908
Total 104 2286 2390
• Calculate AR for the following table

• Solution
• AR = Ie – Io
= 27/482 – 77/1908
= 0.0156 = 156/10, 000
• Interpretation:
• Among 10, 000 oral contraceptive users 156 cases are
attributed due to use of oral contraceptive
• We can prevent 156 cases among oral contraceptive
user by elimination oral contraceptive use
• Attributable risk percent (AR%)
• Estimate the proportion of the disease among the exposed
that is attributable to the exposure, or
• The proportion of the disease in the exposed group that could

be prevented by eliminating the exposure.
AR% =(Ie-Io) x 100
Ie
• From above example calculate AR%?
7/11/2022 41
• Population attributable risk (PAR):
• Measures the excess risk among the general population
that is due to exposure of interest
• PAR = Risk in population -Risk in unexposed
• PAR = AR x Proportion of exposed in the population
• Measures public health impact of an exposure
• Estimate the excess rate of disease in the total study

population that is attributable to the exposure.

• Example: research was conducted to assess the
association between cigarette smoking and death
from lung cancer. The following finding were
obtained.
• AR = 89 per 100,000 per year
• Prevalence rate of cigarette smoking = 20%
• Calculate PAR?
• Both AR and PAR are used to estimate the effect of
disease incidence of eliminating a given risk factor,
• AR estimate reduction in disease incidence only in

those exposed,
• PAR estimates reduction in disease incidence in the

population as a whole.

• Population attributable risk percent (PAR%)
• Estimate the proportion of disease in the study population that is
attributable to the exposure and thus could be eliminated if the
exposure were eliminated.
PAR% = PAR x 100
incidence rate in total population
• Example: PAR = 17.8 per 100,000 per year
• Mortality rate in the total population 24.8 per 100.000
• Calculate PAR%
• Give interpretation ?
• Exercise
– From 2000 – 2005 a research group enrolled 2000 pregnant
women in an epidemiological study to investigate the effect of
anemia in delivery outcome. Among the 650 women who had
anemia, 140 experience a preterm delivery or low birth weight of
their baby. Among the 1350 women who had no anemia, 40
experienced a preterm delivery or low birth weight of their babies.
Prevalence of anemia among the population is 15%.
• Name the exposure and outcome variable.
• What kind of study is appropriate?
• Calculate the appropriate measures of association and give interpretation.
• Calculate the AR and give interpretation.
• Calculate the PAR and give interpretation.
• Possible out came in studying the relationship between
disease and exposure
• No association between exposure and disease
– AR = 0, RR = 1
• Positive association between exposure and disease(more

exposure, more disease).
– AR>0, RR>1
• Negative association between exposure and disease(more

exposure, less disease).
– AR<0,(Negative), RR<1(fraction)
• Absolute Risk Reduction (ARR)
• Absolute difference in the rates of events between the
two groups
• It gives an indication of the baseline risk and

treatment effect
• ARR = 0 means there is no difference between the

two groups therefore the treatment had no effect.
48
• Cont…
• ARR = risk of the outcome in the control group - risk of
the outcome in the treatment group.
• This is also known as the absolute risk difference.

• Example: A researcher include outcomes of cancer after two
years of treatment by the new drug, which is 15% (0.15) of the
control group died and 10% (0.10) of the treatment group
died.
– RR = 0.10/0.15 = 0.67
• Interpretation: Since the RR < 1, the treatment decreases the

risk of death. 49
• Cont…
– ARR = 0.15 - 0.10 = 0.05 or 5%
• The absolute benefit of treatment is a 5% reduction in the

death rate.
50
• Relative Risk Reduction (RRR)
• Is the complement of the RR
• It is protective efficacy of the treatment
• Most commonly reported measure of treatment

effects
• It tells us the reduction in the rate of the outcome in

the treatment group relative to that in the control
group
51
• Cont…
• RRR= absolute risk reduction / risk of the outcome
in the control group. An alternative way to calculate
the RRR subtract the RR from 1 (RRR = 1 - RR)
– From above example the RRR = 0.05/0.15 = 0.33 or 33%
– RRR = 1 - 0.67 = 0.33 or 33%
• The treatment reduced the risk of death by 33%

relative to that occurring in the control group.
52
• Number Needed to Treat (NNT)
• It represents the number of patients we need to treat
with the experimental therapy in order to prevent 1 bad
outcome
• It incorporates the duration of treatment.
• Clinical significance can be determined to some extent by
looking at the NNTs, but also by weighing the NNTs
against any harms or adverse effects of therapy.
53
• Cont…
• NNT = inverse of the ARR and is calculated as 1 /
ARR
• From above example NNT = 1/ 0.05 = 20
• We need to treat 20 people for 2 years in order to

prevent 1 death.
54
• Precise of estimate on treatment effect
• Looking at the confidence intervals (CI) for each estimate
• If the confidence interval is fairly narrow the estimate is a

precise reflection of the population value.
• It also provides information about the statistical

significance of the result.
• If the value corresponding to no effect falls outside the

95% confidence interval then the result is statistically
significant at the 0.05 level.
55

Measures of Association

Uploaded by

Copyright:

Available Formats

Measures of Association

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Measures of Association

Uploaded by

Copyright:

Available Formats

Session five

Sisay S. (BSc, MPH in Epidemiology)

7/11/2022 Sisay S (Bsc., MPH in Epidemiology) 1

• Calculate and interpret risk ratio and odd ratio and

• Calculate and interpret attributable risk and risk

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 2

• Is there a relationship between the exposure and

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 3

• Requiring comparing two group.

– With outcome Vs without outcome

• It quantifies the relationship b/n exposure and

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 4

– Sharing needles, drinking alcohol

– Getting education about certain behavioural change

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 5

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 6

• Comparisons within and between populations.

• Risk factor identification.

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 7

• A 2x2 table contains 4 cells = a, b, c and d

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 8

• b= Number of individuals who are exposed but not

• c=number of individual who are not exposed but have

• d=number of individuals who are not exposed and do

• c+d = the total number of individuals who are not exposed

• a+c = the total number of individuals with disease

• b+d = the total number of individuals without disease

• a+b+c+d = total sample size of the study

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 10

• Difference Measures: excess risk

• These measures may be applied at:

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 11

• Odd ratio (depending on the type of study design)

• Relative risk (risk ratio) (RR) for:

• Odd ratio (OR) for:

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 12

• Relative risk = Incidence of disease among exposed

• RR= Risk in exposed (a/(a+b))

• It indicates the likelihood of developing the disease

• It is direct measure of risk.

• It usually used in cohort and experimental study

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 15

• RR always a positive number.

• That means the risk acquiring the disease is the same

• RR > 1 indicates that the risk of acquiring the disease is

• RR < 1 indicate the risk of acquiring the disease is less

• Suggests that the factor exposure may be protective

• If RR is 3.0 or more association is high

• 1.5 to 2.9 association is moderate

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 18

• Give interpretation for the result?

• Interpretation: Women who used oral contraceptive

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 20

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 21

7/11/2022 Sisay S. (Bsc. MPH in Epidemiology) 22

• It is usually used in case control and cross sectional study

• Cases and controls are predetermined and we are calculating to

• Indicates the likelihoods of having an exposure among cases

• The probability of an event in a study population is

• The odds of the event occurring in the study