1.4 Measure of Association

Question 1

Draw a 2x2 table for outcome and exposure.

Answer 1

Left hand side (Exposure). Top (Outcome).
Top left square (+ exposure, + outcome) = a.
Top right (+exposure,-outcome) = b.
Bottom left (-exposure,+outcome) = c.
Bottom right (-exposure,-outcome) = d.

Question 2

How to calculate risk ratio?

Answer 2

RR = cumulative incidence among exposed (a/a+b) / cumulative incidence among not exposed (c/c+d)

Question 3

How to calculate rate ratio?

Answer 3

RR = incidence rate among exposed (a/person-time) / incidence rate among not exposed (c/person-time)

Question 4

How to calculate prevalence ratio?

Answer 4

PR = prevalence among exposed / prevalence among not exposed

Question 5

How to calculate odds ratio (for case control studies)?

Answer 5

OR = odds of having exposure among cases / odds of having exposure among controls

Question 6

How to calculate odds ratio (for cohort or cross-sectional studies?

Answer 6

OR = odds of having disease among exposed (a/b) / odds of having disease among not exposed (c/d)
Alternative formula: OR = (axd)/(bxc)

Question 7

Under what circumstances are Risk Ratio, Rate Ratio and Odds Ratio roughly equivalent?

Answer 7

When the disease is rare

Question 8

Describe Attributable Risk

Answer 8

The rate (proportion) of a disease or other outcome in exposed individuals that can be attributed to the exposure. Reflects the amount, usually a percentage, by which the risk of disease is reduced by elimination of exposure.

Question 9

Calculate Attributable Risk

Answer 9

Rate Difference: IR(exposed) - IR(unexposed)
E.g. 285.31-22.52 = 262.79
Compared to males who never smoked, there was an extra 263 cases of lung cancer per 100,000 per son years in male smokers
Ris Difference: CI(exposed) - CI(unexposed)
e.g. 2.05-0.16=1.86% or 18.6 cases per 100.
Compared to males who never smoked, 18.6 of 20.5 cases per 1000 males can be attributed to smoking.

Question 10

Population Attributable Risk

Answer 10

How much of the incidence of disease in the total population can we attribute to the exposure.

Question 11

Calculate Population Attributable Risk

Answer 11

Rate = IR(total population) - IR(unexposed)
e.g. 205.22-22.52-182.70
In the total population, 183 cases of lung cancer per 100,000 person years can be attributed to smoking
Risk = CI(total population) - CI(unexposed)
e.g. 1.47-0.16 = 1.30 or 13 cases per 1000
Of the 14.7 cases of lung cancer per 1000 males, 13 can be attributed to smoking.

Question 12

What is and how do you calculate the Population Attributable Fraction?

Answer 12

What is the proportion of disease that occurred in the population that could be avoided if there was no exposure.
PAF=(PAR/IRtotal)x100
PAF=((IRtotal-IRunexposed)/IRtotal)x100
e.g. PAF=89%.
89% of lung cancer in the male population could be attributed to smoking and (in theory) could have been avoided if no one ever smoked.

Question 13

What measures of association determine ‘how many times more likely’ is a disease given an exposure?

Answer 13

Risk Rate, Odds Ratio, Prevalence Ratio

Question 14

What measures of association determine ‘how much extra’ cases there are of a disease given exposure?

Answer 14

Attributable Risk, Population Attributable Risk

Question 15

What measures of associated determine ‘proportion of cases due to exposure’?

Answer 15

Attributable Fraction, Population Attributable Fraction