2. Measure of Disease Occurrence and Association

Question 1

Aims of Epidemiology

Answer 1

1. Understand etiology of disease
2. Explain local disease patterns
3. Describe natural history of disease
4. Administrative uses

Question 2

Epidemiologic Cases Definition vs. Clinical Diagnosis

Answer 2

Epidemiologic case definitions often depend on clinical diagnoses, however, epidemiology and clinical medicine have different goals

–Clinicians need the accurate diagnosis for treatment and prognosis

–Epidemiologists are focused on causes and a disease definition for one purpose may not be optimal for the other

ETIOLOGY

Question 3

Closed/Open Population

Answer 3

• A closed population adds no new members over time and no loss of members to death [most cohort starts this way]
• An open population may gain new members over time from immigration and birth, or lose members who are still alive through emigration

–Any study population with the loss to follow-up is an open population

Question 4

What is risk and how to measure?

Answer 4

Risk: Probability that disease develops in a person within a specified time period.

1. Incidence proportion (cumulative incidence) IP

–Measure of average risk

2. Incidence density (or rate) ID

–Measure of average rate of disease occurrence over a specified amount of time (person-time)

Question 5

IP Incidence Proportion

Answer 5

Difficult to measure directly because:

1. can only be measured in a closed population
2. loss to follow-up
3. competing risks
4. how do we consider and treat recurrent disease?

Question 6

Techniques for Estimation of Incidence Based on Persons at Risk (Cumulative Incidence)

Answer 6

1. Life Table of the Actuarial Type
2. Kaplan-Meier Method

Question 7

ID Incidence Density

Answer 7

1. Assumption: Time is interchangeable
2. we have a better idea of what we are measuring

Question 8

Comparing IP and ID

Answer 8

IP= 1-e^(-ID∙t)

Assumption:

• The population is closed
• there are no competing risks
• The number of events is small in proportion to the number at risk

Question 9

Prevalence

Answer 9

Prevalence = frequency of existing cases

Point Prevalence = frequency at any one point in time

(cross-sectional study)

Period Prevalence = usually what is meant by prevalence.

Prevalence over some period of time.

Point prevalence = Incidence x Duration (when prevalence small)

Question 10

Odds

Answer 10

•Odds is the ratio of the probability of the event of interest to that of the non-event

For example: Incidence Odds = q/(1-q)

Where q = # of disease events

Question 11

Measures of Association (2 types)

Answer 11

•Absolute difference: public health preventive strategy/activities because want to an absolute decrease in risk of an outcome

e.g. IP exposed - IP non-exposed

•Relative difference: etiologic studies looking for disease causation (usually more informative)

​

Question 12

Measures of effect and association with Counterfactual model of causation

Answer 12

A₁/T₁: exposed

A₀/T₀: the same group of people over the same time period but non-exposed. However, we must rely on a reference group that estimates the rate (A₂/T₂)

Question 13

Measures of Association in Cohort studies

Answer 13

1. Risk Ratio (RR)/Relative Risk: Ratio of two Incidence proportions
2. Rate Ratio (RR): Ratio of two incidence densities
3. Odds Ratio (OR): Ratio of two odds.

Rare disease assumption: if the risk of disease is low, OR approximates RR.

4. Attributable Risk (AR)/Risk difference: estimates the absolute excess risk associated with a given exposure.
5. Population AR: estimates the proportion of the disease risk in the total population associated with exposure

PopAR approximates AR in the exposed when the exposure prevalence is high

Question 14

Interpretation of Attributable risk (also Percent AR)

Answer 14

Assuming a causal relationship, without bias, reducing HPN to normal BP should lower risk in the exposed group from 1.8% to 0.3 %. 1.5% is the absolute excess incidence that would be prevented by eliminating severe HPN.

Percent AR:

If causal, 83.3% of the total risk in the exposed (those with HPN) can be attributed to the exposure (e.g., HPN)

Question 15

percent efficacy

Answer 15

RCT’s, you may wish to test the efficacy of a vaccine:

Risk in Vaccine = 5% (i.e. “unexposed”)

Risk in Placebo = 15% (i.e. “exposed”)

Question 16

Calculation and Interpretation of PopAR (%)

Know:

Prevalence of exposure in population (Pe) = 0.40
Prevalence of non-exposure in population (1- Pe) = (1- 0.40) = 0.60

Risk in exposed individuals (q+)= 0.20

Risk in unexposed individuals (q-) = 0.15

Answer 16

Step 1:

Risk in total population (weighted sum of risks ) =

[(q+) x (Pe)] + [(q-) x (1-Pe)] = (0.20 x 0.40) + (0.15 x 0.60) = 0.17

Step 2:

Population Attributable Risk (PopAR) = Difference between risk in total population and that in the unexposed= 0.17 - 0.15 =0.02

Step 3:

%PopAR = [(PopAR)/(Risk in total population)] x 100% = (0.02/0.17) x 100% = 12%

Interpretation: If the relationship is causal and the effect of the exposure is completely reversible, then exposure cessation would be expected to decrease the total population risk from 0.17 to 0.15 (the risk in unexposed) or 12% of the risk in the total population.

Question 17

Measure of Association in Cross-Sectional Studies

Answer 17

Point Prevalence Ratio (PR) = Prevalence in exposed/ Prevalence in unexposed

Point prevalence = incidence * duration * (1 - prevalence of exposre in population)

Question 18

Can the Point Prevalence Ratio Estimate the Risk Ratio?

Answer 18

–Depends on relationship between incidence and point prevalence

If the PR is used to estimate the RR, then 2 types of bias will differentiate the measures

• (Dur+)/(Dur-) = Duration Ratio Bias: exposure influences disease duration. e.g. exposure will cause people die fast but from prevalence, the exposure looks positive!
• (1-prev.+)/(1-prev.-) = Pt Prev. Complement Ratio Bias: when prevalence is different between exposed and non-exposed

Question 19

Measure of Association in Case-Control Studies

Answer 19

Odds ratio