Incidence and Prevalence Quiz 1

Question 1

Ecological study

Answer 1

Aggregate measures: association between exposure and outcome at the group level

• good for hypothesis generation
• Ecological fallacy (aggregation bias): inappropriately drawing conclusions from population level to individual level

Question 2

Cross sectional study

Answer 2

Unit is the individual Can determine associations between exposures and diseases at individual level

• may be used to assess prevalence, prevalence ratio -not good for estimating risk
• incidence-prevalence bias when exposure affects survival - temporal order of exposure and outcome cannot be established

Question 3

Case control study

Answer 3

• individuals selected according to disease status
• cases and controls derived from same population
• study question: what is the frequency of exposure among cases vs controls
• measure of association= odds ratio
• may suffer from selection bias and recall bias

Question 4

Cohort study

Answer 4

Goal to calculate risk or rate of specific disease or event

• usual goal to calculate association between a specific exposure and risk or rate of a specific disease
• bias loss to follow up

Question 5

Calculation of prevalence

Answer 5

Prevalence = # of existing cases of disease/total population at risk at a given point in time/Total population at risk ata given point in time

Question 6

Point prevalence

Answer 6

• numerator = # of cases at one point in time (cross-sectional)
• denominator = population at time of assessment

Question 7

Period prevalence

Answer 7

• numerator = # all existing cases and new cases that occur during a given time period
• denominator = average reference population over the specified time period
• good when difficult to know when a disease began (chronic illness etc)
• lifetime prevalence = the proportion of a population that at some point in their life have experienced the condition

Question 8

Study Design Flow Chart

Answer 8

Question 9

Factors affecting prevalence

Answer 9

1. faster recovery time -
2. increased incidence ^
3. better reporting ^
4. in-migration of noncases -
5. in-migration of cases ^
6. improved cure rate -
7. new treatment extends survival time ^
8. out-migrartion of cases -
9. decreased mortality of cases ^
10. increased mortality among cases -
11. decreased incidence -

Question 10

Odds

Answer 10

Odds are defined as the ratio of the probability to the ‘non event’ (all other outcomes)

• probability of an event = p
• probability of a non-event = 1-p
• Odds = p/ (1-p)

Question 11

Why use odds?

Answer 11

• Odds rartio can approximate risk under cirtain circumstances
• Logistic regression has favorable statistical properties, and produces odds ratios

Question 12

What is incidence?

Answer 12

The proportion or rate of NEW events or cases that develop in a population of individuals at risk during a specific time interval

• implies follow-up over time (cohort studies)

Question 13

Cumulative incidence

Answer 13

• The probability of developing a disease (event) ina population at risk, during a specified period of time (also called RISK)

Question 14

Incidence rate/density

Answer 14

The rate of developing a disease (event) in a population at risk per unit time.​

• ​Rates have person time in denominator

Question 15

Special types of incidence

Answer 15

• Mortality rates (usually per year)
• Case fatality rate (proportion not rate)
• Attack rate (proportion, not a rate)\

Question 16

Incidence rate

Answer 16

*calculated using aggregate data

incidence rate= # of new cases/avg population*timeoverwhich cases occur

Assumptions:

• the risk of an event is constant over the observed time interval.
  
  • ​ie the risk of n persons followed for t time is the same as t persons followed for n time​
• independence between censoring and events
• does not require a fixed closed cohort
• population and demographics relatively stable throughout the time period (lack of secular trends

Question 17

Incidence density

Answer 17

ID= # of events /person time

Assumptions:

• the risk of an event is constant over the observed time interval.
  
  • ​ie the risk of n persons followed for t time is the same as t persons followed for n time​
• independence between censoring and events
• does not require a fixed closed cohort
• population and demographics relatively stable throughout the time period (lack of secular trends

Question 18

Incidence density example

Answer 18

n=10

events = 6 = numerator

person months = 115

Person years = 9.58

ID= 6/(115/12) = 6/9.58py = .63 events per person year * 100 = 63 events per 100 person years

Question 19

When would you use calendar time instead of follow up time?

Answer 19

• Interim analyses of long term studies with dynamic cohorts
• If an exposure or outcome changed greatly over time (secular trend)
  
  • includes birth cohort and period effects

Question 20

Which method do I use?

Answer 20

• In general use the method that atkes imnto account all the info available to you
  
  • if you have individula level data about each participants time on study, don’t calculate rate using avg population at risk. Instead, calculate incidence density–it will be more accurate
  • Exceptions when comparing studies and more crude measures are used

Question 21

Cumulative incidence (CI)

Answer 21

• most intuitive way to estimate risk
• Always a proportion 0-1
• Is unitless, but must specify a time period
  
  • ​example: 60yo risk of heart disease 2% (over what time period?)

Question 22

What is the most appropriate study design to calculate cumulative incidence?

Answer 22

Cohort study

Question 23

Cumulative Incidence cont.

Answer 23

• When follow up is complete (no censoring, pop steady throughout study), CI = # of events (over a specified time period)/ initial population
• example: 1,000 men aged 69-89 initially free of prostate cancer are followed for 10 years. Over this time, 45 men develop prostate cancer. What is thew CI?
  
  • ​CI = 45/1000 over 10 years = .0.045 over 10 years = 4.5% over 10yrs

Question 24

Life table method

Answer 24

When we know people are censored but can assume they are censored uniformly through each follow-up time period:

• CI= # number of new cases or events/number of subjects at start of period minus 1/2 # of censored

Question 25

Life table vs Kaplan-Meier Method

Answer 25

Life table method calculates **conditional probability** of an event within each interval * intervals are usually, though not necessarily, of equal length * Must assume ceonsoring occurs at a uniform rate within each time interval What id you cannot assume uniform censoring? * Use KM method * Involves calculating conditional probability of each event at the time it occurs--requires data on specific event times and censoring times * Denominator is the study population at risk at the time of each event * allows you to calculate the cumulative probability of survival, and the compliment of survival is risk (CI = 1 - cumulative probability of survival)

Question 26

Calculating cumulative incidence using KM method

Answer 26

1. when an event occurs, eneter the number at risk at the time of event and the number of events at that time 2. Divide the # of events by # at risk to obtain the conditional probability of the event over the time interval 3. Subtract 1-(conditional probability of event) to obtain conditional probability of survival over that interval 4. multiply the probability of surviving that time interval by the probability of surviving the previous interval to obtain the cumulative probability of survival 5. Subtract 1-( cumulative p of survival) to obtain cumulative probability of the event over the entire duratrion of the study

Question 27

Survival estimates examples

Answer 27

Question 28

Attack rate

Answer 28

* Attack rate is defined as the prportion who become ill after a specified exposure. Often used in outbreak situation * ​Example: An outbreak of gastroenteritis with 50 cases in 2 months among a popualtion at risk of 2,500. Attack rate = 50/2,500 = 0.02 over 2 moths (is this a rate?)

Question 29

Case fatality rate

Answer 29

Case fatality rate is defined as the prportion (percentage) of cases that result in death. Used to measure disease severity. * CFR = # of deaths/# of cases (\*100) = % case deaths

Question 30

Cumulative incidence vs cumulative density

Answer 30

* Numerator = # of individuals who develop outcome during observation period; same for both CI and CD * Denominator: * ​for cumulative incidence, the denominator is the total # of people at risk being followed * for incidence density, the denominator is the total amount of person time at risk for the outcome * CI is a useful measure of the risk of a disease occurring in a particular population, over a particular time period * Incidence density is an estimate of the rate at which the disease develops. * Incidence Densities are good for studies with dynamic populations and in studies with fixed populations that have relatively long follow-up time

Question 31

Incidence summary

Answer 31

* Important the denominator is accurate * ​Must exclude prevalent cases at baseline if no longer at risk (could lead to underestimation of incidence) * A common mistake is to fail to specify the units of time after calculating incidence, especially for cumulative incidence

Question 32

Relationship between incidence and prevalence

Answer 32

prevalence/1-prevalence = incidence\*duration Example: incidence rate = 45.9/100,000 Avg duration of illness = 0.5 years prevalence = 45.9/100,000 \* 0.05 = .00023 or .023% \*rare disease: 1-prevalence is close to 1

Question 33

Which study design produces prevalence data?

Answer 33

* Cross-sectional * Cohort (baseline data)

Question 34

Which study design produces incidence data?

Answer 34

* Cohort, clinical trials -\> longitudinal follow-up required