Unit 2

Module 1

Quick review of study designs

 Observational studies: observe the participants and no intervention

o Limitation: Confounding
 Risk factors don’t happen in isolation
 Very hard to isolate one risk factor
 Avoid confounding
 During design phase: randomize or match cases
 During analysis phase: use multivariate regression to statistically
adjust for confounders
o You cannot control for all confounders and there is always
residual confounding

o Cross sectional study: Prevalence

 Looks at a sample of the population at a single time point
 Limitation
 Because the measurement its at the same time you don’t know if
the disease came first or the exposure
 Confounding
 No causation
o Case control study
 Use for rare diseases or outbreaks
 Find people who have already developed the disease
 Fin the difference retrospectively of the risk factors in the case and control
group
 Limitation
 Getting appropriate controls is difficult
 Recall bias
 Confounding
 Risk factor may have come after disease
 Cannot calculate disease frequency because selection of cases can’t
o Prospective cohort study
 Measure risk factor and exposure before they develop disease
 Follow up
 Allows identification of causality because you can be sure the exposure
came before disease
 Can calculate risks or rates of developing disease
 Limitations
 Long time
 Confounding
 Loss to follow up
o Retrospective cohort study
 Exposure are measured before the outcome
  Look for databases in which the data on exposure and outcome has already
been collected for some other purposes
 Limitations
 Data quality may be limited
o Nested case control study
 Case control study nested within a prospective cohort study
 You measure one factor only in a small group
 The measurement of the factor is done from samples acquired in the pre
outcome stage where there was only exposure
 You avoid an extremely expensive measurement on all the cases, and only
do it in a reserved group

 Experimental: intervention
o Randomized clinical trial
 You assign the interventions
 Advantages
 Randomization minimizes confounding
 Blinding minimizes bias
 Limitations
 Expensive
 May not be generalizable because of strict eligibility criteria

Unit 2

Module 2

Measures of disease frequency

 3 mayor statistic to measurements disease frequency

o Incidence
 Rate at which people are developing a disease
 Needs follow up
 Only in Cohort and randomize trial

 Example

# Of GI events Person year of Calculation Incidence Rate

follow up
Vioxx 56 2315
Naproxen 121 2316

o Cumulative risk
 Proportion of people who develop a disease in a specified time period
 Less precise
  Requires follow up
 Calculated from cohort or randomized studies
 Used when there is not enough information about the timing of events and
timing of participation

o Prevalence
 Calculated from cross sectional study
 Only look at everybody at one time point
 Proportion or percentage of people who have a disease at a given point in
time – includes old and new cases
 If people live for a long time with disease, then the prevalence will look
much higher

 Person-year of follow up
o Follow a certain number of people for a certain number of time
o You multiply the person times the average time of follow up

o Example

 Total number of people in group: 4047

 Person – years of follow up: 2315
 Average time of follow up: 0.57