Study Design Flashcards
Characteristics Of Descriptive study
- case report, case series , survey
- describe but do not compare
- hypothesis generating
- No inferences
Characteristics of Analytic studies
- experimental and Observational
- Make comparisons btwn groups
- inferences abt E and O associations
- hypothesis testing
Characteristics of Cross - sectional studies(advantages/disadvantages)
lower causal inference because they measure prevalence not incidence and b/c of the inability to refute revers-causation (ie which camefirst- E or O )
- best for permanent exposures
- snapshot- one point in time
- fast
- Inexpensive
- can use all MoA if random sampling
- less bias than C-Cstudies
- recall bias
Steps in Cross Sectional study.
I. Sample without regard to E or O
- Randomly select sample from source pop
- take measurements of each person in study group to determine E and O status
- Classify based on E and O
- Compare prevalence or odds of outcome in E positive and E negative
Steps in a cohort study
I. Clear/ consice objectives
- Define target and source pop
- Define unit of concern (ie indindual or groups)
- Define E and O
- set followup period
Longitudinal Cohort.
You do not know E status of ppl so you select a single group of participants, heterogenous with respect to E of Interest
How do you sample in Cohort?
sample based on E Status or large hetero group with respect to E
How do you ensure validity in Cohorts?
- Ensure representation of target pop (external)
- Ensure comparability with respect to other factors that are not associated with E of interest. (internal) , by:
a) exclusion /restricted sampling : Identify CFV’s and include only 1 level
b) Matching based on CFV
) Analytic Control : through Mantel Haenazel Or regression
3 . Follow up equally - clear diagnostic criteria
Risk-based cohort
- closed pop/ group: all subjects Observed for full study period
-short risk period - measures risk of O in E positive and E negative groups
-Egroups defined at beginning of study period. - Analysis:
I . bi variable risk-based (MOE and M0A)
2. Stratified Analysis
3. Multivariable regression cotten logistic)
Rate Based cohort
- open pop
-rates used to measure disease Freq. - Incidence of disease is expressed relative to the time at risk in each level of exposure :
i. e: sum of amount of time at risk contributed by each memher in each grap (denominator)
→ contribute to time at risk until they develop O change E Status, are lost to follow up , W D or study period ends
Analysis:
1. Rates
2. Surviva l Analysis
Sampling in case control
sample purposively based on O status
Simple Overview : Case control
- select a ft of individuals that have newly the O of interest(cases)
- compare to a group who do not have O of interest(controls)
- frequency of exposure factors are compared between cases + controls
controls need to be:
FROM THE SAME STUDY BASEAS CASES
Not just “healthy ppl” they are Simply ppl without O of interest
they have the same exposure to exposure factors as cases
what is the goal of rate based sampling .
controls Mirror exposure-time at risk of non-cases in source pop
What is the goal of risk based sampling ?
controls represent exposure in source Pop
How to select controls for primary base rate-based sampling
- select from source pop at the end of study period proportional to time at risk. → requires stable pop (Ie exposure does not vary) TAR data known
- selected at fixed time points during study from risk set→ assumes Stable pop TAR data NOT known
- selected from risk set, matched timewise to cases → AKA incidence density sampling select control when case occurs , matched analysis necessary if exposure varies with time
How to select controls from a rate-based, secondary base study:
- select Controls from all non-case ADMISSIONS at end of Study → select controls from norn- case diagnoses that are not associated with exposure
- selected from registry at fixed time points
- selected from regsistry Matched - time wise to case admissions
How to select controls for a risk-based primary base Study
- Random selection of non-cases in source pop → generally occurs atthe end of the risk/study period(However can be selected matched timewise to cases )
How to select controls for risk-based secondary base study
- Random selection of non- cases in the same secondary base registry → Generally occurs atthe end of the risk /study period - but can be selected time wise to cases.
primary Base
Well- defined source pop ( literally or conceptually have a list of all ppl in source pop )
ex federal disease registry.
secondary Base
one or more steps removed from the actual source pop
ex refferals to a clinic
Risk set
Those non-cases eligible to be cases at that point in time
incidence Density sampling
# of controls randomly selected at the time a case arises from the risk set → well suited for when exposure changes with calendar time (Matched analysis needed) → in rate based design people initially identified as controls can become cases → mostcommon, do not need stable pop Or TAR data
