Overview of Epidemiological Study Design Flashcards
Definition of epidemiology
The study of the distribution and determinants of health-related states or events in specified populations and the application of this study to control health problems. The basic science underlying much of public health and preventive medicine.
What are the 2 opportunities to account for bias?
Design phase, Analysis phase
What are the 3 methods to account for confounding in the design phase?
Randomization, Restriction, Matching
What are the 4 methods to account for confounding in the analysis phase?
Standardization, Stratification, Multivariable adjustment, Causal inference analysis
What are the 4 basic types of study design?
Cross-sectional (prevalence) study, Case-control study (Retrospective, Prospective [(nested in a prospective cohort)], Cohort study (Mostly prospective), Randomized controlled trial (Prospective)
How are incidence and prevalence measured in prospective vs. retrospective studies?
Prospective design is ideal for estimating incidence(# of newly diagnosed cases in population) of disease since new cases can only be clearly identified prospectively. Prospective and retrospective can be used to estimate prevalence (proportion of individuals in population with the disease) or association of outcome with a specific candidate of predictors.
Name 3/6 advantages of cross-sectional studies
Relatively easy, quick and inexpensive
Often good first step for new study issue
Good design for hypothesis generation
Can estimate overall and specific disease prevalence and sometimes rates
Can estimate exposure proportions in the population
Can study multiple exposures or multiple outcomes or diseases
Name 3/5 disadvantages of cross-sectional studies
Impractical for rare exposure or diseases
Not useful for establishing temporal relation
Confounding is difficult to control
Reverse causation
What is the most used type of study design in population studies?
Case-control study
How can power be increased in matching in case-control studies?
Using more than one control per case (eg 4:1)
Name 3/4 advantages of case-control studies
Quick, Efficient
Relatively inexpensive
Best design for rare diseases
Can study multiple exposures at the same time
Name 3/5 disadvantages of case-control studies
Only assess one disease outcome
Inefficient if exposure is rare
High potential for bias (selection and recall) and confounding–Difficult to choose appropriate controls (selection bias?)–Accurate exposure information? Recall bias?
Temporal sequence problems: reverse causation
Can’t calculate incidence, population relative risk or attributable risk
What is a nested case-control study?
A case-control study situated within a prospective cohort study.
Whenever a case occurs, a control can be selected from among the cohort members who do not have the disease at that point in time (the risk set).
Risk set sampling of controls allows estimation of the incidence rate ratio.
What are 3 advantages of nested case-control studies?
Remove recall bias because data collected before development of disease.
Allows for the time-frame element to be included in the case-control.
Often more cost-effective than a cohort.
What is a case-cohort study?
The selection of a random sample of noncases from the parent cohort and all incident cases.
No matching is performed.
What modification needs to be performed on the hazard ratio for a case-cohort study?
Cox proportional hazards regression models need to be weighted.
What are 2 advantages of case-cohort studies?
More efficient for several disease outcomes using the same randomly chosen subcohort.
Reduces selection bias and information bias
What measure of risk can be measured in a case-cohort study?
Subcohort can be used to calculate person-time risk and disease risk (risk ratio) can be measured.
What is the order of events in a prospective cohort study?
Exposure/Study beginning –> disease occurrence
What is the order of events in a retrospective cohort study?
Exposure–>Disease occurrence–>Study begins
What is the best observational design to establish a temporal relation between exposure and outcome of interest?
Prospective cohort study
What are 3/6 advantages of cohort studies?
Estimate of risk and rate of disease occurrence over time
Studying rare exposures
Multiple outcomes of a single exposure
Establish a temporal relationship between exposure and outcome and natural history of disease
Lower potential for bias than a case-control study –avoid recall and selection bias
Results are considered more conclusive than results from case-control studies
The longer a cohort study continues, the stronger it can become
What are 3/7 disadvantages of cohort studies?
Very large sample sizes, especially for rare outcomes
Expensive and time-consuming
Larger sample size than case-control
Data collection is usually very expensive
Long time commitment for follow-up
Potential threats to internal validity
–Selection bias
–Attrition problem (Loss to follow-up)
–Differences in the quality of measurement of exposure or disease (information bias)
Can not infer causal relation
What Does the Optimal Cohort Study Look Like? Name 4/8 elements
Large in scale, diverse (>500K) Address multiple diseases/risk factors Highly efficient recruitment, data collection, sample processing Linked personal electronic records High sample and data quality State-of-the-art technology Cost effective Data available for qualified researchers
Name Hill’s causal criteria (5)
Biological plausibility Temporality Specificity Consistency Strength –Quantitative strength –Biologic gradient (dose-response)
What are 2 pros of intervention studies?
True control of exposure (experiment)→prevent confounding
If randomized, no selection bias
What are 3/7 disadvantages of intervention studies?
Feasibility Ethical consideration \$\$$ Time Adherence with regimen Volunteer “bias” Only one exposure
