Retrospective and Observational Study Designs Flashcards
Clinical studies
Discovery an effect
Test a hypothesis
Based on statistics (probability)
Introduction
Research questions Construct a working hypothesis Design experiments Observe and record results Interpret data
Study Design is directly linked to:
Objectives of the study
Measurements to be taken
Interpretation of results
Application of results
Common mistakes
Inefficient or incorrect designs
Too may or too few pts
Use incorrect procedures
Plan for too many response measurement
Observational Retrospective
Case reports/studies Case-control Cohort Cross-sectional - Studies that begin in present and follow the subjects into the PAST
Observational Prospective
Cohort
Cross-sectional
- Studies that begin in present and follow the subjects into the FUTURE
Utility of Observational Studies
Health issues in large groups of people Interventional studies are unethical Extremely rare events or long lag-time between exposure and outcome NO intervention Compares outcomes/experiences
Retrospective study
Hypothesis generating
Exploratory
Pilot in anticipation of a prospective study
- Clarify the study question and hypothesis
- Inform the design of the prospective study (size, etc)
Define case study/series
A description of one or a series of interesting cases
Observation –> hypothesis
Case study Pros
Readily available information
Fast
Case Study Cons
Depedent on availability and accuracy of chart records
Subject to selection bias
Uncontrolled
Least robust
Define bias
A type of systematic error
“Any tendency which prevents unprejudiced consideration of a question”
Case Control
Begin with outcome and look back for features of those with and without outcome
- Hypothesize cause of disease and ID RK
Case Control Pros
Fast and few
Lots of new information
Only feasible option
Case Control Cons
Highly susceptible to bias
Data collected for other reasons
Separate sampling of cases and controls
Difficult to assess incidence or prevalence
Define Cohort Study
Start with a population at risk
ID cohorts of interest - those with RF and without
Cohort Study Utility and Limits
Assess incidence and ID potential cause/predictors Subject to bias Losses to followup Expensive and long Not useful in rare events
Cross-Sectional Study
Measurements are made at one time point and no followup
Describe an outcome of interest and its distribution in the population of interest
PREVALENCE estimate
SNAPSHOT in time
Define Serial Surveys
Series of cross-sectional studies performed at different time points
Subjects in each group are different (births, deaths, migration)
Serial Surveys Pros
Fast
Inexpensive
Part of a cohort
Serial Surveys Cons
Cannot establish cause-effect relationships
Not practical for rare events
