Observational Studies

Overview

• Observational studies are studies in which subjects are observed for the effect of exposures on outcomes without intervention by researchers
• Often done when randomization is not possible for logistical or ethical reasons

Descriptive Studies 

• Case report
  • detailed description of a patient’s symptoms, signs, treatment, and disease course
  • conducted for hypothesis generation or to raise awareness of emerging diseases, rare pathologies or unusual presentations 
  • not useful for hypothesis testing
• Case series
  • a group of case reports of patients with a similar exposure, treatment, or disease 
  • no comparison group
• not useful for hypothesis testing

Prospective and Retrospective Studies

• Prospective studies
  • patients enrolled before data collection
  • participants followed over time 
  • data collected on exposures and outcomes as they arise 
• Retrospective studies 
  • involve collecting historical information from a sample of patients 
• enrollment occurs after exposure and outcome have  already occurred

Case-Control Study

• Retrospective (almost always) 
• Compares patients with and without disease to asses their odds of a certain exposure   
• Odds ratio (OR) is measure of disease association     
  • OR = odds of exposure among cases/odds of exposure in non-cases = ad/bc
    • odds of exposure among cases = # cases with exposure/# cases without exposure = a/c
    • odds of exposure among non-cases = # non-cases with exposure/# non-cases without exposure = b/d
• “How much more likely is it that patients with cirrhosis have been exposed to heavy alcohol use compared to controls?”

Cohort Study

• Can be either prospective or retrospective 
• Compares groups with and without an exposure to assess associations with subsequent disease  
• Relative risk (RR) is measure of disease association 
  • RR = incidence rate in exposed group/incidence rate in unexposed group = [a/(a+b)]/[c/(c+d)]
    • incidence rate in exposed group = # of exposed cases/(# total exposed cases and non-cases) = a/(a+b)
    • incidence rate in unexposed group = # of unexposed cases/(# total unexposed cases and non-cases) = c/(c+d)
• “How much more likely are patients to develop cirrhosis if they are exposed to heavy alcohol use?”

Cross-Sectional Study

• Exposure and outcome are assessed simultaneously  
• Ascertains association not causality
  • best used for gathering information and generating further hypotheses 
• Prevalence is measure of disease association 
• “How much higher is the proportion of cirrhosis patients with heavy alcohol use compared to those without cirrhosis?”

Ecological Study

• Exposure and outcome assessed at the population level 
  • population defined geographically or temporally 
• Prevalence and incidence are measures of disease association
• Often used for rapid hypothesis generation 
• “How much higher is the proportion of cirrhosis patients in a country where average alcohol use is high compared to a country in which alcohol is illegal and of low use?”
• Disadvantage is the potential for “ecological fallacy”
  • making spurious risk associations between populations which may not represent true risk association at the individual level
• e.g. assuming that people who live in a country with high alcohol use rates must be more likely to get cirrhosis by virtue of living in that country

Sibling Studies

• Measures relative influence of environment vs genetics on trait variation
• Twin concordance study
  • compares how often monozygotic and dizygotic twins both have the same trait or disease
• Adoption study
  • 2 designs
    • examines differences and similarities between child and adoptive vs birth parents
    • examines differences and similarities between siblings raised by adoptive vs birth parents