Week 15 : Experimental Designs for Testing Hypothesis

• Type of Study - Alternative Name - Unit of Study

  Obeservational Studies

  • Descriptive Studies
  • Analytical Studies
  • Ecological - Correlational - Population
  • Cross - Sectional - Prevalence - Individuals
  • Case - Control - Case - Reference - Individuals
  • Cohort - Follow up - Individuals
• Type of Study - Alternative Name - Unit of Study

  Experimental Studies - Intervention Studies

  • Randomized Controlled Trials - Clinical Trials - Individuals
  • Cluster Randomized Controlled Trials - None - Groups
  • Field Trials
  • Community Trials - Community Intervention Studies - Healthy People Communities
• Clinical Trials - used to test therapeutic interventions in ill persons (patients)

  Field Trials - used to test preventive interventions in well persons in the community
• Clinical Trial (CT) vs. Field Trial (FT)

  Setting

  • CT - healty facilities
  • FT - outside clinical settings

  Participants

  • CT - patients in hospitals or outpatient clinics
  • FT - participants are living at home

  Inclusion and Exclusion Criteria

  • CT - more stringent inclusion and exclusion criteria
  • FT - less stringent inclusin and exclusion criteria

  External Validity

  • CT - more problems with external validity, limited generalizability
  • FT - fewer problems with external validity, less limited generalizability

  Randomization

  • CT - by individuals
  • FT - by groups
• Experimental Study Designs

  • good for hypothesis setting
  • the investigator has more control on assigning participants (treatment and control group)
  • examines if the relationship is causal
  • subjects followed over time
  • assign, add, and remove variables to determine the effect of one variable to another
  • best design for confounding variables (inclusion and exclusion criteria)
  • suitable for testing treatment efficacy (before and after exposure can be seen)
  • major advanatge : strength of causal inference
• Characteristics of Experimental Study Design

  • analytic (can be performed in the extracted data)
  • longitudinal (exposure development of the disease)
  • manipulation of independent variable
  • randomization (random allocation)
  • comparison group - treatment and control group
• Experimental Design Population and Variables

  • Reference Population - group of interest
  • Experimental Population - group that is studied
  • Efficacy - extent to which specific factor produces a beneficial result under ideal conditions (can it work)
  • Effectiveness - if it can be applied in the field practice (real world), (doest it work)
• Types of Experimental Study Designs

  • According to Type of Exposure
  • According to Unit of Analysis
  • According to Type of Comparison
• According to Type of Exposure

  • Therepeutic - people have the disease

  Intervention and Prophylactic

  • Intervention - product in addition to ongoing treatment plan
  • Prophylactic - efficacy of the drug to prevent disease

  According to Unit of Analysis

  • Clinical Trials
  • Community Trials

  According to Type of Comparison

  • Between Group Designs - two or more groups receiving different interventions
  • Within Group Designs - single group before and after intervention
• Randomized Blinded Trial - gold standard in medical research (stringent)
• Steps in Randomized Blinded Trial

  Select sample from the population (assembling cohort)

  • Inclusion Criteria - set of qualities that an individual shall process to be able to qualify to join
  • Exclusion Criteria - disqualifies entry to study

  Measuring Baseline Variables

  • characterize study cohort
  • for dichotomous outcome, presence or absence of disease
  • for continous outcome, effect of an exposure varies continously from time to time
• Steps in Randomized Blinded Trial

  Applying Interventions

  • compares outcome in groups that received different interventions
  • blinding - prevents participants from knowing certain information that may influence and taint results
  • blinding prevents unintended interventions
  • choice of experimental treatment: particular drug/dose or health education procedure
  • choice of comparison group: they should be clear of any active treatment
  • placebo - pill or substance that is given to the control group
• Steps in Randomized Blinded Trial

  Follow-up cohorts (assuring compliance)

  • degree of credibility is measured on the number of individuals who will be able to finish the course
  • effect of intervention is reduced to degree that subjects fail to comply

  Measure of Outcome Variables

  • Continous Outcome Variable - t test
  • Dichotomous Outcome Variable - chi square
• Special Types of Randomized Blinded Designs

  • Run-in Design - useful design for increasing the proportion
  • Factorial Design - answer two seperate research questions in a single sample of subjects
  • Randomized Matched Pairs - pairing participants based on some factors like age and sex
  • Pre-randomizations - randomizatuon before obtaining informed consent
  • Group Randomization - randomize naturally occuring cluster of subjects
• Randomized Controlled Clinical Trials

  • patients are enrolled in a study and randomly assigned to one of the following group
  • intervention/treatment group - receives the experimental treatment
  • control group - receives the non experimental treatment
• Randomized Controlled Clinical Trials

  • Single Blind Study - patients are unaware on which treatment they will receive
  • Double Blind Study - both participants and investigators are blinded
  • Triple Blind Study/Allocation Concealment - investigators, participants, and the one who delivers the interventions are blinded
• Randomized Controlled Field Trials

  • intervention is usually preventive
  • conducted in the community
  • appropriate participants are randomly allocated to receive the preventive measure
  • participants followed over time to determine the rate of disease
• Experimental Study Design

  Advantage

  • strongest evidence for cause and effect
  • only possible design for some research questions
  • produce faster and cheaper answers

  Disadvantage

  • costly in time and money
  • many questions are not suitable for experimental designs
  • standardized interventions different from common practuce
  • restricts scope and narrows study design