Experimental design

• In epidemiology, researchers are interested in measuring or assessing the relationship of exposure with a disease or an outcome
• Observational Study Design


  Allows nature to take its course. The investigator measures the outcome but does not intervene. It can be descriptive or analytical.
• Experimental Study Design


  Involves an active attempt to change a disease determinant like exposure or a behavior, progress of a disease through treatment.
• Types of epidemiological studies


  • Observational
  • Experimental
• One type of experimental study is the randomized controlled trial or RCT

  
• Types of epidemiological studies, their alternative names and unit of study


  • Cohort studies
  • Case-control studies
  • Cross-sectional studies
  • Randomized controlled trials

  
• Experimental Studies


  • Are cohort studies in which the investigator manipulates the predictor variables (the intervention) and observes the effect on an outcome
  • Examine the relationship between the predictor and outcome variables in a cohort of subjects followed over time
• Experimental Study Research Design


  A procedure for devising an experimental setting where in the researcher administers or withholds a factor to see how the dependent variable is affected the independent variable

  
• Variables involved in an experimental study design


  • Independent variable (predictor variable)
  • Dependent variable (outcome variable)
  • Extraneous or confounding variable

  
• Experimental study design is best if the investigator wants to control confounding variables
• Experimental study design is recommended and suitable for testing the efficacy of treatment programs
• Experimental Study Design


  • Analytic- statistical analysis can be performed in the extracted data
  • Longitudinal- there is monitoring of the predictor value (exposure) up to the development of disease (outcome)
  • Manipulation of the IV by the researcher
  • Randomization (random allocation)
  • Use of comparison group

  
• Reference Population


  Group of interest
• Experimental Population


  Group actually studied
• Efficacy


  Extent to which a specific factor produces a beneficial result under ideal conditions
• Effectiveness


  Extent to which a specific factor produces a beneficial result when applied in the field practice
• Types of Experimental Study Design (according to type of exposure)


  • Therapeutic
  • Intervention
  • Prophylactic

  
• Types of Experimental Study Design (according to unit of analysis)


  • Clinical Trials
  • Community Trials

  
• Types of Experimental Study Design (according to type of comparison)


  • Between Groups Designs
  • Within-Group Designs

  
• Randomized Controlled Trial


  One of the best examples of experimental study design

  
• Steps in Randomized Blinded Trial


  1. Select sample from the population (assembling cohort)

  2. Measuring Baseline Variables

  3. Randomization of Study Subjects
• Inclusion Criteria


  Refers to the set of qualities that an individual shall possess to be able to qualify to join the research
• Exclusion Criteria


  The set of characteristics that disqualifies an individual to join the study
• Dichotomous outcome


  The outcome of interest is not present on the onset of the experiment and will be checked eventually for the progression of the disease. And one exposure is to one disease.
• Continuous outcome


  The effect of an exposure varies continuously from time to time, the differences in the changes between the measurements shall be computed and be compared from one group to another
• Randomization


  Allows equal probability among the sample population to be included in a certain group in the study for example if they will be assigned as an experimental or control group
• Continuously from time to time, the differences in the changes between the measurements shall be computed and be compared from one group to another

  1. Compute differences in changes

  2. Compare differences between groups
• Study effects of antihypertensive drugs


  • Use the difference between two groups in the degree of change in blood pressure over the course of the study
• Randomization


  • Cornerstone or identity of randomized blinded trial
  • Allows equal probability among the sample population to be included in a certain group in the study (e.g. experimental or control group)
• Randomization


  1. Lottery or fishbowl method

  2. Number generating system
• Applying Interventions


  • Compares the outcome in groups of subjects that receive different interventions
  • Blinding is recommended to prevent the participants from knowing certain information that may influence them
• Blinding


  • Eliminates the influence of confounding variables present at the time of randomization
  • Does not protect the study from confounding by variables that develop during the period of follow-up
  • Prevents unintended interventions (Post-randomization confounding variables)
• Experimental Treatment


  Picking a particular drug or dose or health education procedure from among several promising options when there is no clear basis for knowing which is best
• Comparison Group


  Choosing the appropriate comparison or control group, it should be clear of any active treatment in a way that can be blinded which usually means by identical placebo
• Placebo


  A pill or a substance that is given to the control group that has no physical effect on the individual
• Follow-up Cohorts (Assuring Compliance)


  1. Individuals enrolled in the study to be followed up or monitored from time to time for their adherence

  2. Ensure the taste, dosage and duration of the drug shall be well tolerated by the study participant
• Measure Outcome Variables


  • Outcome should be assessed accurately and precisely
  • Continuous outcome variables have the advantage over dichotomous ones of enhancing the power of the study permitting a smaller sample size
  • If the outcome is dichotomous, power depends more on the number of events than on the overall number of subjects
• Run-in design


  Useful design for increasing the proportion of study subjects who comply with the intervention and follow-up procedures
• Factorial Design


  Answers two separate research questions in a single sample of subjects
• Randomization Matched pairs


  Pairing participants based on some factors like age sex then randomly assigning which member of each pair goes to the study group