Experimental Design - Part II PDF

EXPERIMENTAL STUDY DESIGNS AFAF TAHOON CERTIFIED RESEARCH COORDINATOR CAIRO UNIVERSITY one of 2 conditions not present  Pre experimental True experimental O N E S H OT C A S E POST TEST CONTROL...

EXPERIMENTAL STUDY DESIGNS AFAF TAHOON CERTIFIED RESEARCH COORDINATOR CAIRO UNIVERSITY one of 2 conditions not present  Pre experimental True experimental O N E S H OT C A S E POST TEST CONTROL No control S T U DY Both groups are post-tested No random assignment Conclusion is drawn from the difference Single measure is taken between these groups ONE GROUP PRE-POST PRE-POST CONTROL Outcome measured twice Degree of change in each group EXPERIMENTAL S TAT I C G R O U P S O LO M O N 4 G R O U P Randomly selected subjects are Has control group not receive the ttt placed into 4 groups STUDY DESIGNS No random assignment 2 groups post test only No pretest measure 2 groups pre-post test  why not STATIC ADV. randomize OVER RCT ? ETHICAL CONCERN Exposure is critical as surgery , chemical PRACTICAL CONCERN No time to randomize participants or collect pre-intervention measurements  Time series Multiple time series QUASI- EXPERIMENTAL researcher can better understand the true value of the dependent No control group Testing effect + control group variable before the intervention starts and intervention had lasting effects MATCHING SOLUTION FOR LACK OF RANDOMIZATION matched pair STATIC ADV. OVER QUASI NO MEASUREMENT BIAS MULTIPLE measurement could affect calibration Clear temporal sequence N O T S U S C E P T I B L E T O R E G R E S S I O N T O WA R D S T H E M E A N The problem will be that their next measurement will naturally become less extreme, providing a fake effect REGRESSION TOWARD THE MEAN TRUE EXPERIMENTAL TRAIL EXAMPLE OF OBSERVATIONAL STUDY AS TTT  time frame ?! ttt was not randomly assign COHORT VS RCT If RCT will be highly unethical (imagine randomizing people to either smoke or not) Cause effect relationship biased coffee & cancer with smoking confounder RCT BLINDING CONCEALMENT R A N D O M I Z AT I O N C O M PA R I S O N TRAILS SETTING NO.OF PA R T I C I PA N T S PA R T I C I PA N T S E X P O S U R E & RESPONSE TO INTERVENTION  SETTINGS PA R T I C I PA N T S K I N D OBJECTIVES DESIGN CLASSIFICATION SETTINGS AND PARTICIPANT'S TYPE Community trial Participant is entire community rather than individuals Filled trail Healthy participants WHO at risk in general PPL Pragmatic trial Real world situation Explanatory trail Controlled environment  reported blood outcomes chemistry Internal vs external validity INTERVENTION TRAILS CAUSALITY TRAILS PHASE 1 is it safe PHASE 2 does it work PHASE 3 is it better PHASE 4 can it stay in the market matched pair trail historical trail DESIGNS Parallel cross over factorial standard treatment or not according to aim KT and US EXAMPLES PARALLEL DEAIGN Most common used Simple Clear temporal sequence Baseline characteristics is performed either before or immediately following the randomization There should be an equal number in each arm  CROSSOVER Removing patient variation Each of the participants is given all the study interventions in successive periods Order of receive each of the study interventions is determined at random NOT ALLOWED chronic conditions that are fairly stable over time or interventions that last a short time within the patient and that do not interfere with one another Allow the response of a subject to treatment A to be contrasted with the same subject's response to treatment B PARALLEL VS CROSSOVER Number of participants parallel is more participants Variance within –between variance is more in parallel Time required cross over is more HISTORICAL TRAIL All participants receive the ttt being studied The trial results are compared to: a previous study control group or normative data The study is neither randomized nor is the data collected concurrently Oncological studies SUMMARY OF EXPERIMENTAL DESIGNS static group KEY TAKEAWAYS  THANK YOU Afaf Tahoon [email protected] [email protected]