APPLICATION OF PROPENSITY SCORE IN NON-RANDOMIZED CLINICAL TRIAL

[Objective]To discuss and evaluate the effect and practicability of propensity score in analyzing non-randomized clinical trial, and write the SAS program based on the propensity score. [Methods] Calculated the propensity score of every object by employing the logistic regression to get a new sample with covariates balancing as treatment group, and then it was compared with the control group by using of quintile stratification. [Results] Analyzing the non-randomized clinical trial data with quintile stratification method could adjust the imbalance of concomitant variable between two groups. [Conclusion] When randomized trials are not feasible, propensity score is an effective method to analyze the observational data.