CONFIDENCE LIMITS IN EVALUATING CONTROLLED THERAPEUTIC TRIALS