Basic concepts and statistical aspects

Author

Christian Mélot
M.D., Ph.D., ULB, M. Sci. Biostatistics, Limburgs Universitair Centrum
Professor of Biostatistics at ULB,
Head, Department of Emergencies, Erasme Hospital

Summary

Clinical studies can be observational or experimental.

• Observational studies are considered weaker in terms of proof.
• The randomised controlled clinical trial (RCT) is considered as the gold standard. One distinguishes:
• Inter-patient designs (parallel groups): a design with parallel groups requires a larger number of subjects
• Intra-patient designs (crossover): a design with a crossover design has the disadvantage that the effect of the first treatment might persist during the administration of a second treatment (carry-over effect).

The real effect of the treatment is altered by:

• systematic (bias). The purpose of controlled randomised trials is to control bias with randomisation, double blinding, use of a control group and intention to treat analysis.
• random (chance) errors. The effect of chance is controlled by increasing the number of subjects included, which increases the power of the study.

The control group might be treated by a placebo (negative control) or the best treatment currently available (positive control). The primary and secondary endpoints are defined in the protocol. The double blinded evaluation reduces the bias linked to the observer and the patient. In the course of a trial, some subjects drop out. Therefore there are two ways to calculate the response rate:

• according to the number of subjects who completed the study (per protocol analysis)
• according to the number of subjects initially randomised (intention to treat analysis). The intention to treat analysis is more adequate because it corresponds better to real life.