Design principles and statistical considerations in periodontal clinical trials

Although clinical trials are a small subset of conducted biomedical research, they have become powerful investigational tools for the evaluation and approval of new treatments by clinician groups and regulatory agencies like the US Food and Drug Administration. To impact the delivery of care, trials in general must meet three criteria: clarity, comparability, and generalizability. Accordingly, trials can offer meaningful data if they have procedures which are well defined and subjects who represent a reasonably homogeneous population. The evaluation of periodontitis interventions presents several challenges due to the disease's heterogeneity and its irregular, episodic pattern; nevertheless, the intent of these novel interventions is to prevent, diagnose, inhibit, or reverse periodontal disease progression. Careful consideration of the trial's objectives should dictate clinical endpoints (primary and surrogate), comparison groups (placebo, standard therapy, test therapy), and equivalence versus superiority as the basis for conclusions. Several design elements such as control population specification, randomization, masking, sample size calculation, and standardization of procedures for patient care and assessment can decrease potential bias and variability. In both parallel and paired (split-mouth) design trials, multiplicities of endpoints, treatments, and subgroups require strategies which address the broader scope of chance findings without excessive loss of study power. Also, the longitudinal assessment of multiple periodontal sites within patients produces correlated data structures for which analytic methods need to account for the appropriate sampling unit. With these design and analytic elements, clinical trials can provide important evidence to investigators, patients, and governmental agencies for the introduction of novel interventions in periodontal practice.