Statistical Outcome of Random Versus Selected Withdrawal of Dental Implants

Purpose: When performing clinical trials, missing data from withdrawn patients should be evaluated differently, depending on the reason for the withdrawal of the patients. The question is, if a certain type of patient drops out, will that affect the result? Could a randomly selected sample of a study population be used for analyses instead of evaluating each and every patient? The purpose of this study was to answer these questions. Materials and Methods: Detailed information on 1,738 implants in 487 patients was pooled together in a new database and used for statistical evaluations. Random or selected withdrawals were pulled from the database. Chi-squared tests were used for significance tests, and lifetables were used for survival analysis. Results: There was a difference in the outcome depending on whether the withdrawals were randomly chosen or selected. Random withdrawals could represent, in this study, as much as 50% of the included patients without changing the statistical results. If selected withdrawals were based on which jaw was treated, the statistical outcome did change, but it did not change if withdrawals were based on gender or age. Conclusion: Evaluation of reasons for withdrawals and withdrawn patient characteristics are of utmost importance when evaluating clinical trials. A randomly selected sample of the entire population could, however, be expected to give the same statistical value as the entire group, if the original material were large enough. Therefore, the use of study samples may more easily enable clinicians to do follow-up investigations.