Performing routine protocol biopsies after liver transplantation is not a practice that is done in most hospital settings. The current practice is to perform a biopsy only if a patient exhibits signs or symptoms of graft dysfunction. This practice is primarily due to reasons such as cost and risk to the patient. However, there continues to be much research published on various outcomes in this population and the question lies in the accuracy of the research findings from studies using only non-systematic biopsy information.; The National Institute of Diabetes, Digestive and Kidney Diseases Liver Transplantation Database is a liver transplantation database that includes both systematic and non-systematic biopsies on recipients from three medical centers across the country. This dataset is considered the 'complete' dataset since it included information at predetermined time intervals and when a patient exhibited signs or symptoms of a complication. Simulated populations based on the 'complete' dataset were generated to test whether the predictors of acute rejection from data of non-systemic (only if signs or symptoms of complications were present) follow-up supported those from the 'complete' dataset. Two different statistical methodologies were evaluated. One was the Anderson-Gill method of analyzing multiple failures and a 'first episode' method which is the common practice in most research studies.; The results of the simulations indicated that if the outcome of interest is highly correlated with signs or symptoms of disease, then the 'complete' and the 'complications only' datasets showed similar results. If the correlation is decreased, the results from the non-protocol datasets are disparate. Thus, complication only data collection methods would not be appropriate for diseases where the outcome is not highly correlated with physical symptoms. Also, the Anderson-Gill as compared to the 'first episode', methodology was shown to be the preferred analytical method for multiple failure data.
展开▼
