SPS Plenary Session | May 21, 2019 8:15 am | Kevin Couloures and Daniel Tsze
Dr. Couloures began by explaining what data mining is and then provided some examples of the studies that have been completed so far using the Pediatric Sedation Research Consortium database. He summarized research findings and use of the database for the purpose of quality improvement (both for multi-site and single-site studies). Another potential use of the database is to develop tools (e.g. scoring) that could provide pre-sedation risk for adverse events, for example. Future directions for the data may include the use of machine learning algorithms. Looking at data without bias may show unanticipated findings.
The database has about 500,000 cases thus far, which could provide enough of a sample to provide trends on rare events. However, the earlier data (over 10 years ago) may not be as relevant to contemporary practice. The use of the same data for prediction and validation is also possible through statistical methods.
Dr. Tsze discussed prospective trials using the PSRC and contrasted their utility with randomized control trials. Some benefits of prospective trials include: they can accommodate novelty for the exposure/intervention, data is high quality, and data can be collected that are not available retrospectively. Dr. Tsze explained that the large number of potential participating institutions and the expertise and existing infrastructure of the PSRC makes it a unique group with great research potential. He then summarized the current prospective PSRC study on Intranasal Dexmedetomidine at PSRC sites. This study will provide feasibility of future prospective studies and information to develop communication between sites, the management of multiple IRBs, and the collaboration amongst multiple sites, including the development of a data collection tool.
Currently 397 patients have been enrolled during the data collection period. Screenshots of the study data collection tool were presented and described by Dr. Tsze. Some unique characteristics of the data collection tool include item response technology to reduce the amount of information required during data entry, methods of collecting time to recovery from sedation, and additional granularity related to intranasal administration of medication and sedation scoring at five-minute intervals. There were several lessons learned about development of tools, including need for back-up plans and need to have funding for these types of studies. Dr. Tsze invited members to consider and discuss potential research questions that have not been explored previously.