No abstract

a regularly scheduled comprehensive review, we are able to track trends on an individual and group basis. Therefore, developing new methods to improve our data prepares us better for challenges of external audits, such as CIBMTR and FACT (Foundation for the Accreditation of Cellular Therapy).

No abstract

hematopoietic cell transplant (HCT) event. Multiplying this by the number of transplants that are part of the CIBMTR Outcomes Research Database underlines the need for structures that document this 'data about data' (Metadata). The difficulty is showing how the complete set of data around a transplant event is connected in a way that transcends the collection method, i.e. form. Our aim was to demonstrate that a domain-driven architecture aligns with the forms-based model, and eases the introduction of collecting more data about HCT transplant events. The Biomedical Research Integrated Domain Group (BRIDG) model is publically available. The goal of BRIDG group is to have a common view of the data exchanged for semantic interoperability. The BRIDG model is intended to balance the concerns of the larger health care community, while being specific enough to apply to a particular subject area such as HCT. We extracted all Common Data Elements (CDEs) for all CIBMTR-mandated forms and associated each element to one of three contexts, Recipient, Donor and stem-cell product; most elements were in the Recipient context. Because no element could be described in isolation, instance diagrams were created to describe how one simple concept needed multiple BRIDG entities to be fully described. Some CDEs were described as relationships between entities rather than an attribute of an entity. We extended the generic BRIDG model to contain all identified elements. To this end, we requested the expansion of the BRIDG model to include a 'PerformedSubstanceExtraction' entity at the same level of inheritance as 'Perform-edSubstanceAdministration'. While the collection of a stem-cell product could have been described in the context of a 'Product' and a 'SpecimenCollection', the richness of the relationship between a Donor and a Recipient would not have been as obvious. This as well as other enhancements to the BRIDG model were included in BRIDG version 3.2. With this effort, we have documented each data point (CDE) collected on all the CIBMTR-mandated forms and the relationships between them. We intend to use the UML BRIDG model with the added HCT content as the specification for a physical database model. This physical model will help remove barriers that transplant centers experience in electronic transfer of HCT data to the Stem Cell Transplant Outcomes Database by providing a foundation upon which to develop their own in-house data systems, and eventual development of Electronic Medical Record (EMR) integration engines to submit data to the Outcomes Database.

surveillance cultures (P < .0001). When MMCI cases were compared to single MCI cases, significant differences included: female gender (P < .0349), receipt of allo-HSCT (P < .0373), vancomycin (P < .0012) and cefepime (P < .0009) use. There was no difference in the number of patients who experienced neutropenic fever between MMCI, MCI (P < .275) or culture negative cases (P < .1247). Strept mitis or C.difficileinfection occurred concommitantly or preceded the second MCI in 29(47%) cases. Overall mortality was significantly higher in MMCI cases when compared to cases without any positive cultures (P < .001) or patients with a single MCI (P < .0197). There was no difference in overall mortality for patients who developed MMCI <72hours (polymicrobial) versus MMCI>72 (non-polymicrobial; P < .2990). MMCI are an infrequent but serious cause of adverse events which occur during HSCT. Patients who are at high risk for developing MMCI require increased vigilance and early aggressive antibiotic therapy.