Harmonization of Databases for Traumatic Spinal Cord Injury

Research Objectives: To harmonize data of people with traumatic spinal cord injury (TSCI) from the National Spinal Cord Injury Model Systems (SCIMS) Database and Veterans Health Administration (VHA) Spinal Cord Injuries and Disorders (SCI/D) Registry.

Design: Not applicable.

Setting: Not applicable.

Participants: Not applicable.

Interventions: Not applicable.

Main Outcome Measures: Not applicable.

Results: The information and documents from the two databases, which includes variable list, data collection forms, and data dictionary, were assembled and reviewed by the researchers from the two databases. The SCIMS Database contains information of participants at the time of injury (Form I) and follow-up interview (Form II). Form I data set (678 variables) includes baseline demographic and clinical information. Form II data set (337 variables) includes socio-demographic and outcome data obtained during follow-up interviews. The VHA SCI/D Registry includes the data of Veterans’ each VHA health care utilization at inpatient or outpatient settings. The data collected are demographic, injury, and medical characteristics; psychosocial well-being; functional outcomes; as well as SCI/D specialty and non-SCI/D specialty inpatient and outpatient care, health process measures, and laboratory results. The common variables were found in VHA SCI/D Registry for 12 Form I variables and 75 Form II variables. All 12 Form I variables and 21 Form II variables are easy find information. The crosswalk was created for these easy find variables. Additional works are needed to get common variables from VHA SCI/D Registry for 53 Form II variables.

Conclusions: This study identified and harmonized common variables between the SCIMS Database and VHA Spinal SCI/D Registry, which could provide a foundation for further expanding with other spinal cord injury (SCI) databases (e.g., Rick Hansen SCI Registry in Canada), linkage with other data sources (e.g., census data through geographic identifiers), and probabilistic matching with data collected from other clinical settings.

Author(s) Disclosures: Nothing to disclose