Statement of purpose Crash report data is generally limited to events that occurred just prior to the crash, the crash event itself, and the conditions of people involved immediately after the crash. Linking crash data to other relevant sources of existing data would allow us to better place crash events both in the context of the drivers’ histories and their short- and long-term outcomes, thus gaining a more comprehensive picture of the causes and implications of crashes.

Methods/approach Since 2011, we have been developing the New Jersey Traffic Safety Outcomes Program (NJ-TSO) data warehouse, a unique source of individual-level data that can be used to answer a host of important research questions in traffic safety. In all, 8 statewide data sources from 2004–2014 will be linked: (1) driver licensing (n=10 million); (2) traffic-related citations; (3) police-reported crashes (n=6 million drivers); (4) Census; (5) electronic health record data for NJ residents who were patients of the CHOP healthcare network (n=150,000); (6) birth certificate; (7) death certificate; and (8) hospital discharge. Databases were/will be linked using a combination of hierarchical deterministic and probabilistic methods.

Results We will describe in detail: collaborations established in order to successfully conduct this project; the processes by which databases were linked and how validity was assessed; identifiable and analytic data elements available in the NJ-TSO; and exemplar novel research questions that have and will be addressed via the NJ-TSO.

Conclusions Though not typically created for research purposes, administrative databases can provide a wealth of data useful for and important to traffic safety research. The value of each of these data sources to traffic safety increases exponentially when linked, thus catalysing further advancements in knowledge.

Significance and Contribution This innovative project involves development of one of the most comprehensive traffic safety data warehouses to date.