Concordance of motor vehicle crash, emergency department, and inpatient hospitalization data sets in the identification of drugs in injured drivers.

Prescription drug overdoses, abuse, and sales have increased dramatically in the United States in the last decade. The purpose of the present study was to link crash data with emergency department (ED) and inpatient hospitalization data to assess the concordance between the data sets in the identification of the presence of drugs among injured motor vehicle drivers (passenger cars, passenger trucks, light trucks, and semi-trucks) in Kentucky. Kentucky CRASH data were probabilistically linked to ED data sets for years 2008-2010 and to inpatient hospitalization data sets for years 2000-2010. Statistical analyses were performed. Of the 72,529 linked crash/ED visits, there were 473 drivers with an associated nondependent abuse of drugs diagnosis in the ED, and 930 drivers had drug involvement recorded in the CRASH data (only 163 cases overlapped with drug involvement both recorded in CRASH data and coded as nondependent abuse of drugs in the ED); 64 drivers had multiple drug types present in their system. Of the 20,860 total linked crash/inpatient hospitalization cases, there were 973 drivers diagnosed with nondependent abuse of drugs in the inpatient hospitalization record and 499 drivers had drug involvement recorded in the CRASH data (only 207 overlapped); 250 drivers were diagnosed with multiple drugs in their system. Surveillance data from multiple public health data sets is necessary to identify the presence of drugs in injured drivers involved in motor vehicle crashes. The use of a single surveillance data set alone may significantly underreport the number of drugged drivers who were injured in a motor vehicle collision.