Purpose. To evaluate the U.S. Coast Guard's Boating Accident Report Database (BARD) and estimate the incidence of recreational boating injuries.

Data and Methods. We compared 2002 BARD data with boating injury cases from mortality data and hospital discharge (HD) and emergency department (ED) discharge data. We constructed boating injury incidence estimates for three severities: 1) fatal; 2) non-fatal, hospital-admitted; and 3) non-fatal, non-admitted, medically treated. We compared state-by-state counts from BARD with those from mortality and HD and accepted whichever count was higher. For 14 states lacking HD data, we factored up the BARD counts based on average undercount from HD states. The process for estimating non-admitted incidence was more complex. We followed a similar procedure with ED data from 11 states. We multiplied the ED estimates by a national ratio of cases treated in doctor's offices vs. EDs.

Results. In 2002, recreational boating accidents resulted in an estimated 763 fatal, 2,200 hospital-admitted, and 28,700 non-admitted injuries. BARD generally identified more boating deaths than mortality data. For several states, BARD also recorded more hospital admissions than HD data. The quality of reporting in BARD and the quality of E coding in HDD varied widely between states. BARD missed most non-admitted injuries, but sparse data make it difficult to produce a precise alternative estimate.

Conclusions. BARD captures all but 1-2% of recreational boating deaths, but it appears to miss about 20% of hospital-admitted injuries and 90% of non-admitted injuries. The quality of reporting of boating accidents varies widely by state.