Alcohol and fatal injury: the use of routinely collected fatality data in community prevention evaluation

This study analyzed patterns of alcohol involvement among coroners' cases (which typically include blood alcohol content information) to develop a method for weighting death-certificate cases (which typically do not include blood alcohol content information) for likelihood of alcohol involvement for purposes of alcohol-involved injury intervention evaluation.

The coroners' data analyzed here were collected from four California communities and correspond to all injury coroner cases between 1987 and 1996 in those communities. The death-certificate data were provided by the State of California, Department of Health Services, and correspond to all injury deaths occurring in that state between 1980 and 1996. Each injury fatality in the death-certificate data was assigned a probability of alcohol involvement based upon the coroners' data. These were then summed to provide an estimate of, or "surrogate measure" for, the total number of alcohol-involved fatal injuries per 10,000 California residents aggregated across the state of California by month for the period from January 1980 to 1990. As a test of this estimate, we examined its responsiveness to an intervention designed to reduce alcohol-involved injuries which was implemented in California in 1990 using a time series analysis technique (ARIMA) that corrects for serial autocorrelation typically found in time ordered data.

This analysis found an effect during the postintervention period (p = 0.046). An alternative model testing for intervention effects on all injury fatalities did not find an effect.

This surrogate measure seems responsive to intervention effects and may provide a useful tool for interventions designed to reduce alcohol-involved injuries.