Another minute of searching has led me to this article:
http://aje.oxfordjournals.org/cgi/co...ract/157/2/131
Association of Alcohol-related Laws with Deaths due to Motor Vehicle and Motorcycle Crashes in the United States, 1980–1997
From 1980 through 1997, laws mandating a BAC of 0.08 g/dl per se and zero tolerance laws had modest associations with a reduction in overall traffic mortality: approximately 3 percent and 4 percent, respectively. Both had stronger associations with reductions in crash fatalities attributable to alcohol: approximately 14 percent and 12 percent, respectively. Administrative license revocation laws were associated with a 5 percent reduction in overall mortality and alcohol-related mortality. Implementation of mandatory jail terms for persons first convicted for driving under the influence of alcohol was associated with an overall 2 percent increase in traffic mortality, while there was a 5 percent reduction in alcohol-related traffic mortality. Implementation of laws that explicitly allowed police to set up sobriety checkpoints for controlling drinking and driving did not appear to be related to a reduction in overall traffic mortality or alcohol-related traffic mortality.
Our study had several limitations. With the methods used, we were unable to measure any police effort to enforce a law. Substantial variations might occur within a state and between states. We did not account for any policy, law, or program applied at a county or city level. This is especially important for the evaluation of sobriety checkpoints, which have been mostly applied at local levels and at specific times (47). Another potential limitation, one that is common to most studies that deal with this subject, is the high proportion of missing information on BACs. We used multiple imputation methods in an attempt to reduce the bias that may result from only considering data without missing values. We dealt with missing BAC values by employing the multiple imputation methods suggested by Rubin et al. (43); we think the method of Rubin et al. is an improvement over the method of Klein (44), because continuous rather than categorical values of BAC are imputed and because multiple imputation allowed us to account for the uncertainty regarding the missing BAC values. Finally, defining denominators to evaluate the association of laws with deaths is difficult for motorcycle crashes. Data on total vehicle miles driven for motorcycles is not available by state and year, because not all states report this information. Data on registered motorcycles are available but serve only as a rough estimate of motorcycle use.
Our study had several strengths. We used models appropriate for longitudinal count data, took into account changes in mortality over time, and evaluated possible interactions between the laws. We calculated alcohol-related deaths based on the relative risk of being in a fatal crash due to a specific driver BAC, as described by other investigators (38, 48), because this is a more accurate assessment of alcohol-relatedness than proxy measures such as single nighttime crashes (11, 49) or police-reported alcohol consumption (49, 50). This explained the smaller percentage of alcohol-related deaths that we estimated in comparison with some other studies (1, 6, 30, 49, 51, 52).
BAC 0.08 g/dl per se laws
Our finding of a 14 percent reduction in alcohol-related deaths due to implementation of BAC 0.08 g/dl per se laws is similar to that of several recent studies (9, 10, 53). Another study (6) found only five of 11 states to have a significant reduction in mortality. Our findings differ from the findings of a North Carolina study (14) that reported no overall association of these laws with a reduction in alcohol-related deaths.
Two studies (6, 54) used the ratio of the number of fatalities for drinking drivers to the number of fatalities with no drinking drivers as their dependent variable. Another study (9) used a proportion of fatalities rather than the actual count. This use of ratios could bias estimates as described by Kronmal (55) and others (56–59). A recent systematic review of studies reported an estimate similar to ours (60).
Zero tolerance laws
Our estimate of a 12 percent reduction in alcohol-related mortality due to the implementation of zero tolerance laws is consistent with findings from other studies as reported in a systematic review (12). Previous studies estimated reductions in mortality between 11 percent and 33 percent after implementation of zero tolerance laws in the United States and Australia. Our finding that the association of these laws was greater among alcohol-related crashes supports the view that these laws affected their target population.
Administrative license revocation laws
Administrative license revocation laws were associated with a 5 percent reduction in overall motorcycle mortality and a 2 percent reduction in alcohol-related motorcycle mortality. This is consistent with findings by Whetten-Goldstein et al. (49), who reported a statistically significant fatality rate difference of –0.04 per 1,000 persons, and Zador et al. (61), who reported a 9 percent reduction in nighttime fatal crashes.
Sobriety checkpoints
Enacting laws that allowed police to set up sobriety checkpoints did not appear to be related to a reduction in overall mortality, and it had a minimal, statistically nonsignificant association with lower alcohol-related mortality. Other studies (47, 62–64) have reported a benefit from sobriety checkpoints. Those studies used information about the degree of enforcement, whereas we could not account for actual enforcement.
Mandatory jail terms upon first conviction
For all alcohol-related deaths, mandatory jail terms imposed upon the first conviction for driving under the influence of alcohol were associated with a 5 percent reduction in mortality. Previous studies have differed in their conclusions regarding these measures (49, 65–67), ranging from no effect (49, 67) to a 40 percent decrease in reoffending (65).
Other laws
Primary seat-belt laws were associated with reductions in traffic mortality in all crashes and in alcohol-related crashes. This association was slightly weaker for alcohol-related crashes—a finding consistent with evidence that intoxicated drivers are less likely to comply with seat-belt laws (68–70). Secondary seat-belt laws appeared to have no association with reductions in traffic mortality among any of the groups studied. This is consistent with other evidence that primary seat-belt laws are more strongly related to mortality reductions than are secondary seat-belt laws (68, 71).
Universal helmet laws appeared to be strongly related to reductions in traffic mortality in all motorcycle crashes, as well as in alcohol-related crashes. Several studies have evaluated the relation of universal helmet laws with mortality (15, 16, 18, 72–74), estimating changes that range from 12 percent to 73 percent. Our results are generally consistent with those findings. Selective helmet laws were not associated with significant reductions in motorcycle fatalities.
This study provided information on the effect of alcohol-related laws in the 50 states and the District of Columbia during a period of 18 years. Our results support recent policy measures that set a national level of 0.08 mg/dl for BAC (75). Additional measures such as zero tolerance laws and administrative license revocation laws may also have reduced mortality due to drunk driving.
I would do more searching for the lazy drunks but I have to get back to work; my lunch is over.