Objective:​ Before October 1, 2002, Maryland's regulations for relicensing drivers with 2 recorded alcohol-related traffic violations distinguished between offenders with 5 or more years between their first and second violations and those with less than 5 years. Our research examined whether this policy was supported by differential probabilities of recidivism and violation-free survival.

Methods:​ We compared recidivism rates and survival probabilities among the 2 latency subgroups and 2 control groups (first offenders and drivers with no previous alcohol-related traffic violation). Data were extracted from Maryland's driver record database and segregated files and analyzed by age quintiles using Cox proportional hazards models containing identifiers for risk factors, including prior violations. All drivers (N = 64,536) were matched on age quintile, gender, and month of offenders’ index violations. Effects of violation histories on survival and recidivism probabilities were measured by contrasts of regression coefficients.

Results:​ Among second offenders, the shorter latency subgroup consistently had higher recidivism and lower violation-free survival than the longer latency subgroup, whose rates fell between those of first offenders and the shorter latency subgroup. Although highly significant, the subgroup differences were small and paled by comparison to differences between first and zero offenders in probability of a subsequent violation.

Conclusions:​ An earlier study that showed similar overall recidivism for these latency subgroups helped encourage Maryland to change its regulations governing license reinstatement. New regulations issued October 1, 2002, focused on 2 alcohol violations “during any period of time” where investigation indicated alcoholism or unaddressed alcohol problems. To obtain relicensure, these offenders could be required to enter or complete a lengthy certified alcohol treatment program. Our current results are consistent with these requirements. License reinstatement should be primarily guided by the extent of alcohol impairment, especially because both latency subgroups showed higher risks of recidivism than first offenders, who themselves had comparatively high risk.