Accident Analysis and Prevention 36 (2004) 783–794 Cost savings from a sustained compulsory breath testing and media campaign in New Zealand Ted Miller a,∗ , Michael Blewden b , Jia-fang Zhang b a Pacific Institute for Research and Evaluation, 11710 Beltsville Drive, Calverton, MD 20705, USA b The Centre for Social and Health Outcomes Research and Evaluation, Massey University, PO Box 6137, Wellesley Street, Auckland, New Zealand Received 25 July 2003; accepted 28 July 2003 Abstract This paper evaluates three approaches to compulsory breath testing (CBT) where all drivers stopped are tested: (1) intensive, moderate- profile CBT (plus zero alcohol tolerance for drivers under age 20, which was implemented simultaneously, remains in effect, and unavoidably is commingled with CBT in the effectiveness estimates); (2) CBT plus an enhanced media campaign; and (3) shifting to aggressively visible booze buses, which also streamlined drunk-driver processing, plus enhanced community campaigns against drunk-driving. Approaches 1 and 2 were implemented throughout New Zealand (NZ) in 1993 and 1995. Booze buses and community programs were added for about one-third of the country in late 1996. ARIMA time series models estimated the impact on serious and fatal injury crashes between 10 p.m. and 3 a.m., a proxy for alcohol-related crashes. A benefit–cost analysis assessed return on investment. Cost savings were analyzed from four perspectives: societal, governmental, drunk-drivers’, and people other than drunk-drivers (external cost). CBT plus zero tolerance reduced expected night-time crashes by 22.1% and enhanced media by 13.9%. Booze buses yielded a further 27.4% reduction where implemented. The program and associated crash reduction persisted until at least 2001 (the most recent data available). Estimated societal benefit–cost ratios were 14 for CBT, 19 for CBT plus enhanced media, and 26 for the comprehensive package. Government saved more than it spent on the program, especially with booze buses. Aggressive CBT plus zero alcohol tolerance for youth, media blitzes, and booze buses proved dramatically effective. Together, these four interventions halved late night serious and fatal injury crashes. Sustained effort seems to be critical. Better outcomes may be achieved with staged, increasingly visible and inescapable checkpoints than with an “ideal” initial program. It appears CBT is best implemented in conjunction with broader community-centered efforts to reduce drunk-driving. © 2003 Elsevier Ltd. All rights reserved. Keywords: Benefit–cost; Drunk-driving; Breath testing; Checkpoints; New Zealand; Accident; Injury 1. Introduction Of 572 road fatalities in New Zealand (NZ) in 1996, driver alcohol was a factor in 156. This toll was the lowest since at least 1980. It was less than half the alcohol-involved deaths in 1990 or the annual average from 1986 to 1990 (LTSA, 1997a). How was this dramatic decline achieved? Several related programs contributed. A major push on host responsibility and premise monitoring began in 1990 with the imple- mentation of the 1989 Sale of Liquor Act. In April 1993, random compulsory breath testing (CBT) was implemented, with a target of 1.5 million breath tests annually in a coun- try with 2.3 million registered vehicles (LTSA, 1997b). Blood alcohol limits (BALs) for drivers under the legal ∗ Corresponding author. E-mail address: miller@pire.org (T. Miller). alcohol purchase age of 20 were simultaneously (and per- manently as of this writing) lowered to 30 mg/100 ml from the long-standing limit of 80 mg/100 ml that applies to most drivers (so-called zero tolerance for youth). Day-time speed camera enforcement and a broader speed reduction cam- paign also were implemented, reducing day-time crashes (Mara et al., 1996). In October 1995, a retooled, aggres- sive road safety advertising campaign began, modeled on a successful campaign from Victoria, Australia (CRDD, 1996). Finally, in October 1996, the northern half of New Zealand’s North Island (including Auckland) unveiled ag- gressively visible “booze buses” and raised CBT’s profile, with the intent of increasing its general deterrence effect as a disincentive to alcohol-impaired driving. Simultaneously, this area increased community involvement in CBT and other efforts against drunk-driving. Early evaluations of the NZ CBT program yielded promising results (Mara et al., 1996; New Zealand Police, 0001-4575/$ – see front matter © 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.aap.2003.07.003