1 Takemi Program, Department of International Health, Harvard Shool of Publi Health, Boston, MA, USA; 2 Department of Emergeny Mediine, St Mary s Hospital, Kurume, Japan; 3 Division of Pharmaoepidemiology and Pharmaoeonomis, Brigham and Women s Hospital, Boston, MA, USA; 4 Department of Health Poliy and Management, Harvard Shool of Publi Health, Boston, MA, USA; 5 Department of Environmental Health, Harvard Shool of Publi Health, Boston, MA, USA Correspondene to: Dr T Nagata, Department of Emergeny Mediine, St Mary s Hospital, 422 Tsubukuhonmahi, Kurume-ity, 830-0047, Japan; nagata.takashi@gmail.om Aepted 29 November 2007 This paper is freely available online under the BMJ Journals unloked sheme, see http:// injuryprevention.bmj.om/info/ unloked.dtl Effetiveness of a law to redue alohol-impaired driving in Japan T Nagata, 1,2 S Setoguhi, 3 D Hemenway, 4 M J Perry 5 ABSTRACT Objetive: To estimate the effet of a new road traffi law against alohol-impaired driving in Japan. Methods: Japan passed a new road traffi law in June 2002 intended to redue alohol-impaired driving by dereasing the permissible blood alohol level and by inreasing penalties. Using data olleted from polie reports, the number of traffi fatalities and injuries were analyzed by time series. Results: Simple omparisons of the average of all severe traffi injuries, traffi fatalities, alohol-impaired traffi injuries, alohol-impaired severe traffi injuries, and alohol-impaired traffi fatalities per billion kilometers driven showed redutions after enatment of the new road traffi law in June 2002. The rate of alohol-related traffi fatalities per billion kilometers driven dereased by 38% in the post-law period. In segmented regression analyses with adjustment for baseline trends, seasonality, and autoorrelation, all traffi injuries, severe traffi injuries, alohol-impaired traffi injuries, alohol-impaired severe traffi injuries, and alohol-impaired traffi fatalities per billion kilometers driven delined signifiantly from baseline after the new traffi law. Conlusion: Large, immediate publi health benefits resulted from the new road traffi law in Japan. Traffi injury is a worldwide publi health issue. Annually, more than a million people are killed on the world s roads; in the USA there are over 40 000 motor vehile fatalities eah year. 1 In April 2004, The World Health Organization (WHO) and World Bank released the World report on road traffi injury prevention. 2 It stated that, in 1990, road traffi injuries were the ninth leading ontributor to the global burden of disease, but are predited to beome the third leading ontributor by 2020, unless appropriate ation is taken. Alohol-impaired driving (AID) is a leading ause of traffi fatality both in developed and developing ountries. A review of studies in low- and middleinome ountries found that blood alohol was deteted in 33 66% of fatally injured drivers. 2 Although drinking and driving legislation, inluding administrative liense suspension, random sreening, and lowering the legal blood alohol limit, has been shown to redue traffi fatalities, many ountries have not enated suh measures. 3 In Japan, the absolute number of traffi deaths fell from 11 451 in 1992 to 7358 in 2004, an average derease of 3 4% per year. The National Polie Ageny (NPA) states that this redution is due to improvements in poliy, roads, vehile engineering, driver behavior, and the nation s emergeny medial system. 4 Original artile In the 1990s, blood alohol was deteted in 14 16% of fatally injured drivers in Japan. 5 To redue alohol-related traffi fatalities, in June 2002, the Japanese government enated a new road traffi law against AID. The law redued the allowable breath alohol test from 0.25 to 0.15 mg/l and blood alohol onentration (BAC) from 0.05% to 0.03%, and inreased the penalties for AID. The fine inreased from approximately 50 000 to 500 000 Yen (US$425 to $4250 in 2002), and driver s liense points were severely imposed. Moreover the law made bartenders and passengers ulpable in addition to arrested drivers. 4 5 This study evaluates the impat of this new law on traffi injuries and fatalities in Japan using time series analysis. We test the null hypothesis that the new road traffi law did not redue AID traffi fatalities, AID severe traffi injuries, or AID traffi injuries; nor did it redue traffi fatalities, severe traffi injuries, or all traffi injuries. The alternative hypothesis is that there were signifiant hanges in the time series interept or slope after enatment of the law. METHODS Data and their soure Two data soures were used, the NPA Crash Reports and the Ministry of Land Infrastruture and Transport Driving Data. The NPA Crash Reports are managed by the traffi bureau of the NPA. 4 Data on traffi rashes are reported by loal polie stations and then entralized by the NPA. The loal polie stations use a standardized protool for data olletion, and all data on injuries and deaths are supplied by emergeny medial are professionals through medial ertifiates. Initially all the traffi inidents are olleted in the loal polie stations, then are olleted in the headquarters of the prefeture polie agenies, and finally systematially reported to the NPA. However, the number of traffi injuries in the NPA Crash Reports is thought to be underestimated. 6 Data on traffi rashes were provided anonymously through the website of the NPA in Deember 2005. We olleted monthly data on the total number of traffi fatalities, severe traffi injuries, and all traffi injuries as well as the total number of AID traffi fatalities, AID severe traffi injuries, and all AID traffi injuries from January 1998 to Deember 2004. 4 7 The Ministry of Land Infrastruture and Transport provides publi monthly and annual data on total kilometers driven in Japan. Total kilometers driven is defined as the total distane driven by all of the registered vehiles in Japan. Total kilometers driven are measured on the basis Injury Prevention 2008;14:19 23. doi:10.1136/ip.2007.015719 19
Original artile Table 1 Inidene of traffi injuries in Japan (2004) Per 100 000 Total population Per billion km driven All traffi injuries 1 190 487 932.3 1239.6 All severe traffi injuries 72 777 57.0 75.8 All traffi fatalities 7 358 5.8 7.7 AID traffi injuries 22 291 17.5 23.2 AID severe traffi injuries 2 081 1.6 2.2 AID traffi fatalities 761 0.6 0.8 Total population 127 687 000 Total km driven 960 384 603 AID, alohol-impaired driving. of speifiations by the ensus law. About 30 000 samples among 77 000 000 registered vehiles in Japan are hosen randomly and anonymously. Monthly total kilometers driven by eah vehile is exatly measured aording to the odometer readings. We obtained the monthly data on total kilometers driven from January 1998 to Deember 2004 from the website of the Ministry of Land Infrastruture and Transport. 8 The study used data available in the publi domain, and was given human subjets exemption by the Harvard Shool of Publi Health Human Subjets Committee. Traffi injuries, traffi fatalities, and rates In Japan, the sample of traffi injuries in motor vehile inidents inludes those of drivers and passengers, and 78% of traffi fatalities and 82% of all traffi injuries our to drivers in the offiial polie report. 4 Traffi fatalities are defined as deaths within 24 h of a rash on a road involving a vehile with an engine, the death being the result of the rash. Thus, fatality data inlude all motor vehile-related deaths (eg, truks, motorars, motoryles, biyles, pedestrians). Severe traffi injuries are defined as injuries that resulted in more than 30 days of medial are after the inident, whih inludes both hospitalization and outpatient treatment. All traffi injuries are defined as any traffi injuries reported to the NPA, whih inlude severe traffi injuries and fatalities. AID traffi fatalities, severe traffi injuries, and total traffi injuries are ounted as the number of inident ases involving AID. Involvement of AID is determined by the polie using breath tests. Before June 2002, AID was defined as a measurement of alohol.0.25 mg/l by a breath test or.0.05% by blood alohol test. After June 2002, AID was defined as.0.15 mg/l by a breath test or.0.03% by blood alohol test. The number of AID ases was divided by total kilometers driven to alulate the rates for traffi fatalities, severe traffi injuries, and all traffi injuries. In this study, we alulated rates using total number of traffi fatalities, severe traffi injuries, all traffi injuries, in addition to AID traffi fatalities, AID severe traffi injuries, and AID traffi injuries divided by total kilometers driven. Statistial analysis First, simple ounts were made of the annual nationwide traffi fatalities from January 1998 to May 2002 (pre-law) ompared with June 2002 to Deember 2004 (post-law). We next onduted segmented regression analyses to evaluate the effetiveness of the June 2002 traffi law on AID after adjustment for pre-intervention trends and for seasonality of traffi injuries. 9 Beause of the large number of observations, we assumed normal distributions of monthly rates and onduted least squares linear regression. We tested for first- and higherorder autoorrelation by alulating ordinal and generalized Table 2 Average national traffi fatalities, severe traffi injuries and traffi injuries per billion kilometers driven before and after enatment of the new road traffi law, Japan (January 1998 May 2002 ompared with June 2002 Deember 2004) Perentage Pre-law Post-law hange All traffi injuries 1156.57 1229.02 +6.1 All severe traffi injuries 80.76 77.71 23.8 All traffi fatalities 9.34 8.00 214.3 AID traffi injuries 36.74 24.77 232.6 AID severe traffi injuries 3.68 2.33 236.7 AID traffi fatalities 1.39 0.86 238.1 AID, alohol-impaired driving. Durbin Watson statistis. 10 When autoorrelation was deteted, we used stepwise autoregression to selet the order of the autoregressive error models. 11 Calendar month served as the unit of analysis, with the primary outome measure being the event rate. Six events (monthly rate of traffi fatalities, severe traffi injuries, all traffi injuries, in addition to aloholrelated traffi fatalities, severe traffi injuries, all traffi injuries) were modeled. There were 84 monthly intervals: 53 before the intervention, and 31 after the intervention. The formula for the basi model is as follows: Y t = b 0 + b 1* Time t + b 2* Intervention t + b 3* Time after Intervention t + b 4* February + b 5* Marh + b 6* April + b 7* May + b 8* June + b 9* July + b 10* August + b 11* September + b 12* Otober + b 13* November + b 14* Deember + V i where: Y t is a monthly rate of traffi fatalities, severe injuries, and total injuries per vehile kilometer driven per month; b i represent the oeffiients of the model t; Time is a ontinuous variable indiating the time in months at time t from the start of the observation period (in this study it started in January 1998 through Deember 2004); Intervention is an indiator for time t ourring before (intervention = 0) or after (intervention = 1) the new law, implemented at month 54 (June, 2002) in the series; Time after Intervention is a ontinuous variable ounting the number of months after the intervention at time t, oded 0 before the new traffi law and (time253) after the new traffi law; and V i is autoregressive errors. All analyses were onduted using SAS (SAS Institute, Cary, North Carolina, USA). A hange in either the level or the slope of the time trend beginning at the intervention date suggests an effet of the law. RESULTS In Japan in 2004, there were 7358 traffi fatalities within 24 h of a rash, 5.8 per 100 000 persons (table 1). Traffi injuries totaled 1 190 487 or 932 per 100 000 persons. Severe injuries, totaled 72 777, or 57 per 100 000 persons. The proportion of traffi fatalities to severe traffi injuries to total traffi injuries was 1:9.9:162. Just over 10% of all traffi fatalities were aloholrelated, under 3% of all serious traffi injuries were aloholrelated, and 1.9% of all injuries were alohol-related. Table 2 ompares the average annual injuries rates per billion kilometers between January 1998 and May 2002 (pre-law) with the average rates between June 2002 and Deember 2004 (postlaw). AID traffi fatalities fell by 38%, and AID traffi injuries fell by 33%; all traffi fatalities fell by 14%. For severe traffi injuries, AID traffi injuries, AID severe traffi injuries, and AID traffi fatalities per billion kilometers driven, there were redutions after the new traffi law was enated in June 2002. 20 Injury Prevention 2008;14:19 23. doi:10.1136/ip.2007.015719
Figure 1 Observed and predited alohol-impaired driving traffi fatalities (per billion kilometers driven) using a stepwise autoregression model, Japan, 1998 2004. There was no evidene of a redution for all traffi injuries (+6%) during the same periods. In segmented regression analyses with adjustment for baseline trends, seasonality, and autoorrelation, we found a signifiant hange in the level of AID traffi fatalities (fig 1), AID severe injuries (not shown), and all AID traffi injuries per billion kilometers driven (fig 2). The baseline rate was signifiantly (p,0.01) redued by 20%, 23% and 32% for AID traffi fatalities, severe traffi injuries and all traffi injuries, respetively. We found no signifiant hange in the slope for time trend before and after the intervention for AID traffi fatalities (p = 0.14) (fig 1) and severe traffi injury (p = 0.07) (not shown). However, we observed a qualitative hange in the slope (0.1 to 20.2) for total AID injuries (p,0.01) (fig 2). For all traffi fatality rates, we observed a 4% redution in the baseline rates (p = 0.06), but the hange in the slope (20.024 to 20.017) was not signifiant (p = 0.51) (fig 3). We found 4% redutions in the baseline rates for both severe traffi injuries (p = 0.05) (not shown) and all traffi injuries (p,0.01) (fig 4). The slope in the trend hanged from 0.06 to 20.16 for severe traffi injuries (p,0.01) and from 3.89 to 1.64 for all traffi injuries (p,0.01). Figure 2 Observed and predited alohol-impaired driving traffi injuries (per billion kilometers driven) using a stepwise autoregression model, Japan, 1998 2004. Original artile DISCUSSION Following the 2002 new road traffi law, whih inreased the penalties for AID, Japan experiened a signifiant and substantial redution in AID traffi problems: fatalities, severe injuries, and all injuries. These results were onsistent aross all traffi injuries (alohol-related plus non-alohol-related), severe traffi injuries, and also total traffi fatalities (p = 0.06). In Japan, AID traffi fatalities had been dereasing sine the early 1990s, and it appears that the law had an effet over and above this trend. Hitosugi et al 12 and Imai 13 reported a derease in traffi fatalities after the new law, but did not measure the effet against the already delining traffi fatalities sine 1993 in Japan, nor were trends standardized by kilometers driven. This study used time series analysis to demonstrate the effet of the new traffi law, aounting for both the trend and driving exposure. 14 16 Alohol remains a major ontributor to traffi death throughout the world, although muh progress has been made. Liense suspension, illegal and administrative laws per se, seletive and regular enforement patrols, and sobriety hekpoints have been effetive in reduing the harm aused by AID. 2 17 The introdution of a legal BAC limit has been effetive in the UK, Canada, The Netherlands, and Japan. 2 In the USA, Injury Prevention 2008;14:19 23. doi:10.1136/ip.2007.015719 21
Original artile Figure 3 Observed and predited all traffi fatalities (per billion kilometers driven) using a stepwise autoregression model, Japan, 1998 2004. the introdution of the 0.08% BAC Law is ontroversial, but appears to have redued AID; however, its effetiveness in reduing traffi injuries and fatalities has been debated. 14 15 An earlier 1970 law in Japan setting the legal BAC at 0.05% appeared to redue traffi fatalities; however, it was not 18 19 statistially evaluated. Japan publiized the new law in a variety of ways, inluding mass media ampaigns enouraging publi transportation when drinking instead of driving. Not only the NPA but all the governmental agenies atively worked to announe the introdution of the road traffi law for AID. They also onduted a media ampaign. The offiial advertisement for this ampaign is memorable. In it, a drunken man tries to go home in his ar. The bartender requests that he pay $4250 (equal to the drinking driving penalty in Japan), whih is more than the average monthly salary of most workers. This advertisement gave viewers a strong impression of how severe the penalty would be. In this study, we find that the redution in AID traffi injuries and fatalities in Japan, 2002, was a suessful publi health intervention, and believe the lessons learned from this poliy hange will be useful for other ountries trying to ope with AID problems. Figure 4 Observed and predited all traffi injuries (per billion kilometers driven) using a stepwise autoregression model, Japan, 1998 2004. We measured the effet of the new road traffi law on the derease in vitims of traffi rashes in a time series analysis. Time series analysis refers to a large series of observations made on the same variables repeatedly over time, and it is widely used in eonomi, physial, demographi, and publi health studies to evaluate and predit the trend of a single harateristi in 20 21 different time intervals. Some weaknesses of simple pre post omparison studies are that long-term trends, slow or delayed effets, regression to the mean, and the orrelated data are not properly onsidered. Time series analysis an overome these weaknesses, and haraterize the hange in trend assoiated with the intervention more learly. The new road traffi law was assoiated with signifiant redutions in the baseline for all traffi injuries, all severe traffi injuries, AID traffi injuries, AID severe traffi injuries, and AID traffi fatalities per billion kilometers driven, and the total redution in all traffi fatalities was borderline signifiant (p = 0.06). Only 15% of traffi fatalities in Japan were aloholrelated before the law, whih has had its main impat on alohol-related driving. 4 12 After the law, alohol-related fatalities fell to 11% (table 2). The primary effet of the new traffi law was to hange the behavior of people who intended to drive 22 Injury Prevention 2008;14:19 23. doi:10.1136/ip.2007.015719
Key points Eah year, more than 7000 people die as the result of road traffi injuries in Japan. In June 2002, the Japanese government enated a new road traffi law to redue AID. By applying time series analysis, we determined that baseline rates of all traffi injuries, severe traffi injuries, AID traffi injuries, AID severe traffi injuries, and AID traffi fatalities per billion kilometers driven delined signifiantly after the new traffi law. The Japanese poliy appears to have been a suessful publi health measure. a motor vehile after drinking, whih affeted AID injuries but not the overall traffi fatality rate. This study has various limitations. Firstly, the lowering of the definition of AID may have inreased the number of aloholrelated traffi rashes after the law, whih may have aused bias in the estimation of alohol-related traffi events over time. For example, before the new law, drivers with a BAC of 0.04% would not have been ounted as AID, whereas after the new traffi law was implemented, they would be. Unfortunately, the numeri results of breath or blood alohol testing were not available, nor was there information on the extent of missing data on breath or blood alohol tests in lassifying aloholrelated motor vehile events in Japan. Use of a onsistent definition of AID throughout the period of observation was not possible beause of hanges in the law. However, the fat that signifiant redutions in AID events were seen after the law was enated, even though the definition beame more inlusive, suggests that AID redutions were onsiderable, and perhaps even larger than our estimates. Seondly, data were not available on the proportion of drivers in non-fatal, serious injury, and fatal rashes who were tested for alohol. In addition, the NPA does not have a standard protool stipulating the aeptable time lag between ollision and breath testing, and therefore it is unknown. Thirdly, it was not possible to systematially ontrol for external onfounders suh as other new traffi laws, ampaigns, and overall alohol onsumption. However, to our knowledge no other major hanges ourred between 1998 and 2003 that may have had a signifiant effet on traffi injuries. The ommitment of law enforement in the pre- and post-law period appears to be the same. A study showed that 140 460 people were arrested for traffi violation between January and June 2002 (pre-law), and 133 937 were arrested between January and June 2003 (post-law). 4 7 The differene was not statistially signifiant. Fourthly, it was not possible to stratify the data aording to age groups, and therefore information about high-risk groups for traffi rashes, inluding alohol-related ones, was not available from the publi domain datasets used in this study. More detailed data will be neessary to examine the impat of the law on different age groups. Our study finds that implementation of the new Japanese road traffi law was followed by a substantial deline in fatal and non-fatal motor vehile injuries assoiated with AID. The Japanese poliy appears to have been a very suessful publi health measure. Original artile CONCLUSION Despite some limitations, this study, by applying time series analysis, shows that the new Japanese road traffi law implemented in June 2002 had an effet on delining traffi injuries and fatalities. This suess was obtained in several ways: dereasing the permissible blood alohol level, inreasing penalties, and an ative media ampaign. By desribing the suess of the AID intervention in Japan, the findings of this study may be helpful for other ountries oping with the problems of AID. IMPLICATIONS FOR PREVENTION National poliies an have a signifiant effet in reduing AID by using multidisiplinary approahes that inlude enforement and eduational ampaigns targeting driver behavior. Aknowledgements: We thank Dr Yeong-Su Ju, Takemi Program in International Health, Harvard Shool of Publi Health, for supporting the data analysis, and Danielle Stokley, Department of Environmental Health, Harvard Shool of Publi Health, for assistane with editing the manusript. Funding: MP reeived partial support for this work in the form of grant No T42 OH 008416-02, from the US National Institute for Oupational Safety and Health (NIOSH). Competing interests: None. REFERENCES 1. National Highway Traffi Safety Administration. State laws and praties for BAC testing and reporting drivers involved in fatal rashes. Washington, DC: US Department of Transportation, 2004. Report DOT HS 809756. 2. World Health Organization. World report on road traffi injury prevention. Geneva: WHO, 2004. 3. Mann RE, Madonald S, Stoduto LG, et al. The effets of introduing or lowering legal per se blood alohol limits for driving: an international review. Aid Anal Prev 2001;33:569 83. 4. National Polie Ageny. Polie report about fatal traffi rashes in 2003. Tokyo: NPA, 2004 (in Japanese). 5. Shimizu S. Tougher penalties for drink driving in Japan. Addition 2001;96:1866. 6. Nakahara S, Wakai S. Underreporting of traffi injuries involving hildren in Japan. Inj Prev 2001;7:212 14. 7. Institute for Traffi Aident Researh and Data Analysis, Japan. http:// www.itarda.or.jp/english/eg_home.html (aessed 12 De 2007). 8. Ministry of Land Infrastruture and Transport, Japan. http://toukei.mlit.go.jp/ jidousya/jidousya.html (aessed 12 De 2007). 9. Gillings D, Maku D, Siegel E. Analysis of interrupted time series mortality trends: an example to evaluate regionalized perinatal are. Am J Publi Health 1981;71:38 46. 10. Kennedy P. Autoorrelated disturbanes. In: A guide to eonometris. 4th edn. Cambridge, MA: MIT Press, 1998:123 5. 11. University of California. SAS/ETS user s guide. The AUTOREG proedure.http:// www.u.edu/sashtml/ets/hap8/index.htm (aessed 12 De 2008). 12. Hitosugi M, Sorimahi Y, Kurosawa A, et al. Risk of death due to alohol-impaired driving in Japan. Lanet 2003;361:1132. 13. Imai H. The new traffi law and redution of alohol related fatal rashes in Japan. Inj Prev 2003;9:382. 14. Tippetts AS, Voas RB, Fell JC. A meta-analysis of.08 BAC laws in 19 jurisditions in the United States. Aid Anal Prev 2005;37:149 61. 15. Hingson R, Heeren T, Winter M. Lowering state legal blood alohol limits to 0.08%: the effets on fatal motor vehile rashes. Am J Publi Health 1996;86:1297 9. 16. Foss RD, Stewart JR, Reinfurt DD. Evaluation of the effets of North Carolina s 0.08% BAC law. Aid Anal Prev 2001;33:507 17. 17. Wagenaar AC, Zobek TS, William GD, et al. Methods used in studies of drink-drive ontrol efforts: a meta-analysis of the literature from 1960 to 1991. Aid Anal Prev 1995;2:307 16. 18. Deshapriya EB, Iwase N. Impat of the 1970 legal BAC 0.05 mg% limit legislation on drunk-driver-involved traffi fatalities, aidents, and DWI in Japan. Subst Use Misuse 1998;33:2757 88. 19. Deshapriya EB, Iwase N. Are lower legal blood alohol limits and a ombination of santions desirable in reduing drunken driver-involved traffi fatalities and traffi aidents? Aid Anal Prev 1996;28:721 31. 20. Shadish WR, Cook TD, Campbell DT. Experimental and quasi-experimental design for generalized ausal inferene. US: Houghton Mifflin, 2002. 21. Wanger AK, Soumerai SB, Zhang F. Segmented regression analysis of interrupted time series studies in mediation use researh. J Clin Pharm Ther 2002;27:299 309. Injury Prevention 2008;14:19 23. doi:10.1136/ip.2007.015719 23