Table 1 Studies included based on eligibility criteria: studies with no increases in enforcement activities or with statistical models to account for those increases

Whittam 2006 [18]

Objective: Assess the potential impact of public-service assessments on young drivers between the ages of 16 and 19 years

All crashes among 16–19 year olds

Net change in serious-injury crashes among 16–19 year olds: 18.3 % decrease

4.5 months

 CITS, 1994–1999

 Intervention period: Aug 15–Dec 31, 1996 (excluding Oct)

Intervention site:

 Intervention sites: Kingsport, Johnson City, and Bristol, Tennessee

• 21.6 % reduction during the intervention period (p < 0.05)

 Comparator sites: Hamilton County

Mass media: Paid television and radio announcements, billboard display

Comparator site:

• 3.2 % increase in crashes (p = 0.61)

Net difference:

• 24.8 %

Serious-injury crashes among 16–19 year olds

Intervention site:

• 16.4 % decrease (p = 0.19)

Comparator site:

• 1.9 % increase (p = 0.17)

Net difference: 18.3 %

Murry 1993 [19]

Objective: Evaluate an anti-drinking and driving advertising campaign targeting 15–24 year olds.

Nighttime fatal and incapacitating accidents for 15–24 yo males and females (intervention site: −7.14 %, comparator: +11.8 %, p = 0.06)

Net change −18.9 % (p = 0.05)

6 months

 Jan 1983–Sept 1987, monthly CITS

 Intervention: Wichita, Kansas, USA

 Comparator: Omaha, Nebraska, USA

Mass media: 6-month paid media schedule using television, radio, newspapers, and billboards

Newstead 1995 [20]

Objective: Evaluate various safety measures implemented starting Sept 1989 in Victoria, Australia.

Contribution of drink-driving publicity in reducing nighttime serious casualty crashes: approximately 14 % (average of 1990–1992)

Reduction of nighttime serious casualty crashes in Victoria from 1990 to 1992 was 14 %

3 years

 1983–1992

 ITS

 Intervention: Victoria, Australia

 Comparator: None

Mass Media:

TV advertising, Dec 1989 to Dec 1992, radio, press, outdoor advertising, Sky Channel and cinema

Statistically significant in Victoria (p < 0.05), but not in Melbourne crashes (p = 0.07)

Enforcement: Random breath testing, lowering of freeway speed limit, speed cameras

Tay 2002 [21]

Objective: Evaluate New Zealand’s Supplementary Road Safety Package initiated by Land Transport Safety Authority in 1995

Estimated impact of the advertising campaign on the number of fatal crashes using regression model:

Estimated impact of advertising campaign on the number of fatal crashing has no impact on the target population (male 15–34 years old)

2 years

 ITS, 1988–1996 (108 observations)

 Intervention site: New Zealand

 Comparator site: None

• Male drivers between 35 and 54: 29.91 % decrease

• Female drivers between 15 and 24: 40.21 %

• Female drivers between 25 and 34: 70.04 %

Media campaign: TV, mainly targeting

• No impact on young male drivers (15–34)

18–24 year olds

Enforcement:

Estimated impact of the program before and after implementation of the campaign:

Speed cameras, advanced speed detectors, compulsory breath testing

• Male drivers: −32.9 % (15–24yo) to +4.7 % (55 years and older)

Female drivers: −56.8 % (25–34 %) to −26.7 % (55 years and older)

Jones 2005 [22]

Objective: Evaluate “Smart Roads” program in Pueblo, Colorado aimed at drivers aged 21–34.

Nighttime injury crashes decreased by 39 % in the intervention counties, whereas it increased by 3.3 % in the control counties (p < 0.0001)

Nighttime single-vehicle crashes: net change 28.8 %

4 years

 Before: 1998 to 1999

 After: 2000 to 2001

 CBA

 Intervention group: Pueblo county (intervention site) plus eight other low-population surrounding counties

Mass media:

Television, radio, and newspaper advertisements, billboards, bumper stickers, bus station banners, other collaterals)

Nighttime single-vehicle crashes decreased by 24.8 % in the intervention counties, whereas there was a 4.0 % increase in the control counties (p = 0.01)

 Comparison: all other counties in Colorado

Workplace initiative education program.

Epperlein 1987 [29]

Objective: Evaluate the effect of crackdown on drinking drivers in Arizona

Impact estimates of the anti-drunk-driving publicity campaigns of March, 1982

Nighttime fatal crashes (net change): −16.2 %

22 months

 March 1972-Dec 1983 ITS

 Intervention site: Arizona, USA

Mass media:

• Nighttime fatal crashes −26.8 % (pre-intervention mean/month. 724)

Television, print, and radio advertisements, billboards, posters, bumper stickers (March 1982)

 Comparator site: None (daytime crashes and crashes with no identified drinking drivers used for comparison)

• Daytime fatal crashes −10.6 % (pre-intervention mean/month. 1633)

Enforcement:

Stricter DWI legislation Increasing the minimum drinking age (August 1982)

Net change: −16.2 %

• Drinking drivers in crashes −14.0 % (pre-intervention mean/month. 1036)

• Non-drinking drivers in crashes −0.8 % (pre-intervention mean/month. 11345)

Net change: 13.2 %

Zampetti 2013 [34]

Objective: To verify the effect of intensive vs. basic road safety education programs on the incidence and severity of nonfatal road injuries (NFRTI)

The number of NFRTI

Difference in incidence of NFRTI in the basic site: −0.04 % (p = 0.05)

5 years

 Before: Jun–Aug 2003

• Before: 907,

 After: Jun–Aug 2008

After: 755

 CBA

Incidence of injuries in the basic campaigns (8 municipalities)

 Intervention period: 2003–2008

 Intervention sites: 20 municipalities in the Local Health Authority 1 (LHA1) area in Campania, Italy

Publicity campaigns: Billposting on public transport, bus stops, train stations, in bars and meeting places. Dispatch of brochures, pamphlets, and posters

• Difference in incidence of injuries −0.4 per 1,000 (2003 (before) 1.1, 2008 (after) 0.7)

 No comparator site

Mass media: press conferences, articles in local papers, radio/television interviews, and the LHA1 web site

• Incidence of injuries in the intensive campaigns (12 municipalities)

Sites for intensified approach (12 out of 20 municipalities):

• Difference −0.5 per 1,000; p < 0.001

School campaigns and community conferences, 1-day conference at the end of school year

Worden 1975 (Elder) [35]

Objective: Evaluate Vermont public education campaign on alcohol and highway safety

The proportion of “high-risk” male drivers (those who report consuming three or more drinks at least once a week) above 0.05 g/dL BAC:

Drivers above 0.05 g/dL BAC: −158 %

24 months

 May 1972–May 1974

 CBA

 Intervention site: Vermont

Mass media: Radio, TV, drive-in theater spots.

Fatal crashes: 0 %

 Comparison site: counties with no intervention

• At mid-campaign (May, 1973) decreased 37 % from a baseline of 10 of 48 drivers to 9 of 69 (95 % CI: −72 % ~ +42 %; net change = −158 %)

Enforcement: Stayed high throughout the study period

• Immediately following the campaign (May, 1974) decreased 67 % (95 % CI: −88 % ~ −7 %; net change −111 %)

The proportion of had-been-drinking to total fatal crashes decreased 6 % from a baseline of 9 of 20 to 8 of 19 (95 % CI: −54 % ~ +91 %; net change 0 %)

*Very small sample sizes

Cameron 1998 (Elder) [36]

Objective: Evaluation of the first two years of the New Zealand Supplementary Road Safety Package that was introduced in 1995/1996 (supplements CBT and speed camera programs introduced in 1993)

In 1996–1997, campaign estimated to result in:

Injury crashes

24 months

 Jan 1990–June 1997, quarterly

Arm 1 (Urban): −7 %

 CITS

 Intervention: New Zealand (crashes during high alcohol consumption hours)

• A 33 % decrease in urban high alcohol hour serious injury crashes (95 % CI: −40 % ~ −25 %; net change = −7 %)

Arm 2 (Rural): −18 %

 Comparator: New Zealand (crashes during low alcohol consumption hours)

Mass media: primarily TV advertising campaigns

• A 32 % decrease in rural high alcohol hour serious injury crashes (95 % CI: −41 % ~ −22 %; net change = −18 %)

Enforcement: Sobriety checkpoint

In 1995–1996, campaign estimated to result in:

• A 16 % decrease in urban high alcohol hour serious injury crashes (95 % CI: −24 ~ −6 %; net change = −2 %)

A 6 % decrease in rural high alcohol hour serious injury crashes (95 % CI: −18 % ~ −7 %; net change = −5 %)