Abstract
Objectives. We assessed the 5-year, nationwide impact on road deaths of the raise in the speed limit (November 1, 1993) on 3 major interurban highways in Israel from 90 to 100 kph.
Methods. We compared before–after trends in deaths as well as case fatality—an outcome independent of exposure (defined as vehicle-kilometers traveled).
Results. After the raise, speeds rose by 4.5%–9.1%. Over 5 years, there was a sustained increase in deaths (15%) and case fatality rates (38%) on all interurban roads. Corresponding increases in deaths (13%) and case fatality (24%) on urban roads indicated “speed spillover.”
Conclusions. Immediate increases in case fatality predicted and tracked the sustained increase in deaths from increased speeds of impact. Newtonian fourth power models predicted the effects of “small” increases in speed on large rises in case fatality rates. Countermeasures and congestion reduced the impact on deaths and case-fatality rates by more than half.
On November 1, 1993, the government of Israel increased the enforced speed limit for all vehicles, including trucks, from 90 to 100 kilometers per hour (55.9 to 62.1 mph) on segments (115 km, or 71.4 miles) of 3 major interurban highways connecting its 4 major cities: Tel Aviv, Jerusalem, Haifa, and Beersheba. The government made major improvements on these highways and many other roads and declared the increased speed limit a 1-year “experiment.”1 Simultaneously, it mandated the use of rear seat belts and daytime running lights.
Lower travel speeds and fewer deaths usually follow lowered speed limits.2,3 Higher travel speeds and more deaths follow increased speed limits.4–10 Recent data demonstrate a 17% increase in deaths after a 4% increase in speeds on US interstate highways.11 High-speed driving on highways induces speed adaptation (a situation in which vehicle speed is influenced by the speed and duration of recent travel in the vehicle) on connecting interurban roads, and even urban roads. This so-called spillover effect may persist for 5 to 6 years.12–15 Yet there is still worldwide controversy over the impact of increased speed limits.16–18 One view holds that increased speed limits not only shorten travel time but also are protective when increased vehicle mileage is used to correct for increases in death tolls.16 The US Centers for Disease Control and Prevention, despite 41 967 road deaths in 1997, has not cited higher speed limits as contributing to the high number of deaths.19 The British government, by contrast, is committed to strategies to reduce speeds.20
Israel, with a size of 21 501 km2, provides an ideal setting for observing the effects of speed limits. Israel has a fairly modern car fleet and roads and relatively low drunkdriving rates and is isolated from traffic from neighboring states.21 All its highways are interurban. The 3 highways on which the enforced speed limit was raised serve as the major conduits of Israel’s interurban traffic.
We examined the suddenness, size, distribution, and persistence of nationwide changes in death and injury tolls after the increase in the speed limit on these highways, with specific attention to speed spillover and its nationwide effect on road deaths. We examined the utility of 2 empirically derived models that demonstrate the relations between speed and fatality risks; the models are based on Newtonian physics. The first model demonstrates that case-fatality rates (CFRs) vary to the fourth power of the velocity at vehicular impact with both unbelted22 and belted23 drivers; the second model demonstrates that the number of crashes, injuries, and deaths varies with the first, second, and fourth power, respectively, of increases in average traffic speeds.24,25 We also examined whether an increase in the CFR, a crash-phase outcome (those variables which influence survival in the event of a crash, such as speed of impact, seat belt use, and trauma care) independent of exposure (billion vehicle-kilometers of travel [bvkmt]),26 predicted and tracked sustained trends in increased road death tolls. Finally, we assessed the degree to which protective countermeasures and increased traffic congestion offset and conceal the full impact of increases in travel speeds on road deaths.
Data on speed trends on the 3 high-speed highways (Tel Aviv to Jerusalem [highway 1]; Tel Aviv to Haifa [highway 2]; Tel Aviv to Beersheba [highway 4]) came from sporadic roadside daytime monitoring in the years 1971–1994 using roadside radar and laser cameras.27–30
We collected data on road deaths (up to 30 days after crash injury), serious injuries (hospitalized more than 24 hours), and light injuries (not hospitalized, or hospitalized less than 24 hours), and exposure—as measured by billion vehicle-kilometers of travel—from the Central Bureau of Statistics.27,29 We also used a surrogate measure for the CFR—the proportion killed among all seriously injured (hereafter CFRS, for “CFR surrogate”)—to avoid biases from transient underreporting of light injuries, which the Central Bureau of Statistics estimated to be of the order of 10%.
We carried out a 1-year comparison of deaths and case-fatality rates (CFRs and CFRSs), before and after the increase to 100 kph on November 1, 1993, for high-speed roads, other interurban roads, and all urban roads, and for major crash types and driver subgroups.
We analyzed changes in death rates and CFRs between 3 years before and 5 years after the increase in the speed limit with the Student t test and cumulative summing. This method involves subtracting the differences between monthly totals for deaths from the overall mean of the 3-year control period derived from a baseline of monthly average death totals and testing these differences with simple t tests.31 We then estimated the specific effect of increased speed limits without countermeasures and congestion by comparing the observed change in the number of deaths per year with that attributable (K(ATTRIB)) specifically to the change in CFRs. We made these estimates using the formula
(1)
where K(B) represents persons killed per year 3 years before, and CFRS(B) and CFRS(A) are the proportion of those killed among all those seriously injured before and after the speed limit change, respectively.19 Using models developed by Evans,3 we also estimated the average change in speeds (V) of travel and crash impact, as well as the change in number of deaths per each 1% change in speed, using algebraic fourth-root models in which
(2)
where K(A) and K(B) are persons killed per year 5 years after and before the increase, respectively, and V(A) and V(B) are average speeds on roads after and before the speed increase, respectively.
(A separate autoregressive integrated moving averages analysis of observed–expected ratios for deaths, deaths per billion vehicle-kilometers of travel, and CFRs on interurban and urban roads of the first year after the increased speed limit is available from the authors on request. This analysis used monthly totals for these parameters going back 13 years as a baseline for predicting expected results during the first 6 and subsequent 8 months after the increase in the speed limit.)
Sporadic monitoring from 1971 to 1994 (Table 1) indicated that right after the increase in the speed limit, travel speeds on high-speed roads increased by 4.5% on the slow lane of the Tel Aviv–Haifa road (highway 2), by 9.1% on the fast lane of the Tel Aviv–Jerusalem road (highway 1), and even more on a newly widened stretch of a major connecting road off the Tel Aviv–Jerusalem road (Table 1, highway 40)—all compared with the year before. Other data showed that speeds rose on all 3 highways after the speed limits were raised, and that the mean estimated increase in speeds on the high-speed roads later fell back to a net increase of approximately 4% (range: −4% to 13%) in 1995.27–30
A sudden increase in monthly nationwide death tolls and CFRSs followed the increase in the speed limit on November 1, 1993. The first month after speed limits were increased, deaths (n = 61) increased 32.6% from October (n = 46). Interurban deaths (n = 38) and the CFRS (26.4%) were the highest since November 1990.
In the first year after the increase in the speed limit, deaths increased by 24%, from 257 to 319, and CFRSs increased by 29.5% on all interurban roads combined, compared with corresponding increases of 3%—from 230 to 236—and less than 10%, respectively, on urban roads (Table 2). On newly widened segments of the 3 high-speed highways (and extensions) with 100-kph limits (roads 1, 2, and 3), deaths increased by 67%, from 21 to 35, a reversal of downward trends from 1990, and CFRSs increased by 50%. Even so, 48 (77%) of the 62 added deaths on interurban roads in the 12 months after the increase in the speed limit occurred not on the 3 high-speed roads but on other interurban connecting roads. A separate autoregressive integrated moving averages analysis29 verified that the abrupt, large jump in deaths in the first 6 months after the increase in the speed limit was especially marked on interurban roads and was directly attributable to increases in CFRs and offset the long-term drops in deaths per billion vehiclekilometers of travel. Before and after the increase in the speed limit, 90% or more of those killed in truck crashes were occupants of passenger cars.32 After the increase in the speed limit, much—60%—of the increase in the nationwide road death toll came from large increases in deaths from truck crashes, mainly on interurban roads. Table 2 also shows increased CFRs and deaths in 1-vehicle and motorcycle crashes nationwide and decreased deaths among pedestrians and cyclists. Despite retention of the 90-kph limit by the military, reported deaths involving soldiers—both drivers and occupants—increased 106% (from 15 to 31), and reported CFRs increased 30%.
In the first year following the increase in the speed limit to 100 kph, interurban traffic increased 5% from 11.4 to 12.0 bvkmt, whereas urban traffic increased much more—from 13.2 to 15.5 bvkmt (17%). The number of road deaths per year was weakly correlated (r = 0.15) with billion vehicle-kilometers of travel annually from 1963 to 1995, but negatively correlated with the number of licensed drivers (r = −0.46) from 1970 to 1995. These results rule out more vehicle traffic and drivers as plausible explanations for the sudden large increase in deaths.
The government introduced several countermeasures, including laws requiring rear seat belts and daytime running lights (both mandated on November 1, 1993), more capital investment in upgrading old roads, building of new roads, midline concrete barriers and flyovers, and nighttime lighting. Hospital trauma services increased from 1 to 5, and police enforcement, measured by issuance of speeding tickets, increased approximately fourfold in the years 1994 to 1998 (Cdr E. Efrat, Traffic Police Division, written communication, May 2001).
There were no changes in before–after ratios of billion vehicle-kilometers of travel for trucks to all vehicles (27.3%:27.5%), drivers aged 19 to 24 years–all drivers (17.2%:17.5%), or fuel costs or alcohol sales.
During the entire 5-year period following the increase in the speed limit (November 1, 1993, to October 31, 1998), there were substantial drops in the number of persons reported with serious injuries (Table 3). In July 1995, the monthly death toll (n = 62) peaked. In the third year after the increase in the speed limit, the death toll on interurban roads began to fall from a peak in 1994–1995 (n = 327), corresponding to indications of decreases in average interurban speeds (Table 1) but continued to increase on urban roads. During the entire 5-year period, there were mean increases of 39.2 (15%) and 27.2 (13%) deaths per year on interurban and urban roads, respectively. The corresponding increases in CFRs, which tracked trends in speeds of impact, were much greater: 38% (from 12.5% to 17.3%), and 24% (from 7.9% to 9.8%) (Table 3). Using validated Newtonian models,3 we estimated that without countermeasures and congestion, increased speeds of travel and impact would have resulted in increases of 100.2, and 51.1 deaths per year on interurban and urban roads, respectively. Countermeasures and congestion would have resulted, if not for the increased speed limits, in corresponding reductions of 61 and 23.9 deaths per year. The 5-year nationwide increase in deaths per year (n = 151.3) expected from increased speeds of impact greatly exceeded the observed increase (n = 66.4) in total deaths per year (Table 4).
After the increase in the speed limit from 90 to 100 kph, sporadic data suggested that travel speeds increased on Israel’s 3 major highways and other roads, later falling back somewhat. In the first year after the increase in the speed limit, there were abrupt increases in deaths from increases in travel speeds on the 3 major highways and spillover of these effects to other urban and interurban roads. More than three quarters of the first-year increase in deaths (n = 62) on interurban roads occurred from a systemwide spillover effect from high-speed roads on which the speed limits were legally increased to other interurban roads. All these findings state the case for systemwide increases in real travel speeds.
Our observations state the case for a direct cause-and-effect relation between the increase in the speed limit and the increase in the death toll. First, the step function increase in deaths coincided with the increase in the speed limit. Second, the increase in deaths is attributable specifically to the increase in CFRs—in all vehicle and crash types—a finding that suggests increased speeds of impact. Third, the increases in deaths and CFRs on the high-speed roads were proportionately much greater than on other interurban roads and on urban roads. Fourth, the degree of increase in CFRs and deaths matched that expected from the reported increases in travel speeds based on the validated models. Fifth, time trends in all the modifiers and confounders (enforcement, seat belts, trauma care) should have resulted in reductions, not increases, in death tolls.
We confirmed the utility and validity of predictive Newtonian models in which deaths and CFRs increase in proportion to the fourth power of increases in speeds of travel of all vehicles and impact speeds of crashing vehicles, respectively. In the first year after the 100-kph speed limit was implemented, the observed increase in average travel speeds of 4% to 4.5%, based on sporadic measurements, accords with the increase of 5.5% in average travel speeds predicted from the observed increase of 24% in deaths on all roads.3,23–25 Increases in CFRs of 50% on highways and 26% on other interurban roads in the first year imply that average increases in speeds of impact on these roads were of the order of 11% and 5.6%, respectively. During the entire 5-year period after implementation of the 100-kph speed limit, similar calculations suggest that impact speeds of crashing vehicles increased on interurban and urban roads by some 8.3% and 5.5%, respectively. These increases exceeded the estimated increases of 3.6% and 3.1% in average speeds of travel for all vehicles (Table 4). These calculations imply that each 1% increase in average speeds of travel and impact resulted in increases of approximately 11 deaths per year on interurban roads, and 9 deaths per year on urban roads.
The large increases in the number of deaths and CFRs were seen in crashes involving trucks, motorcycles, single vehicles, and soldiers (both on and off-duty) but not pedestrians and bicyclists. Other evidence suggested that the fatal crash risk of truck drivers from higher speeds increased with longer hours, irregular shifts, and incentive premiums.32,33
The fact that increases in deaths from increases in CFRs occurred in all subgroups except pedestrians and bicyclists rules out changes in the case mix of crash types as the reason for the increase in total CFRs of all crash groups combined. Soldiers—both drivers and occupants—were at increased risk, despite the military’s retention of the 90-kph limit during working hours. For pedestrians and bicyclists, the drop in death tolls may be explained by a trend seen in the United Kingdom, where there has been a reported decrease in walking and cycling on interurban roads.34 The fact that CFRs from 1-vehicle crashes—which are not influenced by vehicle–vehicle interactions—were not less than those from other crash types undermines the claim that increased speed variance35 and not increased speed is the real cause for the increase in deaths.
The observed increase in deaths per year following the increase in the speed limit to 100 kph substantially underestimated the increase in deaths directly attributable to the increase in CFRs. The exponential effect of “small” increases in speed and speed spillover on nationwide increases in CFRs over the next 5 years more than offset the decreases in death risks per vehicle-traveled from protective countermeasures, as well as congestion in urban areas. The persistence of high CFRs indicates that increased speeds of impact were negating the protective effects of newly widened roads, improved lighting, cloverleafs, air bags, rear seat belt laws, more speed enforcement and changes in trauma care, and other countermeasures, as well as increased congestion. Without protective countermeasures and increased traffic congestion, there would have been many more deaths, and without the increase in the speed limit, there would have been many fewer deaths (Table 4).
In Israel, increases in traffic congestion and road safety countermeasures have produced a long-term strong inverse relation between deaths per billion vehicle-kilometers of travel and both the number of vehicles (r = −0.88) and vehicles per population (r = −0.85).19,27 In the United States, car occupant deaths dropped 11% between 1975 and 1997, despite a ninefold increase in cars.36 But from 1992 to 1997, by which time US states were raising speed limits, deaths—and even deaths per billion vehicle-kilometers of travel—did not drop.19 By contrast, in the years 1991 to 1998, road deaths per year fell by 25.2% in the United Kingdom with correspondingly larger decreases in deaths per billion vehicle-kilometers of travel.37
Risks for deaths per billion vehicle-kilometers of travel have always decreased with increases in billion vehiclekilometers of travel38 (“the soccer field is tilted downwards”19); therefore, before and after differences following increased speed limits will underestimate the increases in death tolls from increased speed limits over many years. Studies of the long-term impact of increased speed limits that “correct” for increases in billion vehicle-kilometers of travel may underestimate the full direct impact of increased speed limits and travel speeds on road deaths. Long-term follow-up in Washington State, for example, showed that when corrected for exposure, a 25% increase in deaths on interstates was reduced to a 10% increase.39 These studies ignore the role of countermeasures and increased congestion in producing the falling trends in deaths per billion vehicle-kilometers of travel in urban areas. Congestion from increased billion vehicle-kilometers of travel during the so-called rush hours offsets the effects of higher speeds during other hours, notably during nighttime.
Evans has shown how relations between increases (and decreases) in speed and the increases (and decreases) in death tolls are direct, obey algebraically defined laws derived from Newtonian physics, and are reversible.23 Based on these premises, we suggest that a reduction by 10% in the average speeds of impact during the period we studied would have prevented 121 (or 40%) of the 300 interurban deaths per year, and 85 (or 35.4%) of the 240 urban deaths per year (Table 4). Support for this inference comes from the observation in the United Kingdom that there are even bigger decreases in death tolls—up to 70%—with reductions in speeds from massive use of roadside speed cameras.40 Substantial reductions in deaths, injuries, and crashes from the use of speed cameras state the case for their use.41
We suggest, however, that achieving sustainable major reductions in road death tolls requires not only lower speed limits and increased detection and deterrence of high speeds, but also lower design speeds for cars, and a downward shift in speed distributions, in keeping with the principle of treating sick populations, not just sick individuals.42 The public health stakes involved in applying this principle are enormous, given that globally, there are now more than 1 170 000 road deaths per year43 and more than 40 000 deaths per year in the United States alone. We predicted increases in death tolls from new highways and spillover roads with even higher design speeds and speed limits, a trend now seen in many rapidly motorizing countries.44 Elsewhere, we have suggested using the CFR to track the direct long-term impact of increased travel speeds on death tolls in the United States.19 The CFR, the outcome of concern, is a parameter based on a universe; because it is extremely sensitive to small changes in speed well within the range of sampling and measurement errors, it paradoxically may be a more valid indicator of speed trends than sporadic speed measurements themselves.
Elsewhere, we have addressed the ethical and scientific lapses underlying the decision to increase the speed limit.45,46 In retrospect, the sentinel increases in travel speeds on the study’s highways, CFRs, and deaths in the very first months of the 100-kph “experiment” predicted its subsequent 5-year nationwide impact and stated the case for its cancellation.
In 2002, despite an increase in deaths to 540 from 476 the year before, there were renewed pressures to increase the speed limit still further, to 110 kph–120 kph.
Note. NA = not available. aWeighted mean speed traveled: (sum of number of vehicles per siting multiplied by mean speed for individual siting) divided by total number of vehicles. bWeighted 90th percentile for speed traveled: (sum of number of vehicles per siting multiplied by 90th percentile speed for individual siting) divided by total number of vehicles. cRight lane is slow lane; left lane is fast lane. dFrom 1971 to 1990, mean speeds (and number of sitings) were as follows: 1971: 92.3 (1537); 1975: 97.4 (1338); 1976: 97.3 (1446); 1977: 94.0 (1758); 1980: 88.3 (1860); 1981: 87.6 (1866), 1983: 95.9 (1974); 1988: 103.8 (NA); 1990: 102.8 (NA). Note. CFR = standard case-fatality rate: killed/(killed + seriously injured + lightly injured), expressed as percentage; CFRS = modified case-fatality rate: killed/(killed + seriously injured only), expressed as percentage; NA = not available. aData received from the Israel’s Police National Headquarters; data comprising only killed and seriously injured on 100-kph and 90-kph sections of highways 1 (Tel Aviv–Jerusalem), 2 (Tel Aviv–Haifa), and 4 (Tel Aviv–Ashdod) from 1990 to 1994. bData from Israel Central Bureau of Statistics 1992–1994: Road Accidents with Casualties: Part I. cTruck data received by Israel’s Police National Headquarters; data comprising killed and injured (serious + light) for 11 months of 1993 and 11 months of 1994 (November through September). dSingle-vehicle crash data from Israel Central Bureau of Statistics 1992–1994 Road Accidents with Casualties: Part I. Single-vehicle crashes include categories skidding, overturning, running of the way, and collision with fixed object. eSoldier data received by Israel Defense Forces. Data consists of 8-month periods (November 1992 to July 1993; November 1993 to July 1994). For the CFRS, we combine moderate and serious injuries (Israel Defense Forces have a different classification: slight, moderate, serious, and killed). fCFR ratios for 1993–1994 were corrected using the correction factor provided by the Central Bureau of Statistics to correct for underreporting in 1994. Note. CFR = case-fatality rate: the standard rate killed/all casualties; CFRS = modified case-fatality rate: killed/(killed + seriously injured). Student’s t test compared monthly death and case-fatality rates of November 1990 through October 1993 versus November 1993 through October 1998 for all roads combined, interurban alone, and urban alone. Slightly Injured were corrected for underreporting in 1993 by 1% and in 1994 by 9% as recommended by the Central Bureau of Statistics. Change in 1996 reporting of slightly injured resulted in an increase in reported light injuries by more than 8000 during 1995–1996. Change in reporting (“attendance only” requirement) probably reduced number of seriously injured as well. aCalendar year in this study is from November through October. The speed limit was raised officially on November 1, 1993. bAll data obtained from Israel Central Bureau of Statistics Road Accidents with Casualties: Part I for years 1990 to 1997. * P < .01; ** P < .001. aChange in travel speeds derived from fourth root of ratio of number of deaths after to deaths before increase in speed limit, using equation K(a)/K(b) = V(a)4/V(b)4 to solve for V(a), where K(a) and K(b) and V(a) and V(b) are deaths and average vehicular speeds, respectively, on roads after and before (Evans3,23). bChange in impact speeds calculated as fourth root of ratio of case-fatality rate after increase in speed limit to that before increase in speed limit (see references 23 and 24). cDerived by dividing absolute change in deaths by estimated percentage change in speeds. dSee the means for deaths per year before and after the increase in speed limit in Table 3. eDerived from increase in mean case-fatality rate alone from 12.5% to 17.3% on interurban roads and from 7.9% to 9.8% on urban roads after the increase in the speed limit from 90 to 100 kph on November 1, 1993. We used the formula K(ATTRIB) = {K(B) × (CFRS(A)/CFRS(B))} − K(B), where K(ATTRIB) is persons killed per year attributable to the increase in speed 5 years after the speed limit increase, K(B) is persons killed per year 3 years before the increase in the speed limit, and CFRS(B) and CFRS(A) are the proportion of those killed among all those seriously injured before and after the speed limit increase. We estimated average changes in speed of travel and crash impact and number of deaths per each 1% change in speed, using fourth-root models. fThe difference in observed deaths in the 5-year follow-up period and the absolute change (baseline to follow-up) in scenario 1 gives us expected deaths for scenario 2.
Highway Date No. of Sitings No. of Vehicles Weighted Mean Speed Traveled, kpha Range of SDs (by Siting) Weighted 90th Percentileb Right lanec Before November 1993 Highway 2 (Tel Aviv–Haifa) 2/7/93–9/14/93 5 3 533 90.7 9.9–12.7 108.7 After November 1993 Highway 2 (Tel Aviv–Haifa) 1/11/94–3/23/94 3 3 132 94.8 12.8–13.4 109.9 Left lanec Before November 1993 Highway 2 (Tel Aviv–Haifa) 1971–1990d 10 11 974 94.4 7.7–13.9 NA Highway 2 (Tel Aviv–Haifa) 9/7/93–9/9/93 7 8 076 98.6 12.5–22.9 114.6 Highway 4 (Tel Aviv–Beersheba) 8/23/93–8/25/93 6 8 340 96 13.6–20.0 111.7 Highway 1 (Tel Aviv–Jerusalem) 8/24/93–9/1/93 8 10 000 98.9 11.0–14.7 114 After November 1993 Highway 1 (Tel Aviv–Jerusalem) 4/27/94–5/30/94 11 2 762 107.9 8.6–17.3 126.6 Single-lane connecting road Before November 1993 Highway 40 (off Tel Aviv–Jerusalem) 3/92 3 1 461 72.2 11–15 84.8 After November 1993 Highway 40 7/1/94–7/4/94 8 1 901 90.7 11.2–15.3 113.6 No. Killed 1992–1993 No. Killed 1993–1994 Absolute Change, No. Ratio of Nos. (1993–1994/1992–1993) CFR 1992–1993, % CFR 1993–1994, % CFR Ratio CFRS 1992–1993, % CFRS 1993–1994, % CFRS Ratio Interurban Roads High Speeda 21 35 14 1.7 NA NA NA 14.0 21.0 1.5 Otherb 236 284 48 1.2 2.5f 3.0f 1.2 12.8 16.1 1.3 Allb 257 319 62 1.2 2.1 2.7 1.2 12.9 16.7 1.3 Urban roadsb 230 236 6 1.0 0.9 0.9 1.2 8.7 9.3 1.1 Trucksc Interurban 48 74 26 1.5 3.6 5.5 1.5 NA NA NA Urban 25 37 12 1.5 1.9 3.6 1.9 NA NA NA Single vehicled 85 108 23 1.3 1.80 2.7 1.5 8.8 11.9 1.4 Soldierse Off-duty 14 20 6 1.4 10.6 7.5 0.7 19.4 26.0 1.3 On-duty 1 11 10 11.0 0.40 2.9 8.0 3.6 25.6 7.2 Motorcyclesb 22 39 17 1.8 0.8 1.3 1.6 5.9 9.1 1.5 Pedestriansb 199 189 −10 1.0 3.7 3.8 1.0 13.7 14.2 1.0 Bicyclesb 12 10 −2 0.8 1.2 1.1 0.9 6.6 6.5 1.0 Yeara Killed, No. Seriously Injured, No. CFRS, % Range Interurbanb Before increase 1990–1991 219 1718 11.8 6.99–17.73 1991–1992 307 2078 12.9 10.04–18.07 1992–1993 257 1735 12.9 7.50–19.33 Mean 261 1843.7 12.5 After increase 1993–1994 319 1596 16.7 11.36–26.39 1994–1995 327 1596 17.0 11.58–21.09 1995–1996 293 1407 17.2 12.99–23.53 1996–1997 278 1259 18.1 12.82–24.37 1997–1998 284* 1344 17.4** 12.10–24.42 Mean 300.2 1440.4 17.3 Urbanb Before increase 1990–1991 211 2448 7.94 5.49–12.28 1991–1992 195 2592 7.00 3.46–9.65 1992–1993 230 2410 8.71 4.74–13.86 Mean 212 2483.3 7.9 After increase 1993–1994 236 2303 9.3 7.39–11.27 1994–1995 223 2376 8.6 6.15–11.77 1995–1996 235 2231 9.5 6.35–12.76 1996–1997 249 2156 10.4 5.47–14.04 1997–1998 253* 2008 11.2** 7.11–14.42 Mean 239.2 2214.8 9.8 Mean Deaths per Year November 1990 to October 1993 (Baseline), No. Mean Deaths per Year November 1993 to October 1998 (Follow-up), No. Absolute Change (Baseline to Follow-up), No. Ratio Change (Baseline to Follow-up) Estimated Change in Speeds (Nilssona and Jokschb Formulas), % Mean Change in Deaths per Year for Each 1% Increase in Speed, No.c Interurban roads Observedd 261 300.2 39.2 1.15 3.6a 10.9 Expected Speed limit raised; without increase in countermeasures and congestion (expected scenario 1)e 261 361.2 100.2 1.38 8.3b 12.1 Speed limit not raised; with increase in countermeasures and congestion (expected scenario 2)e 261 200 −61 0.77 −6.4a −9.53 Urban roads Observedd 212 239.2 27.2 1.13 3.1a 8.8 Expected Speed limit raised; without increase in countermeasures and congestion (expected scenario 1)e 212 263.3 51.3 1.24 5.5b 9.3 Speed limit not raised; with increase in countermeasures and congestion (expected scenario 2)f 212 188.1 −23.9 0.89 −2.9a −8.24
Partial support for this work came from university scholarship grants.
We thank Zvi Weinberger of Jerusalem College of Technology, Tali Tal of the Central Bureau of Statistics in Israel, Dr Orly Manor of the Department of Medical Ecology of Hebrew University-Hadassah School of Public Health and Community Medicine, and Professor Gerald Ben-David, Dr Jacob Adler, and Zelda Harris of Metuna for advice and helpful suggestions.
Human Participant Protection No protocol approval was needed for this study.
