IIHS research bibliography

Objective: To assess trends in traffic fatalities on several temporal scales: year to year, by month, by day of week, and by time of day, to determine why some times correspond with higher rates of crash deaths, and to assess how these trends relate to age, the role of the deceased, and alcohol consumption. Method: Traffic fatalities were identified using the Fatality Analysis Reporting System (FARS) for 1998 through 2014 and assessed for their time of occurrence. Three days that, on average, contained particularly high numbers of crash deaths were then assessed in greater detail, considering the age of the deceased, role of the deceased (vehicle occupant, bicyclist, motorcyclist, or pedestrian), and the blood alcohol content of either the driver (for passenger vehicle occupants) or the deceased. Results: Annual crash fatality totals were much lower in 2014 than in 1998, but the decrease was not steady; a marked drop in crash deaths occurred after 2007 and continued until 2014. On average the most fatalities per day occurred in July and August (116 per day), followed closely by June, September, and October. During the week, the greatest number of fatalities on average occur on weekend days, and during the day the most fatalities tend to occur between the hours of 3 p.m. and 7 p.m. Holidays like Independence Day and New Year’s Day show elevated crash fatalities, and a greater percentage of these crashes involved alcohol, when compared with adjacent days. Conclusion: Certain days and times of year stand out as posing an elevated crash risk, and even with the decrease in average daily fatalities over the past decade, these days and times of year have remained consistent.

Objectives: The objective of this study was to determine the extent to which the 2015 increase in U.S. traffic fatalities is consistent with historical relationships between the economy and fatalities and to project the trend in fatalities through 2024 based on those relationships. Methods: Linear regression was used to model the number of vehicle miles traveled (VMT) during 1990-2015 as a function of calendar year and the unemployment rate. Then Poisson regression was used to model annual traffic fatality counts as a function of year, the unemployment rate, and miles of travel. Projections of future unemployment rates by the Bureau of Labor Statistics were used to project future miles traveled and traffic fatality counts. Results: VMT was estimated to increase by an average 1.5% per year if the unemployment rate was unchanged. For each unit increase in the unemployment rate, the VMT declined by an average 1.8% per year. Fatalities were estimated to decline by approximately 2.2% per year if the VMT and unemployment rate were unchanged. Each 1% increase in VMT was associated with a 0.96% increase in fatalities. Finally, after accounting for the effects of time and VMT, for each unit increase in the unemployment rate fatalities declined by 2.1%. Conclusions: The sharp increase in U.S. traffic deaths in 2015 can be tied primarily to the improving economy, although other lesser factors also were involved. However, the long-term decline in traffic deaths seems to be continuing. Assuming a more gradual improvement or leveling of the economy, it is projected that there will be about 34,000 traffic deaths in the U.S. in the year 2024.

The Association for the Advancement of Automotive Medicine (AAAM) is an international, multidisciplinary professional organization dedicated to limiting injuries from motor vehicle crashes. It was founded in 1957 by six practicing physicians whose avocation was motor racing and has grown into a nearly 500-member group of physicians, engineers, epidemiologists, behavioral scientists, and others, representing more than 20 countries. The AAAM members convene every year for presentations of the latest research on traffic injury prevention across diverse disciplines. Refereed papers presented at the annual scientific conference address injury patterns and causation, injury severity scoring approaches, the biomechanics of impact, the relationship between injury risk and vehicle design, the role of alcohol and other driver risk factors in crashes, the long-term economic and social costs of crash injuries, and a comprehensive array of other issues critical to understanding and preventing motor vehicle trauma.

This article is in response to “Can Tax Deadlines Cause Fatal Mistakes?” by D.A. Redelmeier and C.J. Yarnell in CHANCE 26(2). The authors examined the hypothesis that increased stress stemming from the deadline for filing tax returns, usually April 15, affects the number of people involved in fatal crashes (drivers, passengers, pedestrians, bicyclists, motorcyclists, killed, injured, uninjured, unknown). They compared Tax Day with days one week before and one week after, which controls for day of week and seasonality. They found a 6% increase in the number of people involved in fatal crashes on Tax Day, relative to the comparison days. The difference was statistically significant, based on the assumptions of the test they used.

The basic concept of various pay-at-the-pump proposals is that part of the mandatory component of automobile insurance liability premiums would be collected as an additional fee along with each gallon of gasoline. This idea has attracted some support because it appears to be simple and fair, and it could produce environmental benefits because higher gasoline prices would reduce vehicle mileage. Every motorist buys gasoline, proponents of pay-at-the-pump point out. Motorists who buy the most gasoline are presumed to drive the most and, therefore, to be at the greatest risk of a crash. These are the motorists who would pay the most for insurance under pay-at-the-pump plans, so it's fair. Right? In fact, no. The problem is that, contrary to the claims of its advocates, pay-at-the-pump insurance wouldn't improve on present insurance systems. It's basically unfair and could have the unintended effect of compromising safety.

In this article we examine the long-term effects of the sudden, unexpected loss of a spouse or child. In the spouse study, interviews were conducted with 39 individuals who had lost a spouse in a motor vehicle crash 4 to 7 years ago and with 39 matched controls. In the parent study, interviews were conducted with 41 parents who had lost a child in a crash and with 41 matched controls. Control respondents were matched to bereaved respondents case-by-case on the basis of sex, age, income, education, and number and ages of children. Significant differences between bereaved spouses and controls were revealed on several indicators of general functioning, including depression and other psychiatric symptoms, social functioning, psychological well-being, reactivity to good events, and future worries and concerns. For the most part, these differences persisted when variables such as present family income and present marital status were statistically controlled. Comparisons between bereaved and control parents also revealed significant differences on some measures of general functioning (especially depression), but these were not as pervasive as the differences obtained in the spouse study. Responses to questions about current thoughts and feelings suggest that the deceased continued to occupy the thoughts and conversations of bereaved spouses and parents. Moreover, a large percentage of respondents (from 30% to 85%, depending on the question), continued to ruminate about the accident or what might have been done to prevent it, and they appeared to be unable to accept, resolve, or find any meaning in the loss. Taken together, the data provide little support for traditional notions of recovery from the sudden, unexpected loss of a spouse or child.

Much of the information about the effectiveness of highway safety measures comes from simple before and after comparisons. Such comparisons can be misleading because they do not compare what would have happened had no changes taken place to what actually happened after the changes were made. Examples are given and three methods for accurately estimating the effects of safety measures are provided.

Unintentional deaths from suffocation and strangulation account for about 20 percent of all nontransport-related infant and child fatalities in the United States. In the late 1950s, some preventive countermeasures were introduced to reduce the number of deaths resulting from refrigerator or freezer entrapment. A few years later, countermeasures were introduced to prevent deaths resulting from suffocation by plastic bags, inhumation, and mechanical strangulation from wedging in infant cribs. For three of these major causes of suffocation and strangulation deaths among infants and children (refrigerator or freezer entrapment, suffocation by plastic bag, and inhumation at construction sites), there appears to have been a significant decline in incidence; however, there is no evidence of a significant reduction in deaths from mechanical strangulation in cribs. The impact of current countermeasures is discussed, and some suggestions for new or modified approaches are made.

Major mishaps among mobile offshore drilling units worldwide from 1955-1981 were identified from industry and government sources. Based on aanual numbers of rigs in service and typical staffing patterns, annual mishap rates and fatality rates for rig types and mishap categories were computed. While the frequency of major mishaps has increased in recent years, the mishap rate per 100 rig-years of service has remained stable. The overall stability obscures the fact that jack-up rigs have had an increasing mishap rate while the rate for other rig types combined has gradually declined. Although the fatal mishap rate has also remained constant, the annual fatality rate per 100,000 full time equivalent (FTE) workers has risen sharply. This can be attributed to increasing numbers of lives lost in environmental mishaps while deaths from operational mishaps have declined. There were 344 fatalities during the 27-year period. Although an average of some 13 deaths per year worldwide appears minimal, the relatively small size of the workforce gives this number significance particularly when it is noted that 'occupational' fatalities, those occurring in the course of routine operations, are not included. The overall fatality rate secondary to major mishaps was 84.3 per 100,000 FTE worker-years.

The high incidence of illegal passing of school buses reported by school districts in some large urban areas prompted a study to assess the magnitude of the problem, describe its characteristics, and to determine the feasibility for future research. Findings from this preliminary study indicate that illegal passing of stopped school buses is a definite problem, with an average of 6.58 vehicles illegally passing per bus per day. Stops on multi-lane roadways experienced more illegal passes than those on two-lane roads, accounting for over 77 percent of all illegal passes. A survey of motorists suggests that some motorists are confused about the current signaling system. Future research on countermeasures to reduce the incidences of illegal passing of stopped school buses is recommended.

Recent research at the University of Toronto found that safety countermeasures often appear to be more effective than they really are. Under present policies, sites with poor crash histories are recognized and singled out for remedial treatment. A before-and-after study of hte same sites will almost automatically find an improvement even if the corrective treatment is useless. The importance of the regression-to-the-mean is demonstrated, and ways of puring the bias from interpretations of data are presented.

The incidence of hospital-treated facial injuries caused by dog bites was determined from a population-based study involving Dane County, Wisconsin, hospitals during 1978 and 1979. Annual rates were shown to be 152 per 100,000 for ages 0 to 4 years, 128 per 100,000 for ages 5 to 9 years, and 62 per 100,000 for ages 10 to 14 years. Severe facial injuries from dog bites were found almost exclusively in children younger than 10 years. If these rates apply to children in the US population, then an estimated 44,000 facial injuries, 16,000 of them severe, caused by dog bites are seen in hospitals each year. The cumulative incidence of facial injuries from dog bites for children to age 14 years is 1.6%. Most of the published advice for preventing dog bite injuries to the face suggests parental diligence in keeping children away from dogs, but options such as choosing dogs less likely to bite children may be more effective.

House fires kill about 5,000 Americans annually, at a rate (2/100,000) that has remained almost constant for the past 50 years. House-fire deaths were studied in Baltimore, where 55 residents died during a three-year period. More than half of the deaths resulted from cigarette-ignited fires; 39% of the people who died in such fires were not the cigarette smokers themselves. For both blacks and whites, the death rate was highest in census tracts where property rental values were low. The death rate from fires ignited by heating or electrical equipment was nine times as high in the lowest-value census tracts as in the highest.

Effects of a brief educational and purchase program concerning home fires and smoke detectors by two pediatricians were compared to "routine" counseling without such a program using two groups each of 120 patients of well children. Inspection performed four to six weeks after the office visits showed that of 55 experimental group parents without detectors prior to the program, 26 purchased and 19 installed them correctly. No control group parents did so.

This paper considers the analysis of series of Poisson counts generated in multiple jurisdictions. At staggered times in some of the jurisdictions, similar policy changes are made, altering the Poisson intensity. For instance, over a period of several years, roughly half the states repealed or weakened their laws mandating helmet use by motorcyclists. We describe an iterative method that facilitates the log-linear modeling of such situations. Using Monte Carlo simulation, we document conditions under which the mehtod provider unbiased and consisten estimates of the effect of the policy change.

A technique has been developed which enables the decomposition (separation) of a myoelectric signal into its constituent motor unit action potential trains. It consists of a multichannel (via one electrode) myoelectric signal recording procedure, a data compression algorithm, a digital filtering algorithm, and a hybrid visual-computer decomposition scheme. The algorithms have been implemented on a PDP 11/34 computer. Of the four major segments of the technique, the decomposition scheme is by far the most involved. The decomposition algorithm uses a sophisticated template matching routine and details of the firing statistics of the motor units to identify motor unit action potentials in the myoelectric signal, even when they are superimposed with other motor unit action potentials. In general, the algorithms of the decomposition scheme do not run automatically. They require input from the human operator to maintain reliability and accuracy during a decomposition.

In 1978, 345 persons were killed in the United States in jumps and falls from non-crashing motor vehicles: 64 per cent fell; 15 per cent jumped; and it was not known whether the other 21 per cent jumped or fell. Two hundred and one people had been traveling on the exterior of vehicles, especially truck beds, and almost all of these people fell from their vehicles. The other 144 fatalities involved people in passenger compartments. Many of the falls from compartments occurred when occupants opened doors, or when vehicles changed direction. Seventy-seven per cent of those who fell from passenger compartments were males, and 44 per cent were less than five years old. Among those who jumped from vehicle compartments, 62 per cent were women and all were older than 14 years. Fatal falls and jumps from vehicles could be reduced in a variety of ways. These include legislation to prohibit travel on vehicle exteriors, designing vehicles so that doors cannot be opened when in motion, improving door designs, installing signals that provide warning if doors are not closed completely, and using occupant restraints.

Medical examiner records were reviewed for 42 Maryland children younger than 10 years who died of asphyxiation from 1970 through 1978. Twelve children choked on food; six of these deaths involved hot dogs. Eight choked on nonfood objects. Size, shape, and consistency were important, with small, round, pliable products predominating. Twenty-two deaths resulted from suffocation, including four infants who died when plastic bags in their cribs or playpens pressed against their faces. Twelve of the 42 deaths resulted from problems that are now the subject of Consumer Product Safety Commission activity or regulations. Important problems not currently addressed include plastic bags, balloons, and foods that because of their shape or consistency are especially likely to cause asphyxiation.

Firearms in the United States are second only to motor vehicles in the number of fatal injuries they inflict. In 1977 they caused 32,000 deaths, or 15 per 100,000 population. Of these, the manner of death was classed as unintentional ("accident") in 2,000, homicide in 13,000, suicide in 16,000, legal intervention in 250, and undetermined in 900.

A substantial proportion of the effort to reduce injuries in the past has been aimed at people engineering - attemtping to change individual behavior thought to contribute to the incidence of injury. Many such approaches, when subjected to rigorous research, have been found ineffective and some have been harmful. Engineering of the phsycial environment, such as in the introduction of motor vehicle safety standards, has been more effective. However, the adoption and implementation of such engineering has been difficult because of factors in the social environment. The fundamental issue in achieving major gains in reduction of incidence and severity of injuries is wether or not the social environment can be made conducive to the adoption of effective chagnes in the physical environment that are likely to reduce the incidence and/or severity of injuries.

It is a privilege, honor, and distinct pleasure to accept your invitation to give this, the second annual William S. Stone Lecture. Reducing the frequencies with which injuries occur, lessening their severities, and ameliorating their results continue to be the most neglected of the major health problems. Nevertheless, we are well into a long-term transition in concepts, in knowledge, and in their applications. This is increasingly replacing decades of unsubstantiated guesswork and assertions with approaches based on systematic analysis, scientific research, and hard-nosed modern management. Dr. Stone's work and that of many of you here today exemplify this transition, especially in the immediate surgical context. Yet, although many surgical and closely related aspects of the overall trauma field have long made outstanding progress, and continue to do so, progress in the general trauma field lags badly, largely for completely unnecessary reasons. As a result, millions of people are being needlessly hurt each year.

"On the job" motor vehicle deaths number more than 4,000 annually in the U.S. and comprise nearly one-third of all work-related deaths. Yet the Department of Labor has set no standards relating to on-the-road safety of the millions of workers whose jobs entail large amounts of driving, and Department of Transportation standards affecting occupational safety cover only drivers in interstate commerce. Drivers of some commercial vehicles, such as heavy trucks, are at special risk of injury because trucks have usually been exempted for many years from federal motor vehicle safety standards--such as standards for brakes and seatbelts--designed to prevent crashes or protect occupants in crashes. Observations based on a series of 150 fatal crashes involving tractor trailers illustrate the need for better protection of this large population of high-risk workers. Clarification of responsibility within the various federal agencies and application of available knowledge and technology are essential.

The validity of the assumption that loss of a driving license would result in economic hardship for most drivers was examined. Persons who drive to work were asked how they would reach work if a broken leg prevented them from driving. Only 21 per cent said they would not be able to make other travel arrangements. Of those who thought they could find other transportation, two-thirds said the alternative would cost no more than driving themselves. Less than one-fifth said the time required would be more than an additional half hour per day. The results indicate that policies and practices related to license suspension and revocation should not be based on an assumption that job loss or economic hardship would be experienced by most drivers if they were to lose their driving privileges.

An experiment to identify motor vehicle law violators was conducted in Maryland; termed Project Tags, it involved the mass screening of license plates. In December 1973, the Maryland Motor Vehicle Administration, Maryland state police, and the Insurance Institute for Highway Safety launched the project to identify vehicles whose plates indicate they are wanted (i.e. stolen cars or cars involved in felonies) or whose drivers may be operating illegally (i.e., with suspended or revoked licenses). Identification was determined by the use of rapid, computer-based mass screening of the vehicle registration plates of either parked vehicles or those in motion. Laws were extensively researched to ensure that the screening was neither arbitrary nor capricious, and that it did not violate constitutional guarantees. The initial phase of the project, conducted to determine if mass screening would actually work in a variety of environments, was carried out by using a van equipped with observers, keyboard displays, and a minicomputer system containing a list of registration plate numbers of interest to law enforcement officers. The second phase incorporated the mass screening function directly into patrol cars equipped with specially designed query devices that contained the list of registration plate numbers. The observer in the car, using a display console located in front of the passenger position, performed the mass screening function. When the project's results were compared with those of normal police operations in Baltimore County, Md., the comparison showed that mass screening produced more citations for driving while licenses were suspended, revoked, cancelled, or refused than normal operations. Further research is needed to determine how mass screening can be used in other situations.

Research indicates that stressful social and psychological events increase the probability of some illnesses and, perhaps, injury producing events. The usefulness of this knowledge in the amelioration of human damage is, however, in considerable doubt. The history of public health suggests that identification of manipulable necessary conditions for human damage is the key to amelioration of such problems. Such conditions are yet to be identified in the stressor-stress-damage syndrome, but in some cases there are known manipulable necessary conditions that allow the syndrome to be bypassed in the amelioration of human damage.

A statistical model for the evaluation of the effectiveness of motor vehicle inspection programs in reducing highway crashes is presented. The model is based on the assumption that the waiting time between highway crashes follows an exponential distribution. Since highway crashes are relatively rare events, it is assumed that the length of the study period is such that censoring occurs. Under these assumptions, maximum likelihood estimates of the mean waiting time until a crash for the non-inspected (inspected) vehicles is obtained and the corresponding test statistic is derived. As mechanically-caused accidents are but a small part of the overall accident picture and since inspection should only affect this portion, sample size requirements are investigated for various combinations of , (increase in average time until a crash due to the effect of inspection), L (length of study period), and a = ß (probability of Type I error equalling probability of Type II error). For reasonable , the sample required is indeed sizable.

In December 1973, the Maryland Motor Vehicle Administration, Maryland State Police and the Insurance Institute for Highway Safety initiated a demonstration project to test a new technique developed by the Institute for identifying, through rapid mass screening of vehicle registration plates, vehicles and persons that may be involved in certain types of unlawful activity. The techniques (called TAGS because it involves vehicle registration plate data) is still in its trial phase. However, the preliminary results of the project indicate that the TAGS technique may have potential both for strenthening the ability of police to identify and apprehand persons involved in such unlawful activity as driving with suspended or revoked permits, driving stolen vehicles or vehhicles with stolen registration plates and driving unsafe vehicles, and for increasing the rapidity and rate of police identification of stolen cars.

Using accident and inspection data from two states, the effect of periodic motor vehilce inspection on highway crashes is investigated. In both states, accident data from the initial year of the statewide program are examined. Unfortunately, it was not possible to restrict the analysis to mechanically-caused acccidents. In addition, there were serious diffficulties with the phasing-in scheduels and the necessary data file linkages, viz., non-compliance in North Carolina and linkage of inspection, license plate distributino, and corresponding accident information in Florida. As a result, the major contributions of this investigation would be in the statistical methodologies employed. These data from North Carolina and Flordia do not provide evidence of the effectiveness of periodic motor vehicle inspection in reducing highway accidents. However, with the limitations in these data and the probable small effect, if any, of vehicle inspection, it is not unexpected that these studies would fail to detect such an effect.

The potential health hazard created by sublethal concentrations of carbon monoxide from automotive exhaust in the atmosphere has evoked increasing concern. At the same time, the large number of deaths that occur in vehicles when lethal concentrations of carbon monoxide enter the passenger compartment has escaped widespread attention. This paper presents data suggesting that over 500 Americans die each year from carbon monoxide poisoning because their vehicles are defective due to deterioration, damage, or poor automotive design. Most of these deaths occur in non-moving vehicles; therefore they are not classified as “motor vehicles accidents” and rarely are thought of in connection with vehicular deaths. Yet probably no other group of deaths is more closely related to defective motor vehicles. In 1967, 819 deaths in the U.S. were classified as “accidental poisoning by motor vehicle exhaust gas.” As with many types of adult poisoning, information on these deaths is scanty because they rarely reach the hospital alive and therefore are not reported through the poison control information system. Sopher and Masemore’s report on 6 recent carbon monoxide deaths described methods of investigation and disclosed the role played by defective vehicles. Their findings led to this review of all Maryland deaths during the past six years caused by unintentional carbon monoxide poisoning in motor vehicles. The objectives were to document the size of the problem, describe the age and condition of the vehicles and the circumstances under which these deaths occurred, and consider the possible solutions suggested by the data.

Two and three parameter lognormal distributions were fitted to the distributions of certain types of automobile insurance claim payments. Maximum likelihood estimates of the parameters were obtained using both grouped and ungrouped data, for the three parameter cases and the two parameter grouped data cases, the method of scoring worked well. Comparison of the efficiencies of the estimates using grouped data showed that logarithmic grouping was substantially better than equal interval grouping.

The primary purpose is to examine the trends covering the five-year period of the program and to present a follow-up to a previous highway safety research center (HSRC) report that presented statistics for the initial year of the program relative to failure rates and repair charges, and their relationship to vehicle age, mileage and presumed driving environment. Differences also for selected inspection items in various car models are examined with respect to failure rates. Inspection receipts of a special sample of 76,668 privately-owned passenger cars inspected in december 1968 were collected to provide statistics for comparison with those in the 1966 sample. Not only has the overall failure rate decreased from 70.1 percent in the 1966 sample to 33.8 percent in the 1968 sample, but the average repair charge (per vehicle with repair charge assessed) is generally lower this sample than for the previous sample. As in the previous study, urban cars accumulated less mileage and, in general, were newer than rural cars. Moreover, the average repair charge is still approximately 25 cents lower for rural cars than urban cars. In contrast to the 1966 results, however, a difference was found between rural and urban overall failure percentages with the overall urban percentages being significantly greater than the overall rural failure percentages. In addition, the urban failure rates were significantly greater than the rural rates for headlights, steering mechanism, and foot brake. A questionnaire was sent to owners of a number of automobiles spotted on the highways throughout north carolina. Results reveal that (1) 70 percent of the 459 respondents considered the stat's motor vehicle inspection program either of considerable or at least of some value, and (2) 60 percent felt more confident in the safety of their car after inspection, with females expressing significantly more confidence than males.

Carbon monoxide in concentrations above 15% in the blood rarely appears to be a factor in highway crashes. However, carbon monoxide in lower concentrations, and cigarette smoking which can produce such concentrations, have been hypothesized as important hazards to safe driving and one study has reported that smokers have higher crash rates than non-smokers. In the present study, however, fatally injured drivers had evidence of heavy smoking more often than did drivers not involved in crashes but stopped at roadblocks at times and places where crashes had occurred. The excess was attributable to the fact that heavy smokers also are heavy drinkers, and often are problem drinkers. Persons who smoke heavily but do not drink heavily were not over-represented among the fatalities. It is hypothesized that problem drinkers represent a substantial proportion of persons with chronic respiratory disease just as they do among highway fatalities.

The objective of this study is to determine the feasibility of observing large numbers of individual vehicles for the purpose of spotting a "hit", that is, a vehicle that has been reported stolen, or a vehicle owned by a driver whose license has been suspended. During a two-week period in October, teams of observers recorded the plate numbers of 10,106 vehicles observed in parked and moving locations in the greater Seattle area. A total of 16 stolen vehicles were observed. These accounted for .2% of all the observed vehicles. One-third of these stolen vehicles were observed parked near taverns and roadhouses. A comparison of the registered owners' names with a list of 55,000 suspended drivers produced 475 mathces of varying degrees of positive match. For the purposes of this study, a mathc was considered a "hit" if the degree of match was positive enough to warrnat stopping the vehicle for driver identification. Of the 475 matches, 420 were considered "hits". These account for nearly 5% of the total number of observed vehicles. These exploratory results indicate a high yield of "hits" in terms of both stolen vehicles and possible suspended drivers from the observation of large number of vehicles.

The author suggests that many automobile accidents and resultant injuries can be avoided. He discusses the problem in three phases: precrash, crash, and postcrash. Precrash conditions include driver disabilities such as drinking and senility, the road, and the vehicle. The crash phase deals with the vehicle's interior and exterior design and the crash design of the highway; the postcrash phase discusses emergency response systems such as communication, ambulance attendants, first aid knowledge, and emergency hospital facilities.

Immediately after VJ Day in 1945 a discharged GI entered the driver license department of his State to reapply for his driver's license. He smiled calmly as he took the application and sat near the clerk's desk to complete it. In his hurried entrance he barely noticed an attendant refilling the coke machine. Suddenly the calm was shattered when the attendant dropped an empty case of bottles. Without warning the ex-GI rose from his chair like a projectile, screamed, and collapsed. He suffered from "battle fatigue," and, in that condition, was obviously a bad highway risk. Had it not been for a dropped case of coke bottles, this condition would not have been detected.

Fire and burns are the third most common category of fatal childhood injury in the U.S. In 1974, 1,359 children under 15 years of age died of fires or burns, exclusive of scalds and explosions. Childhood burn deaths have long been almost unique among injury deaths in that female rates exceeded male rates. In contrast, death rates are more than twice as high for males as for females for all injuries and age groups combined. On the basis of mortality data for 1959-61, Baker hypothesized that the extraordinary female excess in childhood burn deaths might be attributable to 100sefitting, easily ignited nightgowns and dresses. Since the late 1950's, female clothing styles have changed dramatically and pants are now worn by many females, including young girls. The- present investigation was undertaken to determine whether the sex ratio in childhood burn deaths has also changed and, if so, whether the shift appears to be associated with a decrease in clothing ignitions.