Introduction
    Fortunately, the number of school bus accidents, and particularly accidents involving fatalities and serious injuries are very small. While this is a comforting fact, it poses a serious problem when one wishes to perform comparative studies in order to evaluate the effectiveness of a safety device. To determine statistically from past accident experiences whether seat belts (or any other safety feature), will improve safety, one needs a number of accidents that occurred under very similar circumstances and involved buses some of which were equipped with seat belts and some were not. Then, a determination of the device's effectiveness can be made by comparing fatalities, and the number and severity of injuries that occurred in the two groups of buses. This methodology cannot be used to determine the effectiveness of seat belts. First of all, Type II school buses are required to be equipped with seat belts, while the Type I are not. Therefore, no comparative studies can be done for the same class of vehicles. Although there are some Type I buses that are equipped with seat belts, their number and accident involvement is so small, that a comparative analysis is not possible.

    The next best alternative to a statistical analysis, is to investigate the accidents that did happen, and on the basis of expert judgement, determine what impact the presence of seat belts would have had on fatalities or injuries. Only accident investigations which covered a significant sample of accidents are considered in this chapter. The research team heard and saw in print a number of individual accident accounts that took place in the past within the state of New Jersey. They ranged from the Pemberton bus driver who thanked God for the absence of seat belts on the bus she was driving when it caught fire, to the Newark Police Chief who also thanked God but for the presence of seat belts on a bus that overturned after being struck on its side by a police car. Other accounts included:

A Type I school bus was hit by a truck. An observer felt that compartmentalization was enough since no children were injured. Actually the children helped the injured driver (who was not belted) off the bus.
• A car traveling at a very high speed (50-60 mph) scraped the front bumper of a Type II bus in East Windsor resulting in the injury of a student who "sprained muscle from seat belt" according to the report filed with the Department of Education.

   

• A Type I bus broadsided a Type II bus in Camden. Of the 20 belted children in the Type II bus, only one was injured (suffered a split lip), and so was the monitor who was unbelted. Of the 12 unbelted students in the Type I bus, eight went to the hospital and six had to be carried off on back boards with neck and head injuries.

   

• A bus in Newark was hit by a snow plow and crashed into a building. All students were belted and none were injured.

   

• Seat belts were helpful in the latest serious accident in the state (April 1989 in Montclair)

    All of the above accounts are rather anecdotal. They are quoted here just because they represent information collected during the course of the study. However, they were not used in the decision making process that led to the final recommendations because they do not represent a systematic and unbiased investigative effort.

    Only two systematic studies of school bus accident investigations have been performed in the past for the purpose of evaluating the effectiveness of lap belts. One was conducted by the National Transportation Safety Board (NTSB) and the other by the Texas Transportation Institute of the Texas A & M University System. The remainder of this chapter presents and discusses the findings of these two studies, after a brief summary of the injury scales used to classify accidents.

Injury Scales
    There is a number of ways that one can classify accidents according to their severity. The school bus accident literature uses primarily two injury scales to code school bus occupant injuries. The American National Standards Institute (ANSI) injury scale is the most commonly used. The Abbreviated Injury Scale (AIS) is the second standard classification scheme. The basics of both injury scales are briefly presented here in order to familiarize the reader with their classifications that will be mentioned frequently in subsequent sections.

ANSI Injury Scale
    The ANSI scale classifies non-fatal injuries in the following three levels [2]:

    Level A (Incapacitating Injury) The injury causes disabilities to such a degree that the person injured is not able to walk, to drive, or continue with any activities which was capable of performing before the injury occurred. Level A injuries include severe lacerations, broken or distorted limbs, skull or chest injuries, abdominal injuries, being unconscious at or when taken from the accident scene, and being unable to leave accident scene without assistance. Momentary unconsciousness is excluded.

    Level B (Non Incapacitating Evident Injury) Injuries of a nonincapacitating nature are evident to observers at the scene of the accident where the injury occurred. Level B injuries include lump on head, abrasions, bruises, minor lacerations and others. Limping is excluded (the injury cannot be seen).

    Level C (Possible Injury) Any injury reported or claimed that is not a fatal injury, incapacitating injury, or nonincapacitating evident injury. Level C injuries include momentary unconsciousness, claim of injuries not evident, limping, complain of pain, nausea, and hysteria.

    There may be some variations in coding these injuries in different states. Furthermore, it is not known how accurately police officers apply this scale when reporting the severity of injuries after an accident. The detailed New York State data on injuries in Table 10 can provide a better understanding on the types and severity of injuries coded in the ANSI injury scale. The Texas accident investigations used this injury scale also.

NTSB Study, 1987
    The National Transportation Safety Board investigated the crash performance of large post standard school buses (manufactured after April, 1977 with GVWR greater than 10,000 lbs) [11]. The investigation concentrated on 43 accidents involving 44 school buses (two buses collided with each other in one of the accidents). The study focussed primarily on events during crashes for evaluating how well the standards worked, how the occupants sustained injuries, if any, and how serious the injuries were. The study attempted to analyze also what difference lap belts would have made in the final outcome. Namely, whether the number and severity of injuries would have increased or decreased, if lap belts were present. It is this final aspect of the NTSB stud9 that is of interest at this point, and the sections that follow are devoted to it.

Study Features
    A school bus accident was included in the NTSB study if it met the first and at least one of the remaining three criteria listed below:

    The vehicle was a Type I school bus manufactured after April 1, 1977, and was occupied by school age children, and

    The school bus was involved in a moderate speed collision that disabled the bus (occupant injuries need not have resulted), or

    The school bus overturned, or

    One or more of the school bus occupants was seriously injured or killed in the accident (the accident could be of any type)

    Of the 44 buses studied by the NTSB, half (22) were involved in rollover accidents, but only 8 of them were pure non-collision rollovers (14 were involved in a collision prior to rolling over). Of the remaining 22 buses that were involved in non-rollover collisions, 16 were impacted from the front or rear, three were involved in side impact accidents, and three in multiple impact type collisions. There were a total of 1,166 school bus passengers and 44 drivers occupying the buses studied. There were 13 passenger fatalities and 588 injuries. The injury status of 15 passengers was not known, and 563 did not sustain any injuries. Four of the 44 buses included occupant restraints in the form of lap belts, loop belts (not considered a safety device), and secured wheel chairs. Of the 1,166 passengers 1,119 were unrestrained and 47 restrained (40 by lap belts).

Lap Belt effectiveness Analysis Results
    NTSB analyzed the injuries to unrestrained passengers in order to estimate what the effect of lap belt use might have been. To perform the analysis NTSB tried to seek answers of the following questions for each of the 1,119 unrestrained passengers:

    What injuries sustained by the unrestrained passengers would have been eliminated if they had been lap belted?

    What injuries would have been sustained if the passengers had been lap belted and held in place?

    The NTSB analysis of the potential impact of seat belts on fatalities and injuries of surviving passengers is summarized below:

    Fatalities: Out of the 13 total school bus passenger fatalities, lap belt use would have probably:

• prevented 2 deaths
  
• made no difference for 10 deaths, and
  
• the effect could not be determined for 1 death.

    Furthermore, lap belt use would have caused death to three of the surviving passengers. Thus, the probable net effect is practically zero. The study stated that, if lap belts were present, possibly an additional passenger death would have occurred.

    Injuries: The injuries sustained by the school bus occupants were classified according to the Abbreviated Injury Scale (AIS). The injury levels sustained by the 1,106 surviving passengers were as follows:

• Severe to maximum injuries (AIS 4 or above): 4 passengers
  
• Serious injuries (AIS 3): 24 passengers
  
• Moderate injuries (AIS 2): 58 passengers
  
• Minor or no injuries (AIS 1): 1,020 passengers

    The NTSB study dealt individually with each one of the above injury categories and reached some conclusions which are summarized in Table 4. For some injury categories best and worst estimates were provided, while for other categories only a single estimate was made.

     For AIS 4 and above injuries: Out of the 4 passengers injured at this level, lap belt use probably would have reduced injuries to an AIS 3 or below for 1 passenger, worsened the outcome for 1 passenger, and made no difference for 2 passengers. In effect, lap belt use probably would not have changed to total number of surviving school bus passengers with severe or worse injuries.

     For AIS 3 injuries: NTSB estimated that at best the net effect from the presence of lap belts would have been to reduce the accident severity for 7 passengers in this category (8 reductions and 1 increase). At worst, the net effect would have been to reduce the accident severity for 2 passengers (8 reductions and 6 increases).

     For AIS 2 injuries: Of the 58 passengers in this category, lap belt use probably would have worsened the injury severity for one-fifth (12) of the passengers. At best, the injury severity could have been reduced for 9 passengers with a net effect of increasing the injury severity for 3 individuals. At worst there would be no severity reductions with a net effect of increasing the accident severity for 12 passengers. The Board could not determine the effect that lap belt use would have on the majority of passengers in this category.

     For AIS 1 or no injuries: No conclusions were reached for the remaining (1,020) unrestrained passengers who sustained minor or no injuries. The data were insufficient to make judgements for this category, and the report stated that the ".Board did estimate in some individual cases the effect that lap belt use could have made. Overall, it is not prepared to make the same injury outcome determinations as done for the higher level injuries. It is unlikely that seat belt use would have reduced minor injuries."

     Ejected passengers: There were 15 passengers, among the 1,119 unrestrained, that were totally or partially ejected from the buses. Six of the ejected passengers died. Two sustained critical injuries, 5 severe injuries, 2 serious injuries, 2 moderate injuries, and 4 minor injuries. Discussing the fatalities, NTSB investigators concluded that it was not known ".if they died as a result of injuries sustained outside the bus, during ejection, or as a result of injuries sustained within the bus before ejection. It is not correct to assume automatically that all injuries sustained by ejected passengers occurred as a result of ejection and thus, had they been restrained, injury outcomes would have improved".

     In an overview of the injury outcomes, the NTSB study noted that 90 percent of the 1,119 unrestrained school bus passengers in the study sustained no injuries or only minor (AIS 1) injuries, 5 percent received moderate (AIS 2) injuries, only 4 percent sustained higher than moderate level (AIS 3-6) injuries, and the outcome for 1 percent was not known. Therefore, the school bus passengers fared very well in the accidents investigated, even though they were selected in a way that slanted the sample toward the more serious accidents. This observation, however, simply reiterates the fact that school buses are indeed very safe vehicles overall.

Restrainted Passengers in the NTSB Study
    The NTSB study sample included four buses carrying restrained passengers. Of the 47 restrained occupants of these buses, 40 were lap belted. The NTSB experts felt that they did not have sufficient data to ascertain whether lap belts had a positive or negative impact on the injury severity of these passengers and they refrained from drawing any conclusions on the effectiveness of lap belts on the bases of the belted passengers' injury experience.

Study Conclusions
    On the basis of all the evidence gathered from the 43 accidents it investigated, NTSB summed up its findings in the following two broad conclusions:

    "The Board does not recommend that Federal safety standards be amended to require that all new large school buses be equipped with lap belts for passengers."

    "The Federal safety standards, providing for "compartmentalization", worked well in Safety Board investigated crashes to protect school bus passengers from injury in all types of accidents."

Texas School Bus Accident Investigations-1986
    The Texas Transportation Institute investigated school bus accident data obtained from police reports in the State of Texas for the ten-year period between 1975 and 1984 [13]. During this period there were 12,669 accidents involving school buses in the state. Of these, only 7 percent (887 accidents) were injury or fatality accidents. There were 19 fatal injuries, 160 incapacitating injuries (A level; not able to walk, drive, etc.), 1,648 non-incapacitating injuries (B level, bump on head, abrasions minor lacerations, etc.), and 2,359 possible injuries (limping, complaint of plain).

    The 19 fatalities occurred in 13 accidents and were classified as follows:

• Fell out of the open door 1
  
• Leaning head out of windows 3
  
• Ejected 3
  
• Passenger head struck the bus roof 2
  
• Thrown around within the bus (in rollover) 8
  
• Severe collision impact at the passenger seating position 1
  
• Wheel chair turn over (non crash event) 1
  

     The Texas team analyzed the police report of each accident, and made a determination on whether seat belts would have prevented each of the above fatalities. Their conclusions were:

     Seat belts would have eliminated the fatal injuries for the student who fell out of the bus and was run over by the school bus itself, and for the three students who were leaning their heads out of windows. However, appropriate disciplinary measures, which were not always followed, would have eliminated these fatal injuries also.

     Two of the three ejected passengers would have been saved by seat belts. The ability of a seat belt to prevent injuries to the third passenger was not clear, because the impact took place right at the passenger's location.

     Seat belts would have prevented injuries for the two passengers who were injured because their head struck the roof of the bus and for the 8 passengers who were injured as they were thrown around inside the bus in rollovers.

One passenger was severely injured due to a severe collision impact near the passenger's seating position, and the benefit of a seat belt could not be determined. The passenger who died because his wheel chair turned over was not involved in any crash event, and the accident report had no indication that the bus was specially equipped to carry handicapped children.

In summary, the Texas researchers concluded that seat belts would have saved 12 lives, would have saved four additional ones which could also have been saved if the students were properly disciplined, and no conclusion could be reached for three additional deaths. This assessment was partially questioned by Kyser who expressed concerns about the findings for two accidents that produced 6 fatal injuries [2]. On the basis of his personal on-site inspection in the case of one accident, and from his personal conclusions based on physical evidence and discussions with the accident investigators in the other case, he inferred that there was no evidence which could lead one to state that lap belts would have prevented death for the 6 passengers in these two accidents. The Texas researchers acknowledged in a memorandum that the police officers' information and narratives were marginal at best in their ability to determine seat belt effectiveness in passenger fatalities, and that detailed information of the type collected by Kyser and others could contribute to a better estimate of seat belt effectiveness. However, even if the six disputed fatalities were not prevented by seat belts, there were still six fatalities that seat belts could have prevented, and four more for which seat belts could have provided equal protection as proper discipline.

     Another aspect which was not considered in the Texas study was an assessment of any possible fatal injuries that might have occurred to the less severely injured passengers had they been wearing seat belts. The Texas researchers found that the information on the police accident reports was not sufficient to make such an estimate. The NTSB investigation of 43 accidents that was discussed earlier, concluded that seat belts might have prevented two fatalities but there might have been fatal injuries to three other and less seriously injured passengers had they been belted. The NTSB study pointed out also that in rollover collisions, the fatalities and serious injuries were caused primarily due to the force of impact, and not as much by the dynamics of the rollover. It was not clear from the Texas study to what degree the collision impact or the dynamics of rollovers contributed to the deaths of the eight passengers who were thrown around inside the bus. If the impact was responsible for the fatality, then seat belts probably might not have helped. If both the impact and rollover dynamics contributed to fatalities, then it was most likely that the severity of the fatal injuries might have been reduced by seat belts in the majority of the accidents, just as the Texas study inferred.

     In the cases of incapacitating, non-incapacitating, and possible injures, the effectiveness of seat belts could not be determined by the Texas study. Summarizing their investigation, the researchers observed that the school bus is an extremely safe mode of transportation, since only 19 fatal injuries and 160 incapacitating injuries were sustained in over 12,600 school bus accidents in a 10-year period. Their major conclusions were that:

    The data available do not provide conclusive evidence that lap belts are needed in large school buses.

    With the limited funds available, lap belts in school buses are not safety cost effective.

    Improved vehicle maintenance, bus driver training and rider training may have a greater potential in reducing the frequency and severity of accidents over time.

Review of Accident Investigations
    The NTSB investigation of 43 accidents involving large school buses determined that seat belts could have provided additional protection to passengers that suffered AIS-3 and above level injuries. A large number of undetermined cases in the AIS-2 injury category precludes one from drawing any definitive conclusion on the effectiveness of seat belts for these moderately injured passengers. It is reasonable to conclude that the benefit or harm that could be contributed by seat belts would have been rather marginal. The fact that the sample of accidents analyzed in the NTSB study are slanted towards severe crashes, essentially reduces the chances that any protective device could have made much of a difference. There were 11 serious to critical injuries and only 4 minor to moderate injuries associated with the 15 passengers who were partially or fully ejected from the buses. These injury outcomes are an indication of the extreme severity associated with ejection, and ejection is certainly an event which can be prevented by seat belts.

     Frontal and rear end impacts were the most common collision modes investigated by the NTSB study. Although side impacts cause a substantial proportion of school bus accidents (about 30 percent), there were just three such accidents in the sample of 43. While side impacts were under-represented, rollovers were over-represented in the NTSB sample. Of the 43 cases, 22 were rollovers, contributed either by collision (14) or non-collision (8) incidents, while such accidents contribute to no more than 25 percent of the total. Seat belts are deemed to improve safety in both side impact and rollover accidents. The disproportionately large number of rollover accidents may have balanced the disproportionately small number of side impacts included in the sample and, therefore, the overall anticipated benefits can be considered representative of the results that could have been obtained, if the sample was more representative of actual accident frequencies. Overall, due to the severity of the accidents included in the Safety Board's investigation, the findings only pertain to what might happen to belted passengers during the worst possible school bus accident cases. But, even for these cases, the results are not against seat belts. Belts would have been neutral for fatalities, severe injuries and minor injuries, would have reduced serious injuries, and increased moderate injuries. The trade-off is between the reduction of serious injuries and the increase of moderate injuries and it should be considered as being overall beneficial.

     The very large sample of accidents investigated by the Texas Transportation Institute study makes it more significant in terms of its ability to evaluate the overall effectiveness of seat belts in school buses. The Texas study inferred that 10 out of 19 fatalities could have been prevented by seat belts. Although this number excludes the 6 contested fatalities, it implies an approximate effectiveness rate of 50 percent. An assessment of seat belt effectiveness for injuries could not be made by the Texas accident investigations due to lack of sufficient data. However, the determination of fatality and incapacitating injury frequency by accident type shown in Table 5 is of particular importance. It is evident that 63 percent of the incapacitating injuries are caused by side impacts and rollover accidents, 28 percent are the result of front impact accidents, and rear impacts cause an insignificantly small number of injuries. Since these proportions are based on a very large sample of data, they can provide insights that can be used for the seat belt effectiveness methodology developed in Chapter 8.
Source: School Bus Safety Belt Study, N.J. Institute of Technology, pps 21-30.