Regulations (Preambles to Final Rules) - Table of Contents Regulations (Preambles to Final Rules) - Table of Contents
• Record Type: Powered Industrial Truck Operator Training
• Section: 4
• Title: Section 4 - IV. Studies of Accident and Injury Data and Training Effectiveness

IV. Studies of Accident and Injury Data and Training Effectiveness

This section of the preamble discusses the reports, studies, and other sources of data and information that were analyzed to determine the magnitude and extent of the problems that powered industrial truck operator training can mitigate. It also contains a discussion of the studies that demonstrate how better training can improve safety.

A. Accident and Injury Data

1. The Census of Fatal Occupational Injuries

The Bureau of Labor Statistics (BLS) maintains a database entitled Census of Fatal Occupational Injuries (CFOI). The CFOI is a compilation of information on fatal work injuries that occurred in the 50 States and the District of Columbia. BLS gathers pertinent information from death certificates, workers' compensation reports, and other Federal and State records. Information is verified by using at least two source documents.

The census contains a collection of information on the workers and the circumstances surrounding each fatality. The data are compiled annually.

In April, 1994, BLS published a booklet entitled Fatal Workplace Injuries in 1992: A Collection of Data and Analysis (Ex. 3-4). This booklet contains an article written by Gary A. Helmer entitled Fatalities Involving Forklifts and Other Powered Industrial Carriers, 1991-1992. This report contains information from the CFOI on 170 fatal powered industrial truck accidents. Table 1 lists the reported causes of these accidents.


2. Characteristics of Work-Related Injuries Involving Forklift Trucks

In 1987, Nancy Stout-Wiegand of the National Institute for Occupational Safety and Health (NIOSH) published an article in the Journal of Safety Research (Winter 1987, Vol. 18, No. 4, pp. 179-190) entitled Characteristics of Work-Related Injuries Involving Forklift Trucks (Ex. 8-6). This article contained an analysis of powered industrial truck injuries reported in two occupational injury databases -- the National Electronic Injury Surveillance System (NEISS) and the Bureau of Labor Statistics' Supplementary Data System (SDS).

The NEISS database is composed of records from a national sample of 200 hospital emergency rooms and burn centers handling all types of injuries. The NEISS database was originally established by the Consumer Product Safety Commission, and its original intent was to gather data about accidents involving consumer products rather than industrial injuries. The hospital emergency rooms included in the sample were not necessarily those located in industrial areas, predominantly treating industrial injuries and illnesses. The data from this sample are weighted to represent the nation in numbers and characteristics of traumatic injuries treated in emergency rooms and burn centers.

A subset of this database -- the work-related injuries -- is maintained by NIOSH. Because the NEISS database records only injuries treated in emergency rooms and burn centers, traumatic work injuries treated by private practitioners or by industry or private clinics are not included. Moreover, chronic injuries, such as those caused by overexertion, are not as likely to be treated in emergency rooms as are acute traumatic injuries and, therefore, are probably under-represented in the NEISS database. Other probable sources of error in calculating accident rates include misclassification of the sources of injury or the agent of injury. For example, if an employee fell from the elevated forks of a powered industrial truck, the accident could be misclassified as a fall from elevation rather than a fall from a forklift. Similarly, if an employee were struck in the head by part of a load that fell from a powered industrial truck, the accident could be classified as an "employee struck by falling object" accident. In either case, the accident would have involved a powered industrial truck, but in neither case would it be classified as a powered industrial truck accident.

The Supplementary Data System (SDS) database is composed of workers' compensation claims for injuries involving lost workdays. Thirty states provide information to the SDS system. The SDS system reports the occupations of injured workers and states where the claims are filed. The SDS includes only compensable injuries. The definition of a compensable injury varies from state to state. In some states, injuries are compensable, for example, if they result in one day or more away from work. In other states, the time away from work may be up to 7 days before the injury becomes compensable.

The SDS and NEISS data do not necessarily include the same injuries because injuries treated in emergency rooms do not always result in lost workdays. At the same time, compensable injuries included in the SDS may not have been treated in emergency rooms and thus would not be represented in NEISS. However, both of these databases represent the more serious injuries involving powered industrial trucks, that is, those requiring treatment in emergency rooms and those that result in compensable injuries.

In 1983, the SDS system identified 13,417 workers' compensation claims for lost-workday injuries involving powered industrial trucks. Assuming that the 30 states in the SDS system are representative of and proportional in population to the whole country (50 states), approximately 22,400 compensation claims (5/3 x 13,417) are filed nationally for lost-workday injuries involving powered industrial trucks. This number is comparable to the estimated 24,000 forklift-related injuries that were treated in U.S. emergency rooms in 1983 as reported by NIOSH from information gathered by the NEISS system. In 1985, the NEISS system reported a total of approximately 34,000 powered industrial truck-related accidents that were treated in emergency rooms. This reflects an increase in the number of such accidents reported by NEISS studies of about 39% over the three-year period from 1983 to 1985.

The SDS report also contained a tabulation of the occupations of the injured workers. The breakdown of the occupations of those employees and the corresponding percentage of accidents are listed in Table 2.


3. Industrial Forklift Truck Fatalities -- A Summary

OSHA's Office of Data Analysis (ODA) examined 53 investigative case files involving powered industrial truck fatalities that occurred between 1980 and 1986 (Ex. 3-7). The results of ODA's analysis are summarized in Table 3, below. Note: the columns do not always add to 100 percent in various tables because of rounding.


The single largest cause of accidents was vehicle tipovers (percentages attributed to specific causes may not track those in Table 3 because a single specific cause -- tipover -- may be classified under more than one accident type in that table). These tipovers were attributed to the following: (1) The vehicle was out of control (speeding, elevated loads, mechanical problems, etc.; 7 instances -- 13 percent); (2) the vehicle was run off/over the edge of the surface (4 instances -- 8 percent); (3) the operator attempted to make too sharp a turn (excessive speed, unbalanced load, etc.; 4 instances -- 8 percent); (4) an employee jumped from an overturning vehicle being pulled by another vehicle (2 instances -- 4 percent); (5) the vehicle skidded or slipped on a slippery surface (2 instances -- 4 percent); (6) the wheels on one side of the vehicle ran over a raised surface or object (2 instances -- 4 percent); and (7) the vehicle tipped over when struck by another vehicle (1 instance -- 2 percent).

The second highest number of fatalities reported in the ODA study resulted from accidents when employees were crushed between a vehicle and a surface. These accidents were attributed to the following: (1) The operator got off the vehicle while it was running (7 instances -- 13 percent); (2) a worker on a platform was crushed between the platform and an overhead surface (2 instances -- 4 percent); (3) an employee's leg was caught when a vehicle sideswiped a metal surface (1 instance -- 2 percent); (4) an employee attempted to prevent a vehicle tipover by holding up the overhead guard (1 instance -- 2 percent); (5) an employee changed a tire and the vehicle fell from the jack (1 instance -- 2 percent); and (6) an empty 55 gallon drum used to support the vehicle during maintenance collapsed (1 instance -- 2 percent).

Four of the six accidents where employees were crushed between two vehicles were caused by contact between two moving powered industrial trucks, and the other two involved contact between a powered industrial truck and a stationary vehicle.

Of the five accidents that were identified as being caused by an employee being struck or run over by a vehicle, four were accidents where employees other than the vehicle operator were struck by the vehicle. The remaining one involved an operator trying unsuccessfully to board a free rolling vehicle.

4. Selected Occupational Fatalities Related to Marine Cargo Handling as Found in Reports of OSHA Fatality/Catastrophe Investigations

In 1992, the OSHA Office of Data Analysis (ODA) published a study of fatalities and catastrophes that had occurred in the marine terminal industry (SIC 4491, Marine Cargo Handling) between the years 1975 and 1984. This report is entitled Selected Occupational Fatalities Related to Marine Cargo Handling as Found in Reports of OSHA Fatalities/Catastrophe Investigations (Ex. 27). This report contains an analysis of the causes of and other information about 141 accidents that resulted in 165 fatalities that occurred during the period of the report. Of those accidents, 19 (11.5 percent) were attributed to the unsafe use of powered industrial trucks.

5. The OSHA Fatality/Catastrophe Reports

OSHA records a summary of investigation results of accidents resulting in fatalities, catastrophes, amputations, and hospitalizations of two or more days, and those accidents that have received significant publicity or involved extensive property damage. These summaries are recorded on an OSHA Form 170 and include an abstract describing the activities taking place at the time of each accident and the causes of the accident. These reports are stored in a computerized database system, and cover inspection data from 1984 to 1991. There were 4268 reports of accidents in the system that resulted in 3038 fatalities, 3244 serious injuries, and 1413 "non-serious" injuries (many of the accidents resulted in multiple fatalities and/or injuries).

OSHA queried the database for all reports that contained the keyword "industrial truck." This produced a printout of 208 accidents (Ex. 8-8). These 208 accidents resulted in 147 fatalities, 115 serious injuries, and 34 "non-serious" injuries.

By adding the number of fatalities, serious injuries, and "non-serious" injuries and dividing that sum by the number of accidents, OSHA determined that 1.4 injuries of some nature occurred per serious accident reported. OSHA also determined that 4.8 percent of the fatalities, 3.5 percent of the serious injuries, and 2.4 percent of the "non-serious" injuries were attributable to an accident that involved a powered industrial truck.

These percentages are derived by dividing truck-related fatalities, serious injuries, and other injuries by the corresponding total number of reported fatalities, serious injuries, and other injuries. For example, the 147 forklift fatalities were divided by the 3038 total fatalities to arrive at the 4.8 percent figure.

OSHA examined the OSHA Form 170s to determine the causes of the accidents that were attributable to the use of powered industrial trucks. Table 4 is a compilation of the causes of these accidents.


It should be noted that many of the accidents could have been caused by improper training. For example, when a vehicle tipped over, an employee might have been transporting an unbalanced load because that employee had not been trained about load balance.

Using the OSHA Form 170 data, OSHA also compiled a listing of the industries in which these accidents occurred. Table 5 provides list of industries, and the number of accidents that occurred in those industries. (For a complete listing of the individual industries, see Ex. 3-9.)



6. OSHA Emergency Communications System Reports

OSHA has another internal system for collecting information about serious accidents. This system requires that serious and/or significant accidents be reported to the National Office over the telephone.

This telephone system is part of the OSHA emergency communications system. Regional Administrators are required to file a first report of fatalities, catastrophes, and other important events (such as those that receive significant publicity) with the National Office. The information contained in these reports is disseminated to responsible officials in OSHA. The National Office receives approximately 1200 reports yearly. (See Ex. 8-10.)

None of the reports is screened before the OSHA National Office receives them. Although these reports are not considered statistically significant for the purpose of calculating the total number of serious workplace accidents, OSHA believes that they represent a reasonable sampling of the most serious type of accidents and that the causes of the accidents closely parallel the distribution of the causes of all accidents.

OSHA has examined the First Report of Serious Injury reports for the years 1980-1991 and has identified 247 that involved powered industrial trucks. Table 6 lists the number of reports received each year, the number of those accidents that involved powered industrial trucks (PITs), and the corresponding percentages.


Each of the reports involving powered industrial trucks was examined to determine the causes of the accidents. In some instances, multiple causes were identified. Table 7 lists the number of the accidents that were attributable in whole or in part to each cause.


7. OSHA General Duty Clause Citation Analysis

OSHA's Office of Mechanical Engineering Safety Standards analyzed the citations that were issued between 1979 and 1984 for violations of the general duty clause [section 5(a)(1) of the OSH Act]. During that period, there were 3637 inspections that resulted in the issuance of at least one such citation. (See Ex. 8-11.)

Sixty-five general duty clause citations involved powered industrial truck operation. These citations were issued under the general duty clause because the dangerous condition did not appear to be covered by a specific requirement in Section 1910.178. Each was examined to determine the nature of the violation. Table 8 lists the type and number of violations that were cited.


B. Studies Measuring the Effectiveness of Powered Industrial Truck Safety Training Programs

In 1984, H. Harvey Cohen and Roger C. Jensen, working under contract with the National Institute for Occupational Safety and Health (NIOSH), published an article in the Journal of Safety Research (Fall 1984, Vol. 15, No. 3, pp. 125-135) entitled Measuring the Effectiveness of an Industrial Lift Truck Safety Training Program (Ex. 3-5). The authors analyzed two studies undertaken to measure objectively the effects of safety training on powered industrial truck operators' driving performance and safety practices.

This article detailed the results of an experiment that was conducted to evaluate powered industrial truck operator training using a work sampling procedure to obtain objective data about work practices that correlate with injury risk. Two separate studies were conducted in this experiment, one at each of two similar warehouses. These studies were conducted to assess the value of training and the influence of post-training actions on workers' safety performance. These studies demonstrate that training powered industrial truck operators reduced the operators' error rates (number of unsuccessful operations divided by the total number of operations) and that training combined with feedback further reduced error rates.

The studies were conducted at different warehouses using similar training techniques. The training emphasized those operator driving behaviors that were measurable, frequently observed, capable of being reliably observed, related to frequent accident occurrence, and amenable to corrective action through training. Fourteen driving behaviors were evaluated in these studies. Positive reinforcement during the training (use of praise rather than criticism) was used with some trainees to measure its effectiveness. The experiment was conducted in four phases:

(1) The pre-training phase, during which none of the operators had been trained;

(2) The post-training 1 phase, during which the control group remained untrained, the training group (called the treatment group in the study) had been trained, and the training-plus-feedback group had been trained and had also received performance feedback;

(3) The post-training 2 phase, during which all three groups had been trained but only the training-plus-feedback group had received performance feedback; and

(4) The retention phase, which started three months after the end of the post-training 2 phase (and the end of the feedback program).


Following the initial training (post-training 1), all three groups showed a decrease in their mean error rates, with the training-plus- feedback group showing the largest decrease (from .35 to .27, a 23 percent decrease), followed by the training-only group (from .33 to .27, an 18 percent decrease), and the control group (from .34 to .32, a 6 percent decrease). The control group's reduction in error rate from the pre-training to the post-training 1 phase of the study was attributed to the influence of peer modeling, i.e., the untrained control group operators were copying the behavior of their trained counterparts. Toward the end of the post-training 1 phase, the error rates of the three groups converged, suggesting that the effects of the training program had begun to wear off. Observers also noted that some behaviors were being compromised when employees with different knowledge levels were required to interact, particularly in conflict- avoidance situations such as signaling and yielding at blind intersections.

During the post-training 2 phase of the study, all groups' performance improved. The control group's performance improved by 28 percent (from a mean error rate of .32 to .23), while the training group experienced a four percent improvement (from a mean error rate of .27 to .26) and the training-plus-feedback group had a seven percent improvement (from .27 to .25). There was further evidence of a peer modeling effect because all three groups' performance continued to improve although no additional instruction was given.

The retention phase was conducted three months after the completion of the post-training 2 phase of the study to determine the longer term effects of the training. During this phase of the study, mean error rates were checked, as they were during the other phases of the study. The results of this phase of the study indicate a further improvement in the operators' performance, with the mean error rate decreasing from .25 to .19, a 24 percent improvement in performance. The total performance gain achieved during this study was a 44 percent improvement from the pre-training (baseline) phase through the retention phase (from a mean error rate of .34 to a final error rate of .19). These data indicate that there were significantly fewer errors at each successive phase of the study.

The second study was conducted to verify and extend the findings of the first study. A modified experimental design was used to eliminate the mitigating influence of the untrained control group. In the second study, all operators were trained at the same time and all received performance feedback. Comparisons were made only before and after training. The study was divided into three phases: pre-training; post- training; and retention. The retention phase of the study was again conducted three months after the conclusion of the prior phase.


After the vehicle operators were trained, they experienced a 61 percent improvement in performance scores (from an error rate of .23 to .09). During the retention phase of the study, there was a further reduction of 22 percent in mean error rates (from .09 to .07 mean error rate). The overall improvement in mean error rates between the pre-training error rate (.23) and that achieved during the retention phase (.07) was a reduction of 70 percent.

Not all errors cause accidents; however, most accidents are caused by one or more errors. The final rule is intended to minimize operator errors. The studies show that better training reduces operator errors. OSHA, the authors of the studies described in the preamble, and other experts believe that accidents will be reduced by about the same percentage as the reduction in the error rate. The studies that OSHA has used are among the best available for cause and effect.

[63 FR 66237, Dec. 1, 1998]

Regulations (Preambles to Final Rules) - Table of Contents Regulations (Preambles to Final Rules) - Table of Contents