Document xnqZm15XYxwj5d56XjoewzoJ

Human Factors Perspectives on Vlbmings EVALUATING the effectiveness of warnings under prevailing WORKING CONDITIONS Frank E. Gamer Behavioral Science Applications'--" Yellow Springs, Ohio 45387 ABSTRACT Due to a lack of published test results from which to draw conclusions about the effectiveness of warnings under different working conditions, I conducted a field study as an aid in the formulation of expert opinion during the discovery phase of pending product-liability litigation. My intent was to measure the effectiveness of a label which warned of the risk of delayed lung disease. The design of the warning conformed to the existing requirements for the period of interest - the mid 1960s. To replicate the prevailing conditions in the plant at the time that exposure to the hazard occurred, appropriate engineering and administrative concroIs purposefully were omitted. The warnings failed to cause a significant reduction in the incidence of unsafe behavior. INTRODUCTION Human factors specialists are being called upon more frequently to provide expert consulta tion during product-liability litigation. The content and appearance of warnings, an issue in many cases, is an area in which human factors considerations have been applied extensively. In fact, it has been suggested that a human factors specialist should testify in every case involving the adequacy of warnings (Messina, 1984). Certainly, warnings should be but one element [ of an integrated health and safety program for minimizing exposure to a hazard that cannot be effectively removed from the workplace through design changes. The other elements consist of engineering and administrative controls. Beginning with frequent monitoring, engineering controls serve to isolate the worker from the hazard and may include such methods as enclosure, guarding, exhaust ventilation, and proper job design. Ad ministrative controls, on the other hand, include: (1) enforceable policies which limit the number of workers exposed and, as appropriate, require the wearing of personal protective equipment; (2) strict housekeeping practices; (3) periodic med ical examinations; and (4) compulsory training and educational programs. Thus, the proper role of a warning is to reinforce what has been learned through compulsory education and carefully super vised training. Despite many recent appeals that warnings be tested for effectiveness under representative working conditions (Christensen, 1983; Dorris and Purswell, 1977; Peters, 1984), such testing usually is limited to subjective evaluations of preference for or, perhaps, recall of competing warnings. Why should effectiveness testing under represent ative working conditions be an important consider ation? Clearly, one reason is to minimize pro spective liability, by documenting the comprehen siveness of the foreseeability analysis performed during the development of the warning. Second, if several formats and message contents seem applicable, then realistic effectiveness data can be factored into the design trade-offs. Third, due to a lack of published data from which to draw conclusions about the effectiveness of warn ings under different working conditions, it may be necessary to conduct tests during the discov ery phase of pending litigation, as an aid in the formulation of expert opinion. BACKGROUND Recently, I was asked to provide expert opinion in a case in which a failure to warn was alleged to be a contributing factor in the caus ation of occupational disease. The plaintiffs in the case had been exposed to asbestos fibers In the 1950s and 1960s at an asbestos insulation manufacturing plant. A review of the case ex hibits and plant-related documents revealed that prescribed engineering and administrative con trols were missing, and that warnings, had they been present on bags of raw asbestos entering the plant, would have been the principal method of controlling the hazard. Besides their primary functions in minimiz ing exposure to a hazard, it seems evident that engineering and administrative controls also serve to heighten a worker's awareness of a hazard and the attendant risk. If these controls are indeed missing or deficient, might this not constitute a set of working conditions which could predis pose an otherwise reasonable individual to de value a warning that arguably was adequate in terms of basic design? And would this be espec ially true if the hazardous material had a delayed effect on the health of the individual? The functionally disabling, life-shortening nature of asbestos-related diseases now is well documented in the medical literature (Nicholson, Perkel, and Selikoff, 1982). However, case histories and epidemiological studies underscore the long latent period between initial exposure to asbes tos fibers and the presence of definitive clini cal symptomatology. The question I attempted to answer empiri cally was what immediate impact would a welldesigned warning have had, if "suddenly" presented under these prevailing conditions? In the design Human Factors Perspectives on Whmings of this study, I focussed on unskilled laborers among the plaintiffs who had unloaded and stacked porous bags of asbestos on an irregular basis for varying periods of time. Many had been provided to the plant by temporary employment agencies, and many had handled bags of asbestos for only a fev days or weeks. No exhaust ventilation or res pirators had been provided by the plant, proper dust control and housekeeping had been nonexist ent, and no educational and training programs had been made available to describe the risks and to teach proper work practices and safeguards. More over, although perceived by the workers as annoy ing and bothersome, the asbestos material itself was not considered to be hazardous. Clearly, I could not expose individuals in a field study to asbestos fibers; therefore, I sub stituted granular limestone - a nontoxic material classified as producing nuisance dust. However, the warning was modeled after an appropriate asbestos-related warning for the early to raid 1960s, i.e., not after the asbestos-related warn ings which began to appear at that time nor, for that matter, after the current OSHA-mandated warning. Since the late 1930s, there had been general agreement that workers in asbestos insul ation plants were at risk to develop asbestosis. However, consensus within the medical community concerning a causal relationship between exposure to asbestos and the development of lung cancer and mesothelioma was not achieved until the late 1960s, following publication in December 1965 of the Proceedings of the New York Academy of Sciences International Meeting on the Biological Effects of Asbestos. In designing the format and content of a label which warned of a hypothetical limestonedust hazard, I consulted the open literature and trade journals of the period, 1945 to 1971 revis ions of the Manufacturing Chemists Association Manual L-l (entitled "Warning Labels; A Guide to the Preparation of Warning Labels for Hazardous Chemicals"), and the 1976 ANSI Standard for pre cautionary labeling of hazardous industrial chemicals. Table 1 presents a listing of current guidelines for developing warning labels. These guidelines have been extracted from consensus standards, federal regulations, industry handbooks, government recommendations, textbooks, and journal articles. It is noteworthy that, of the guide lines prescribed today, only hazard-specific color coding and the use of attention-directing symbols and pictographs were not recommended by the 1960s. EXPERIMENT Table 1. Current Guidelines for Warnings I. Does the warning command attention? A. Is the warning conspicuous? 1. ' Is the warning clearly visible? 2. Have attention-directing symbols and pictographs been used? 3. Has the appropriate signal word been used? 4. Has appropriate color-coding been used? 5. Is the size of the warning scaled to the dimensions of the product? 6. Has a border been used to isolate the warning. II. Does the text of the warning: A. Identify the hazard? B. Indicate the degree of risk and the consequences of exposure? C. List conditions under which the product is likely to be a hazard? D. List precautions and means of avoiding the hazard? E. Identify actions to be taken if exposure to the hazard occurs? F. Employ clearly understandable, familiar language? G. Provide a message that: 1. Is accurate? 2. Uses active voice? 3. Is affirmative and avoids "fudge" words? 4. Creates the appropriate concern for safety and perception of the risk by incorporating an urgency that is commensurate with the danger? 5. Is concise? H. Take into account the persons to whoa the warning is addressed? I. Conform to common standards, regulations and practices? III. Is the warning placed in reasonable proximity to the hazard? IV. Is the text readable? A. Has a foreseeability analysis con sidered reading distance, illumination . levels, and other factors? B. Have label life and degradation been considered? C. Have typographic features been considered? trailers by hand, rather than be secured to pallets and transported by forklift. Workplace Subjects #| A government warehouse was selected as the test site, since a wide variety of packaged mater ials is received-from vendors, unloaded from trucks, stored, and eventually reloaded for ship ment to other government installations. Moreover, it is not uncommon for individual containers or bags to be removed from or placed within truck Seventeen men who were experienced in heavy loading and unloading were studied. Since it was important to eliminate the influence of administrative controls that were rigorously enforced by management, the men were not part of the work force assigned to this facility. Instead, they were employed by an agency routinely contacted by the warehouse when f-------------Human Factors Perspectives on Ubmings I jjperienced workers were needed for loading and ^loading shipments after normal duty hours. Procedure Nine of the workers reported to the warehouse from 4 to 8 p.m., and they were assigned to 2 un loading crews of four and five persons, respec tively- The other eight workers were told to report from 9 p.m. to 1 a.m., and they were 8ssigned to 2 loading crews of four persons each. CONTAINS LIMESTONE ___________ DUST IS CREATED WHEN LIMESTONE IS |WARNING!f HANDLED. BREATHING LIMESTONE DUST CAUSES DELAYED LUNG DISEASE. SMOKING INCREASES THE RISK. Do Not Breathe Limestone Dust. The study was conducted on three consecutive days during the two time periods noted above. Handle Limestone Only in Areas Where There 13 Exhaust Ventilation. Again, unloading and loading tasks were performed at different times. Depending on the time period, the workers were told when first arriving that Always Wear an Approved Protective Mask When Handling Limestone. they would be unloading or loading shipments of limestone for the next three days. There was no discussion of a potential health hazard, nor was If Limestone Dust Has Been Inhaled, Contact Plant Management. any safety-related training provided. The un loading crews were instructed to remove individ ual bags stacked in rows in the trailer and to place them on pallets positioned by the edge of Figure 1. Format and Content of Warning, the dock. The loading crews were instructed to , lift individual bags from pallets also positioned by the edge of Che dock and to stack them in exhaust ventilation or exposure to hazardous rows in the trailer. dust. All data were collected unobtrusively. It was decided that undue psychological stress Individual bags of limestone weighed 22.7 could not be caused by having any of the workers kgs and were approximately 48 by 29 by 10 cm in believe that they were at risk for days or weeks. dimension. During each time period, two trucks Therefore, all workers were fully debriefed after were either unloaded or loaded by separate crews work on the same day Chat the warnings were in at adjacent docks. The pallets were filled and troduced. emptied sequentially. Forklifts transported pallets to and from the docks. Each crew handled Results approximately 800 bags of limestone every day. Because the bags were perforated along the sides At the beginning of the second work day (i.e,, and at the top, and because there were no pro prior to the appearance of the warnings), 3 of visions for exhaust ventilation or portable dust the 17 workers requested and then wore res collection, Che dust was very visible and con pirator masks for the remainder of the study. centrated in the air within the trucks and sur Because many of the workers had prior work rounding the docks. The clothing and exposed experience in other dusty environments and since areas of all workers were literally covered with they did communicate with each other during the this dust during the unloading and loading tasks. unloading and loading tasks, it is not surpris ing that some workers, at least generally safety On the first day, the workers acclimated to conscious, would seek protection after being the task requirements. Bags of limestone with exposed to the heavy dust on the first day. In out warning labels were unloaded and loaded in fact, it supports the credibility of the test accordance with the procedures described above. The second day consisted of data collection in the control condition; again, no warning labels environment. However, a change from 0 to 3 in the number of workers wearing respirator masks is not statistically significant (Fisher exact were present on the bags. The third day was devoted to data collection in the experimental probability test, j> > 0.10). What is surpris ing is the finding that when the warning labels condition; warning labels (as depicted in Figure were introduced, only 3 of the 14 workers who 1) were present and clearly visible on every were not wearing respirator masks at the time bag handled by the workers. A within-subjects requested and wore them. This does not repre design was chosen to examine the veracity of sent a statistically significant reduction in claims by certain plaintiffs in the case, that unsafe behavior (Fisher exact probability test, they would have heeded warnings on bags of as j> > 0.10). If we adopt a legal criterion instead bestos even if such warnings appeared after of a statistical criterion, an effective warn they, had begun working at the plant. ing is one that "more probably than not" will reduce the occurrence of unsafe behavior in the Recorded were: (a) the time required to fill and empty each pallet; (b) the number of workers requesting and wearing respirator masks; and (c) the number of workers reporting inadequate workplace. Clearly, the instruction to wear respirator maska was not heeded by a majority of workers. Moreover, none of the 17 workers re Iported the hazard or complained about a lack of Human Factors Perspectives on fthmings exhaust ventilation. If a worker does not explicitly follow the instructions in a warning, can it be assumed that he or she does not appreciate the risk. Perhaps real or perceived pressures from fellow workers or management can cause a worker to continue handling a substance without a respirator. Thus, it is possible that the warnings could have en gendered the appropriate perception of risk even though unsafe behavior continued in the workplace. Since an abrupt change in the time required to complete a practiced task can be a sensitive measure of work-related stress, response time data were entered into a 2 X 2 X 20 repeated measures analysis of variance, representing task (unloading vs. loading), workday (control vs. experimental), and pallet number (1 through 20), respectively. Neither the main effect of workday nor any interaction involving this factor were significant; the time required to fill and empty pallets was not affected by the presence of warn ings. Although the work area was extremely dusty and they observed their co-workers wearing res pirator masks, 11 workers still failed to comply with this precautionary measure after the warnings were introduced. All workers reported during the debriefing that they had seen the warnings. None stated that they had dismissed the warning as false or because it was insufficiently strong in word ing. Rather, they believed that they could con tinue loading or unloading as they had the previ ous two days. REFERENCES Christensen, J. M. (1983). Human factors consid erations in lawsuits. In G.A. Peters and D. B. Esau (Eds.), Safety law; A legal refer ence. Park Ridge, H: American Society of Safety Engineers. Dorris, A. L., & Purswell, J. L. (1977). Warnings and human behavior: Implications for the design of product warnings. Journal of Products Liability. 1_, 255-264. Fischhoff, B. (1977). Cognitive liabilities and product liability. Journal of Products Liability. 1, 207-219. Messina, J. (1984). The human factors expert in torts litigation. Trial. 20. 38-42. Nicholson, W. J., Perkel, G., & Selikoff, I. J. (1982). Occupational exposure to asbestos: Population at risk and projected mortality. American Journal of Industrial Medicine. 3, 259-311. Peters, G. A. (1984). Fifteen cardinal principles to ensure effectiveness of a warning system. Occupational Health and Safety. May. 76-79. While the motivating factors which influence a person's decision to accept or reject risk are not well understood (Fischhoff, 1977), in times of high unemployment workers will take greater risks because they must remain on the job. Further, if exposure to the hazard is painless, albeit annoying, and the disease takes years to be manifest, then denial is a likely defense mech anism. In view of the special emphasis now being placed on warnings as a result of OSHA's new Hazard Communication Standard, hopefully even greater pressure can be exerted for effectiveness testing under a variety of representative working conditions. CONCLUDING COMMENTS Two additional points must be made. First, by the late 1960s, bags of asbestos handled in insulation manufacturing plants should have con tained warnings specifically describing the risks of lung cancer and mesothelioma. Second, this paper should not be cited as proof that great care need not go into the design of warnings. Rather, in addition to training workers to interpret warnings, the findings of this study suggest that workers also should receive training to act in accordance with the warning. 80