Pergamon PII: S0003-4878(98)00062-3 Ann. occup. Hyg.. Vol. 42. No. 8. pp. 565-569. 1998 C 1998 British Occupational Hygiene Society Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain. 0003-4878,98 $ 19.00 + 0.00 Short Communication Does Risk Perception Affect Behaviour and Exposure? A Pilot Study Amongst Asbestos Workersj A. J. STEWART-TAYLOR^ and J. W. CHERRIE [Department of Environmental and Occupational Medicine, University of Aberdeen^ Aberdeen. AB25 IZD^J.K^ Institute of Occupational Medicine, Edinburgh. EH% 9SU U.K. I It is hypothesised that there is a causal association between workers' perceptions of the risks from handling hazardous materials, their behaviour while working and their consequent exposure. This has been investigated in a small group of workers carrying out remedial work on amosite insulating boards and similar products. Risk perception was first assessed using a questionnaire and then the workers' behaviour was recorded alongside task-based measurements of fibre exposure. There was a clear association of higher cumulative exposures when workers used power tools compared to manual methods (about seven times higher). Careful bagging was shown to reduce exposures by a smaller margin, (approximately half). Workers whose perception of the risks was poorer were found to be more likely to use power tools to remove the asbestos containing material. However, fibre exposure was not found to be directly associated with risk perception. Further work is necessary to clarify the validity of the original hypothesis^ 1998 British Occupational Hygiene Society. Published by Elsevier Science Ltd. asbestos containing materials. To minimise the importance of the type of materials being handled we It seems plausible that our perception of the risks have only selected sites where amosite wall boards experienced at work will, to some extent, dictate our or panels, amosite ceiling boards or tiles or amosite behaviour and hence our exposure to these risks. For containing gas flues were being removed. Information example, workers with a heightened awareness of the on the characteristics of the site and work procedures risks arising from exposure to asbestos may take grewere obtained during a preliminary visit. We ater care in their work and this may lead to lower exposure to airborne fibres. Understanding causal attempted to study some men who were employed by links between risk awareness, behaviour and exposure organisations sponsoring the remedial work and some is important in the control of exposure and yet it is an who were employed by a contractor commissioned to area of occupational health research which has not remove asbestos. A self-administered questionnaire was devised to received much study. In this paper we present some preliminary data from a study of asbestos remedial obtain relevant personal information from each workers. We have categorised their awareness of the worker, to assess their knowledge of the health hazrisks from asbestos using questionnaire responses and ards associated with asbestos exposure (three questhen measured their exposure to asbestos, while sim- tions), their perception of the causes of airborne asbestos dust (three questions) and their awareness of the ultaneously observing their behaviour. methods to prevent exposure (one question). The questionnaire was peer reviewed by a group of occumethods pational hygienists, a psychologist and a group of The investigation was carried out at a number of post-graduate occupational hygiene students. The aim sites where remedial work was being undertaken on of this review was to ensure the questions were both appropriate and, as far as possible, unambiguous. For the analysis the individual workers question responses Received 27 January 1998; in final form 9 June 1998. were categorised as 'high', 'medium' or 'low' per% Present address: 4 Fairfield Drive. Whitburn. South Tyneception (scores as 2, I or 0) and these categories were sidesr6 7HE. then summed to give a score for the question category Author to whom correspondence should be addressed. INTRODUCTION
566 A. J. Stewart-Taylor and J. W. Cherrie on the same scale, i.e. 0 to 2. These scores were then added to give an overall risk perception score ranging from 0 to 6. A check-list was developed to assess workers behaviour during asbestos remedial work. The main work patterns of interest were whether the asbestos containing material broke during removal, whether power tools were used, the use of water and/or surfactant based wetting agents, and the manner of bagging asbestos containing waste. Other behavioural information recorded included smoking habits, presence of facial hair, protective equipment worn and use of decontamination facilities. This checklist was also peer reviewed. Asbestos exposure levels were measured during removal and cleaning tasks using the Health and Safety Executive method MDHS 39/3 (HSE, 1990). All samples were independently evaluated on two occasions by one of us (AJS-T). Prior training in the microscopic analytical techniques had been given by an experienced analyst and approximately 10% of the samples were independently checked by a second microscopist. Since both the task duration and the exposure level were variable we have chosen to use the cumulative exposure, i.e. the product of exposure level and duration, as an appropriate summary measure of exposure for individual tasks. These data were analysed on the log-scale using a one-way analysis of variance (ANOVA) to investigate, first, the effect of task and then the effect of the behaviour variables. In the analysis replicate samples from the same worker were treated as independent. Linear regression was used to quantify the relationship between log-transformed exposure and the behaviour variables. Student t-tests were used to assess the significance of differences between replicate evaluations of fibre concentration and differences in risk perception between smokers and non-smokers. RESULTS Six asbestos removal teams, with a total of 17 asbestos workers, agreed to participate in this study. The subjects were all male with ages between 19 and 63 years (mean age 38). The duration of employment in asbestos related work ranged from 2 weeks to 10 years, with a mean duration of 4 years. Nine of the workers were employed by a local government organisation and eight were employed by an asbestos removal contractor. The local authority workers were more likely to be smokers (63% versus 22%) and less likely to be clean shaven (44% versus 88%) or check their respirator prior to beginning work (22% versus 38%). Measured fibre concentrations ranged from 0.01 fibres/ml to 4.4 fibres/ml, from 41 measurements at ten sites. Eight samples were not suitable for evaluation because of excessive dust on the filter. Each worker was sampled on between one and six occasions, with the time taken for individuals to complete a given task varying from i2 to 132 min. There was a strong correlation between the two independent evaluations of fibre concentration and a Student t- test showed that they were not significantly different (p > 0.05). In the remainder of this work we have therefore used the average of the individual evaluations. Figure 1 presents the cumulative exposures for three tasks: removal of asbestos containing material, cleaning after removal and combined removal and cleaning (this latter group was necessary because, on occasion, it was not possible to sample the individual tasks). The median cumulative exposure for the removal and combined removal and cleaning were 22 and 36 fibres/ml.min respectively, while the corresponding value for the cleaning task was 2.6 fibres/ ml.min. In an analysis of variance (ANOVA) of the log-transformed cumulative exposures the differences between tasks was found to be significant (p = 0.01). The importance of the dichotomous task-related behaviour variables (i.e. use or no use of power tools, breakage or non-breakage of asbestos containing materials, careful or careless bagging and hurried or unhurried work) was investigated by sequentially adding these terms to the above ANOVA model after adjusting for task. With the addition of the first behaviour variable (i.e. use of power tools) the task ceased to be significant, although the power tool term was highly significant (p = 0.001). Omitting the task term from the analysis and including all behaviour variables showed that both the use of power tools and the careful bagging were significant determinants of cumulative exposure (/xo.ol), although the other two behaviour variables were not significantly related to exposure (p = 0.06). The model indicated that the use of power tools increased cumulative exposure from approximately 17 fibres/ml.min to 119 fibres/ ml.min(i.e. about 7 times greater) and careful bagging of waste would on average reduce cumulative exposures by about a factor of two. There was a difference in the average worker risk perception score between the smokers (mean = 2.8, 10 men) and non-smokers (mean = 4.3, 7 men) studied, i.e. workers with lower risk perception were more likely to be smokers. This difference was not statistically significant in a Student t-test (p = 0.06). Figure 2 shows the average differences in observed behaviour for those men scoring low in the risk perception categorisation (arbitrarily defined as scores between zero and two) and those scoring higher (i.e. 3 to 5 note none of the workers achieved the maximum score of 6). Power tools were used on 37% of occasions when workers in the lower perception category were sampled compared with 16% of the occasions when workers with better risk perception were sampled. This difference was just statistically significant (p = 0.05). None of the other differences were statistically significant. To investigate whether risk perception had any effect on workers exposure we have produced a box plot of the cumulative exposures for each of the aggre-
100 a x c Ulg ft Fig. I. Box plots ot cumulative exposure tor the three tasks monitored (The narrowest part of the box corresponds to the median, the upper and lower boundaries to the 10% and 90% points. Individual data points shown as solid triangles). Fig. 2. The proportion of workers in ri-hlioi) lo iheii: risk perception,»ho»ere observed «; '; upcatit: ix... in fact, the observed tendency is towards slightly higher exposures for workers with better risk percep- tion. gated risk perception categories described above (Figure 3). Clearly, there are no significant differences in the cumulative exposure between the categories and,
A. J. Stewart-Tavlor and J. W. Cherrie 568 o low high Overall Risk Perception Categories Fig. 3. Box plot of cumulative exposure in relation to the workers risk perception (The narrowest part of the box corresponds to the median, the upper and lower boundaries to the 10% and 90% points. Individual data points shown as solid triangles). DISCUSSION In this work it has been necessary to use cumulative exposure (i.e. the product of exposure level and duration) as the main index of exposure. This is because the exposure measurements are derived from activities of variable duration. If, for example, we had chosen only to analyse exposure level then workers who took longer to remove an asbestos board, perhaps because they did not use power tools, may have erroneously been identified as less exposed. Cumulative exposure is clearly the most appropriate task-related measure of exposure (Cherrie, 1996). The key hypothesis, that risk perception is causally related to behaviour and hence exposure is plausible. It is the basis for the recommendations to employers to provide information, instruction and training for their employees in the hazards and risks they experience. For example, in the Control of Asbestos at Work Regulations in Great Britain employers are required to provide adequate information, instruction and training to employees on the risks arising from asbestos exposure and the precautions that require to be taken (HSE, 1988). A corollary to the hypothesis is that changing peoples perceptions of risks can lead to lower exposures and hence lesser risks. The clear association between behaviour and exposure, with lower exposures when workers carried out careful bagging and much higher exposures when using power tools should help focus employee training. However, our attempt to relate behaviour to risk perception was less successful. Careful bagging of waste was slightly less prevalent in the high risk perception category, while breaking insulation and hur- rying were more prevalent in this group. Only the use of power tools was significantly lower in the high risk perception group. There was also a slightly lower perception of the risks involved with asbestos work amongst those who smoked cigarettes compared with non-smokers. If perception of the risks of lung cancer from asbestos exposure does influence workers behaviour then it might be expected that those who were more aware of the risks would be less likely to smoke. However, Li el al. (1983) noted that risk perception appeared to have little influence on US asbestos workers decisions to stop smoking, with 'saving money' and 'smelling better' as the two most important determinants. Others have noted that workers who are better informed about the risks of solar UV radiation are less likely to protect themselves (Modan and Aziz, 1988). Unfortunately, we were unable to demonstrate any direct link between risk perception and asbestos exposure. This may be because of the relatively small size of our study population and the consequent difficulty in controlling for other variables which may be important exposure determinants, for example the size of the enclosure, the quantity of air extracted from the enclosure etc. McKenna (1988) reviewed the evidence for risk perception in the causation of accidents. He concluded that while risk perception is important it is almost certainly not the sole determining factor directing human behaviour. This is supported by the studies cited above where self image seemed more important in dictating protective behaviour rather than risk perception. Despite this we have shown separate associations between the use of power
Does risk perception affect behaviour and exposure? 569 tools and exposure and the perception of risks and the use of power tools and this is suggestive of a link between risk perception and exposure. Further work is necessary to clarify this position. Acknowledgements We acknowledge the assistance of the managers and asbestos workers whose co-operation enabled this work to be completed. We are also grateful to our colleagues at the Institute of Occupational Medicine for their helpful comments and assistance throughout this work. Particular thanks are due to Dr Anne Soutar for her advice on the statistical analysis of our data. REFERENCES Cherrie, J. W. (1996) Are task-based exposure levels a valuable index of exposure for epidemiology? (Letter). Annals of Occupational Hygiene 40, 715-717. HSE (1990) Asbestos fibres in air: light microscope methods for use with the Control of Asbestos at Work Regulations. (MDHS 39/3). HMSO, London. HSE (1988) The Control of Asbestos at Work Regulations (SI 1987 No. 2115). HMSO, London. Li. V. C. Kim. Y. J.. Terry. P. B., Cuthie, J. C, Roter, D., Emmett, E. A., Harvey, A. and Permutt, S. (1983) Behaviour, attitudinal, and physiologic characteristics of smoking and non-smoking asbestos-exposed shipyard workers. Journal of Occupational Medicine 25, 864-870. McKenna, F. P. (1988) What role should the concept of risk play in theories of accident causation. Ergonomics 31,469-484. Modan, M. and Aziz, E. (1988) Negative association between knowledge of health hazard and preventitive individual behaviour from a study of occupational over exposure to the sun. American Journal of Epidemiology 128, 942.