Associations between safety and contingency measures and occupational accidents on offshore petroleum platforms

OBJECTIVES - The primary purpose of this study was to determine the association between organiza tional factors and accidents among employees on the Norwegian continental shelf. METHODS - A self-administered survey among offshore petroleum personnel was conducted in the spring of 1990. The survey drew respondents from five companies and eight platforms. The response rate was 92% (N = 915). RESULTS - Substantial correlations were found between management and employee commitment and involvement in safety work, social support, altitudes towards safety measures, the personnel's satis faction-dissatisfaction with the safety and contingency measures and accident frequency. CONCLUSIONS - Employee satisfaction-dissatisfaction was associated with injury rates, and also with other organizational factors. The credibility of the theory that employee perception of safety and con tingency measures mirrors the status of accident prevention work was enhanced.

Employ ee perception of safety and contin genc y measures has been found to be the most important factor for perceived job stress and subjective risk assess ment among personnel on offshore oil platform s in the North Sea (1,2). Job stress and percei ved risk was also significantly correl ated with the number of accidents and near-accidents on such installations. Therefore it was relevant to uncover organ izational and social factors exerting influence on the personnel' s satisfaction-dissatisfaction with safety and contingency measures (3). This information is especially important if these factors also affect accident risk.
Self-administered survey s disclose the respondents ' subjective evaluations. One can then ask how consistently do these evaluat ions portray the real status of accident prevention work. Dunham (4) showed that employee assessments of job characteristics were consistent with "objecti ve" evaluati ons of the same factors. Likewi se, emplo yee and supervisor assessments of job characteristics are significantly correlated . The perception of jo b character istics reflects the type of job and the technology and position of the respond ent (4)(5)(6)(7)(8). If the respondents perceive safety and contingency measures in a "realistic" way, greater dissatisfaction on oil platforms should be ex-

128
pected where the injury frequency rate is high ("high injury platforms") than on platform s where the rate is low ("low injury platform s"). In contrast, the respondents' own experience of accidents can also influence their perception of these factors.
The empirical consequences will be the same both when empl oyees perceive safety and continge ncy measures in a "realistic" way and when experience with accidents shapes their perceptions. For practical reasons it can be difficult to condu ct controlled experiments to test these attitudes. However, by comparing satisfaction-dissatisfaction on low-versus high-injury platforms for personnel having the same injury experience, one may be able to enhance the credibility of the theory that employ ee perception of safety and contingency measures "realistically" mirrors the status of the accident preventi on work at the sites in question. Separate analyses must therefore be conducted of employees who them selves have been injured and of those who have never suffered an injury.
If differ ences are found in satisfaction-dissatisfaction on low-injury and high-injury platforms in groups of personnel having the same injur y experience, the credibility of the theory that the personnel 's subjec tive assess ment reflects the true status of accident prevention is heightened. Therefore the satisfaction-di ssatisfaction of workers without record s of injury (ie, those who themsel ves have not experienced an accident when conducting their work) on platforms having low versus high accident experiences must be compared with the satisfaction-dissat-isfacti on of workers with records of injur y (ic, who them selves have experi enced an accident ) on platform s having low versus high accid ent exposure. It is important to analyze these relations in order to be able to outline efficient preventive measures and promote safety.
The spec ific objective of this paper was thus to examine how consistentl y employee evaluations of safety and contingency measures portray the real status of accident prevention measure s on board by analyzin g the relations between the empl oyees' evaluation s of safety and contin gency measures in relation to their accid ent experience.

Sample
A sample intend ed to be representative of the offshore petroleum industry on the Norwegian Continental Shelf was selected. It came from eight platforms and five companies. The response rate was 92% (N =9) 5). (See also reference 3.)

Measures
The respondents were asked to assess their satisfa ction with the safety and cont ingency factors on the oil platform on which they worked, management and emplo yee commitment , involvement in safety work, soc ial support from management, superv isors and colleag ues, and their own attitudes towa rds safe ty measures. Fifty test items were used to measure these aspects. (See reference 3 for a de scription of these indices .) In addition, the respondents' own experience with occu pational accidents and "near-accident s" was measured.
The respondents were asked to report whether they had ex perienced accidents in their time offshore or not and whether they had been exposed "many times," "o nce," or "never" to a situation or incident which could easily have led to injury.
The ratio between self-reported accidents and "ncar-accidents " was I:3. Accident figures published annually by the Norwegian Petroleum Directorate (9) and the respondent' s self-reported accident frequency were co mpared. Accid ent figures, as well as self-reported frequencies, were given range values separately for maintenance and construction personnel , operator personnel, drillers , and techni cal-mech anical personnel, and the range values were identical for statistic al figure s from the Norwegian Petroleum Directorate and self-reported range value s. The self-reported data were found to be valid ( 10).

Analysis
Because of the great number of single items used in the measurement, it was first considered nece ssary to replace single test items by indices which depicted the underlying structure of the evaluations (belief For each individu al the scores on test items belonging to a certain index were then added without differential weighting, and a score for each dimension was dete rmined. For testing the differences in significance due to the responde nts ' injury experience (ie, whether or not they them selves had been injured), the chi-square test and the t-test were used.

Results
The employees' degree of satisfaction-dissatisfaction with safety and contin gency measures was dependent on whether they them selves had been injured while carrying out his work duties. Tho se who had never been inju red were compared with those who had been injured. Th e differences were significant for protective measures (t = 3.61 , N = 857, P = 0.(00), instructions (t =3.70, N =859, P =0.000), and training (t =3.30, N = 863, t =0.00 I). Those who themselves had been inj ured were more dissatisfied than those who had never been injured. The results can be explained as follo ws: (i) a respondent's own injury experience was significantly correlated with dissatisfaction and (ii) the fact that safety and contingency measu res prevent accidents brings about satisfaction with these cond itions. The personnel ' s percept ion of the situation is "realistic" and mirrors the true status of the acc ident prevention work on the platform.
The number of accide nts varied due to platform (X 2 = 14.25, degr ees of freedom (df) =7/884, P = 0.(47). Satisfaction-dissatisfaction among the personnel on platforms where the number of accidents was greater than expected ("high-injury platforms" ) was compared with that among personn el on platforms where the number was lower than expected ("Iowinjury platforms"). The number of platforms was four in both groups. As see n in table I the empl oyees were more dissatisfied with the safety and contingency measures on the high-injury platforms than they were on the low-injury platforms. Separ ate analyses were conducted of empl oyees who themselves had been injured and those who had never suffered an injury durin g their time offshore. Satisfaction among per sonnel with the same injur y experie nce on low-injury and high-injury platform s was compared. Among employees who had never experienced an accident themselves, the differences in satisfaction-dissatisfaction were significant due to platform .
However, in the group of persons who had been injured, there were no statistically significant differences due to platform category (table 2). There can be various reasons for this findin g. The personnel's perceptions could have been closer to the "objec tive" risk among the uninjured employees than amon g the injured ones. Hence, the respondents could have been Scand J Work Environ Health 1994, vol 20, no 2 a "Not satisfied" ="neutral," "dissatisfied," and "strongly dissatisfied". satisfied primarily because the platform had given high priority to safety and contingency measures and not because of their own injury experiences. Since the uninjured employees know the dangers, they are able to tell which measures are necessary for successful accident prevention. They are also able to avoid dangerous situations. Consequently they are not so frequently subject to injury as those whose perceptions are less "objective." The personnel's assessments mirror the true situation in both the group with injuries and that without. An oil platform is a complex organization employing several personnel groups engaged in various duties. Accident frequency and objective risk depend on the personnel group and employment conditions (11, 12). Drilling personnel, maintenance and construction personnel, and technical-mechanical personnel experienced the greatest risk ("high-risk personnel"), while administration, production, and catering personnel experienced the smallest objective risk ("low-risk personnel"). Contractor personnel experienced greater risk than operator personnel.
If the percentage of the persons in the group of high-risk personnel who are operators and contrac-130 tors is the same on both high-and low-injury platforms, the assumption that the assessments are closer to "objective" risk estimates also in these personnel groups gains greater credibility. The chi-square test confirmed this possibility (X 2 = 2.12, df 1, not significant). Nor was the group of injured personnel expected to have any differences in the number of persons belonging to high-risk and low-risk groups both on low-injury and high-injury platforms. This expectation was also confirmed (X 2 = 4.41, df 6, not significant). The results show that (i) the satisfactiondissatisfaction with safety and contingency measures among the personnel were "realistic" (ie, they mirrored the true status of the accident prevention work on the platform and (ii) the conditions affected the number of occupational injuries.

Discussion
This study compared satisfaction-dissatisfaction on low-versus high-injury oil platforms. The credibility was heightened of the theory that employee perception of safety and contingency measures "realistically" mirrors the status of accident prevention work on oil platforms.
When matched pairs of the management of manufacturing firms have been compared (13)(14)(15)(16), commitment and involvement has been the factor most frequently found to characterize firms achieving low accident figures in comparison with firms with high accident rates. In these studies matched pairs of companies were selected so that two members of each pair were comparable. Both firms in each of the pairs were located in the same area, were in the same in-du stry, and were approximately of the sa me size. However, th e firms in each pai r had marked d iffere nces in acc ide nt freq ue ncy. Greater managem ent co m m itment and in vol vem en t in safe ty matt ers was found to be re late d to lower ac cident rat es. Oth er imp ortant fac tors found to se para te lo w-injury and high-injury firm s were work-for ce stability, string ent housekeeping, contro l and in spection routines, formalized accident reporting syste m s, and frequent positive contact between employees and supervisors re garding safety matters. (See also refere nce 17 for a re vie w of the lit erature. ) In the se stud ies lo w versus high accident rates wer e cons idered an effec t of the safety program factors. The studies d id not full y meet the demands for inferrin g caus a l rel ati on s.
Ho wever, to study th ese fa ctors "experimentally" gives ri se in most cases to both practical a nd ethica l pr oblems ( 18,19) that m ak e suc h a de si gn d iffic ult or impossibl e to carry out.
In the present study it was found that empl oyee perception of greater management commitmen t and social su pport we re associated with lower inj ury ra tes. In ad di tio n, the va lid ity of e mp loyee sc lf-report data was es ta blis hed through a comparison of employees' reports of safet y measures to injury rates . Improvin g the fac tors contributing to the variance in sa tis fac tion-dissatisfac tio n w ith th e safe ty an d co nti ngency measures will there fore probabl y also lead to real improve ments in these measures. Th ere is reason to believe th at any improvement of the real status will affect the accident fr equency rate and reduce the number of ac cide nts taking place.