16
Journal of Occupational Accidents, 3 (1982) 273-288 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands 273 AN EVALUATION OF SAFETY INFORMATION SYSTEMS AT SIX MEDIUM-SIZED AND LARGE FIRMS URBAN KJELL6N Occupational Accident Research Unit, Royal Institute of Technology, S-100 44 Stockholm (Sweden) (Received April 7, 1981; accepted September 23,1981) ABSTRACT Kjellkn, U., 1982. An evaluation of safety information systems at six medium-sized and large firms. Journal of Occupational Accidents, 3: 273-288. A study of safety information systems at six firms representing different activity areas (mine, steel mill, wire-rolling mill, housebuilding, repair work and railway work) is presented here. It seeks to chart and evaluate the methods and routines applied to the collection and use of information about accident risks in production. The research material consisted of documents emanating from safety work and the results of interviews with members of the safety organization. On the basis of this material, the routines used in safety work have been examined and the content of the documents analyzed. The conclusion drawn from this study is that the firms did not produce sufficiently comprehensive data on which to base the systematic pursuit of accident prevention. In addition, the information that was available had not been analyzed and summarized in a way such as to underpin decisions assigning priorities to and following up safety measures. INTRODUCTION According to fundamental principles of safety management, the collection of information about accident risks together with the analysis and summariz- ing of this information comprise two important activities in a system for accident prevention (Heinrich et al., 1980; Tarrants, 1980). Scientific proof of the relationship between the supply of data on accident risks and actual experience of accidents is no more than fragmen- tary (Ellis, 1975). A study of eleven matched pairs of companies found that the factor which was chiefly related to a low accident rate, ranking second in importance to top-m~agement involvement, was the quality of the firm’s internal accident statistics (Simonds and Shafai-Sire, 1977). In other studies employing the same conceptual approach, the investigators found that companies with a low accident rate have been more inclined to inquire into minor injuries and narrow escapes than high accident rate companies 0378-6349/82/0000-0000/$02.75 o 1982 Elsevier Scientific Publishing Company

An evaluation of safety information systems at six medium-sized and large firms

Embed Size (px)

Citation preview

Journal of Occupational Accidents, 3 (1982) 273-288 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

273

AN EVALUATION OF SAFETY INFORMATION SYSTEMS AT SIX MEDIUM-SIZED AND LARGE FIRMS

URBAN KJELL6N

Occupational Accident Research Unit, Royal Institute of Technology, S-100 44 Stockholm (Sweden)

(Received April 7, 1981; accepted September 23,1981)

ABSTRACT

Kjellkn, U., 1982. An evaluation of safety information systems at six medium-sized and large firms. Journal of Occupational Accidents, 3: 273-288.

A study of safety information systems at six firms representing different activity areas (mine, steel mill, wire-rolling mill, housebuilding, repair work and railway work) is presented here. It seeks to chart and evaluate the methods and routines applied to the collection and use of information about accident risks in production.

The research material consisted of documents emanating from safety work and the results of interviews with members of the safety organization. On the basis of this material, the routines used in safety work have been examined and the content of the documents analyzed.

The conclusion drawn from this study is that the firms did not produce sufficiently comprehensive data on which to base the systematic pursuit of accident prevention. In addition, the information that was available had not been analyzed and summarized in a way such as to underpin decisions assigning priorities to and following up safety measures.

INTRODUCTION

According to fundamental principles of safety management, the collection of information about accident risks together with the analysis and summariz- ing of this information comprise two important activities in a system for accident prevention (Heinrich et al., 1980; Tarrants, 1980).

Scientific proof of the relationship between the supply of data on accident risks and actual experience of accidents is no more than fragmen- tary (Ellis, 1975). A study of eleven matched pairs of companies found that the factor which was chiefly related to a low accident rate, ranking second in importance to top-m~agement involvement, was the quality of the firm’s internal accident statistics (Simonds and Shafai-Sire, 1977). In other studies employing the same conceptual approach, the investigators found that companies with a low accident rate have been more inclined to inquire into minor injuries and narrow escapes than high accident rate companies

0378-6349/82/0000-0000/$02.75 o 1982 Elsevier Scientific Publishing Company

274

(Cohen et al., 1975) and had more formal hazard inspection procedures (Smith et al., 1978).

The problems associated with corporate safety information systems have been charted in several studies. A critical examination of the accident reporting systems operating at five steel mills identified problems with the following: loss of information due to defective reporting, the quality of the reports and the spreading of information within the firm. An over~ching problem was that the accident reporting system had not been designed as an information system (Adams and Hartwell, 1977). A study of the accident information systems at two major companies found that information from these systems did not facilitate the development of strategies aimed at accident reduction (Edwards, 1981). According to conclusions drawn from two studies, the one in mining and the other in shipbuiIding, there were several communication systems to deal with safety matters. Of these the daily informal reporting of safety matters was more efficient than the formal system when it came to taking remedial action (Nilsson, 1976).

Under the Swedish Work Environment Law (1978), all firms with more than 50 employees are supposed to have a safety committee composed of persons representing the industrial relations parties. It is called upon to plan and monitor the safety work and, in particular, to keep up with the accident trend. The safety committee therefore has access to all the documented information that is produced with the aim of preventing accidents. This makes it appropriate to study the formal safety information system at Swedish companies with reference to the information that is presented to the safety committee.

This article presents an investigation into the safety information systems at six firms concerned with different activities. It includes an analysis of that information about accident risks to which the safety committees have had access on the basis of a process model of accidents (Kjellen and Larsson, 1981). Also studied were the routines used to collect and analyze this in- formation within the firms and the manner in which the information was used to prevent accidents. The object was to evaluate the in-plant safety information systems. One intended use for the result is to plan change- oriented research in this field,

Basic to the evaluation was the premise that there has to be a formal safety info~ation system in order to guarantee a consistently high standard for the accident-prevention work carried on at every workplace within the firm. The safety information system ought to encompass routines for a systematic documenting, analyzing and summarizing of that information about accident risks that is available within the firm and of routines for using such information to prevent accidents.

METHODS

Firms studied

The investigation builds upon data that were collected in studies of

275

accident risks and safety work at six firms. Results from these studies were presented in so-called corporate reports (Carlsson, 1980a; Carlsson, 198Qb; Harms-Ringdahl, 1980; Kjellkn, 1980; Larsson, 1980a; Nilsson, 1980).

Various criteria were weighted to select the companies for study. The criteria were: (1) different types of productian systems and planning systems were to be represented; (2) the safety organization was to be fully developed; and (3) serious accident risks were to be present. For practical reasons, moreover, it was important for the firms to take a positive view of the investigation.

The firms had between 1700 and 37,000 employees, meaning that they can be characterized as medium-sized or large. Four of the firms were privately owned and two’were state-owned. All of them empfoyed salaried personnel (e.g* safety engineers) who worked full-time with safety issues and all had safety committees.

The following activity areas were investigated: e The productian of raw materials is represented by two mines belonging to

two separate firms. At the one mine about 70 employees were engaged in working silver ore; at the other mine about 140 persons were employed in working ferrous and manganese ore.

* Processing industry is represented by a steel miI1 with about 250 employees,

* Fabrication is represented by a wire-rolling mill employing about 200 persons. The steel mill and the rolling mill belonged to the same firm.

* One-off production is represented by a building site, with the ~onst~~tion firm concerned employing about 275 persons.

* Technical service is represented by a business entity engaged in preventive maintenance and repairs, for instance on customers’ premises; this servicing entity employed about 100 men.

a The service sector is represented by a traffic area with about 1700 employees, under the purview of the Swedish State Railways. This investigation concentrated on the work involved in shunting and materials handling. Serious accident risks were present in all the entities studied. To illustrate,

there is the risk of falling rock inside a mine, the risk of cantact with hot matter (molten steel or ingots) in steel mills and rolling mills, risks of falling to a lower level on a building site and risks of slipping, tripping and getting crushed in railway work. The accident rate for the plants entering into the study was between two and seven times the average for Swedish industry,

Model of the accident sequence

The collection of data and method of analysis were based on a model of the accident sequence (Kjellen and Larsson, 1981). Put simply, the model can be said to consist of two levels: the sequence of events and the underly- ing determining factors.

276

The sequence of events is described as a chain of deviations. A deviation is defined here as an event or a state which varies from a norm for what is the faultless and planned production process. The sequence of events is divided into three phases: (1) an initiatory phase, which covers the period beginning with occurrence of

the first deviation and ending with release of the energy which has caused the injury;

(2) a concluding phase, which covers the period beginning with release of the energy and ending with its incipient effect on the human body;

(3) an injury phase, which covers the period during which energy has been absorbed by the body.

The determining factors are properties of the production system which affect the sequence of events, but which compared with that sequence vary only slowly with time. These comprise the parameters of the production system, that is the technical, physical, organizational, economic, social and individual factors which temporally and logically precede the ongoing, goal- controlled production process but which enter into the production system.

The safety measures which the model implies are of two kinds: (1) the elimination of deviations, and (2) the change of determining factors. The second kind of safety measure may either seek to avert or lessen the likeli- hood that a certain type of deviation will arise, or seek to reduce the consequences of this deviation.

The categories of deviations and determining factors under the model have been chosen and defined in such a way that, once they appear in data col- lection and analysis, they are supposed to direct attention towards relevant measures. For instance, there are categories of deviations which are directly related to different systems of production control. This holds for deviations in the flow of material (material control), deviations in the flow of labour power (personnel control), technical deviation in the man/machine system (technical control) and intersecting or parallel activities (activity control). Deviations in the environment and stationary and personal safety equip- ment enter into the traditional domain of the safety organization.

Collection of data

Source inputs for the investigation consisted of: . documents from the safety work from the period 1977-79, . results of interviews with members of the safety organization concerning

routines in the safety work, l results of supplementary interviews with line supervisors and safety

stewards about accident occurrences, and l responses to a questionnaire addressed to about 300 manual workers on

the job. The following documents were studied in the investigation: 182 accident

reports, 55 reports on near accidents, 91 safety-round protocols, and 33 safety-committee protocols.

The interviews concerning routines in the safety work were conducted with the manager of the safety department in each company plus a sample of supervisors and safety stewards.

As supporting material for the interviews about accidents that had occurred, use was made of the accident report and a questionnaire based on the model of the accident sequence. The interviewee was a supervisor or safety steward who was familiar with conditions at the workplace. The time period between the accident occurrence and the interview varied approximately uniformly between one month and three years.

Analysis

Analysis of the research material took in the following: (1) a description and assessment of routines in the safety work; (2) a content analysis of the different types of documents giving information about accident risks which the safety committee has used as a matter of routine, and a corresponding content analysis of the result of the supplemen- tary interviews.

The description and assessment of safety work routines is based on documents from this work and on interviews with members of the safety organization. The following questions were especially studied in connection with assessment of the routines: l Has one tapped the various sources of information about accident risks to

be found within the firm? l Has there been an under-reporting of accidents and near accidents? l Has one analyzed and summarized the information about accident risks in

such a way that it has made a good base on which to rest decisions to assign priorities and to devise and follow up safety measures?

l Has a check been made to ensure that measures have been taken and to follow up their effects?

l Have the results of the analysis and the summary of information about accident risks been distributed to all the corporate functions concerned? The content analysis sought to find out whether or not the safety

information system had documented the information about accident hazards available at the workplaces.

The content analysis of the accident and near-accident reports were based on the freehand-text description of the sequence of events, the factors which contributed to that sequence and the proposals for action. The text was analyzed with respect to the separate observations on events and conditions that were included in it. These were then coded in accordance with the accident sequence model. The information about the injury phase was excluded from this analysis.

A similar method was used to analyze the content of safety-round and safety-committee protocols. Any comment in these records about the risk of an accident was regarded as a deviation during the initiatory phase or as a determining factor in a possible accident.

278

RESULTS

The safety committee’s sources of information about accident risks

A review of the safety committee protocols showed that the committees routinely used data inputs from the following information sources in their work:

accident reports near-accident reports safety-round protocols. Besides these information sources, the safety committees have been

receiving information about accident risks directly from the committee members.

Routines in safety work

Accident reporting l The inquiry into an accident was done by the supervisor, sometimes to-

gether with the shop steward. This procedure included an interview with the injured worker and, in exceptional cases, with witnesses.

l The result of the inquiry was documented on a form supplied by the National Social Insurance Board for reporting an occupational injury or on e.g. a company form.

l At all firms completion of the inquiry was delayed for up to several weeks. One reason for this could be that the supervisor did not immediate- ly find out about the accident.

l The result of the questionnaire-based study suggests that there has been an under-reporting of accidents amounting to as much as three times the reported rate. It was not possible to check up on the responses given in the questionnaire, which together with the statistical uncertainty makes the result highly tentative.

l The accident reports were gathered at central corporate headquarters (one company excepted) for use in the safety work.

l At four of the firms the safety committee went through the accident reports as a matter of routine. But to judge from entries in the protocols, the taking of remedial action was seldom discussed in connection with this run-through.

l At all firms, routines were lacking for follow-up at the workplace to make sure that remedial action had been taken as a consequence of an accident. There were also, with one exception, no routines for feeding the result of the inquiry into the accident back to the employees at the workplace or to the corporate functions concerned.

l At those firms where the accident reports were gathered at central corporate headquarters, tabulations of the accident statistics were made and presented to the safety committee. The number of accidents and the

279

number of workdays lost over a certain period of time for different production units or occupational groups could be read off from these summaries.

Reporting of near accidents Two of the six firms under study provided for the routine reporting of near accidents. At one firm the near-accident reports were written by the chief safety steward upon notification, mainly by the foremen or supervisors, and at the other firm directly by the foremen or supervisors. Near accidents were reported in the studied plants of the two firms on a scale equivalent to between one quarter and twice the number of accidents during the same period. Both firms had developed routines for disseminating information about near accidents at the workplaces; one of these firms had developed routines for following up the close calls with remedial action.

Safety rounds Safety rounds were made at all firms on a routine basis. The rounds were of two types. They could be done once a week or once a month, with foremen and safety stewards taking part, or they could be done once a quarter or once a year, with members of the safety com- mittee taking part. Three of the firms had “directive” safety rounds, meaning that plans had been made to focus on certain problems in particular. The results of the safety rounds were documented either in special protocols or in a so-called safety log. With one exception, use was made of special routines to ensure that measures decided in the safety rounds came into being. For instance, it was common for the safety committee to go through the safety-round protocols as a matter of routine.

Structure of the information sources

The results of the content analysis of the documents from the safety programme of all firms put together are shown in l’able 1. Corresponding distributions for the separate plants have also been studied. The content analysis has shown that: . Over half of the observations on deviations and determining factors in the

accident reports refer to technical and human deviations during the con- cluding phase (Table 1). The corresponding figure for the individual plant varies between 44% and 71%.

l Just over 60% of the observations in the near-accident reports relate either to technical deviations during the initiatory or concluding phase or to physical/technical determining factors. The corresponding figure for the individual plant varies between 44% and 81%.

280

TABLE 1

Distribution of observations in the accident reports, near-accident reports, safety-round protocols and safety-committee protocols (confined to safety problems taken up by members of safety committees) with reference to type of deviation or determining factor

Type of deviation/determining factor*

Deviations Concluding phase:

Technical Human

Initiatory phase : Flow of material Flow of labour power Flow of information Technical Human Intersecting/parallel activity Environmental factor Stationary guard Personal safety equipment

Determining fat tors Physical/technical Organizational/economic Social/individual

Type of information source:

Accident Near- Safety- Members of reports accident round safety com- (%) reports protocols mittees

(%) (%) (%) -

Total observations (deviations + determining factors)

20 30 0 0 32 6 0 0

7 6 1 3 1 1 0 1 1 3 0 6 7 18 36 16 5 9 1 9 1 3 1 3 8 0 17 11 1 1 39 11 2 0 1 9

13 15 4 18 2 6 0 13 1 3 0 0

379 145 536 70

*See Kjellkn and Larsson, 1981.

Over 90% of the observations in the safety-round protocols on problems and shortcomings in production refer to one of the following types of deviations: technical (initiatory phase), environmental factors, stationary guards. The corresponding figure for the individual plant varies between 82% and 92%. The problems directly brought up at meetings of the safety committees have pretty well covered the various categories of deviations during the initiatory phase and determining factors.

It follows that the tendencies in the research material as a whole are to be found in the research material from each individual plant.

Several of the different categories of deviations and determining factors are well represented in the information coming from one or more of the

283

different information sources. This is true, for instance, of technical and human deviations during the concluding phase as well as technical deviations, environmental factors and deviations in stationary guards during the initiatory phase, and physic~~technic~ dete~ining factors.

Other types of deviations and determining factors are reported much less frequently. This is particularly true for deviations in the flow of labour power, intersecting or parallel activities, and social or individual determining factors. None of these accounted for more than 3% of the observations in any infor- mation source.

It should be pointed out that the figures in Table 1 suggest an objectivity which is not real. The table is the result of a data-collecting and analytical process which contained several elements of subjective judgment (see the section on method). This means that the figures are to be regarded as tendencies and not as exact values.

Comparison between the content of the accident reports and the result of the supplementary in teruiews

The content of the 182 accident reports was compared with the results of the supplementary interviews. This analysis sought to find out whether there is latent information at the workplaces about accidents that have occurred and not been documented in the accident reports. This analysis shows that: l The accident reports contained 195 deviations during the concluding

phase; five more turned up during the interviews. l The accident reports contained 124 deviations during the ,initiatory phase

and 100 more deviations turned up in the course of the interviews, giving an average of 0.6 additional deviations per accident for the interviews.

l The accident reports contained 60 determining factors; 353 more such factors turned up during the interviews, giving 1.9 additional determining factors per accident for the interviews. The analysis has also shown that nearly half (47%) of the accident reports

did not contain any information about deviations during the initiatory phase, whereas the corresponding proportion for the interview results was about a quarter (24%).

For 157 of the 182 accidents there were data on the number of absence days in which they resulted. A linear regression analysis showed no significant correlation between the number of observations about deviations in the accident reports and the natural logarithm of the number of days of absence {coefficient of correlation = 0.1).

The mean number of deviations per accident report for each plant varies between 1.1 and 2.2. A linear correlation analysis has shown a negative correlation between the average number of deviations per accident in the accident reports for each plant and the mean number of additional devia- tions which have turned up in the course of the interviews (correlation coefficient = 0.8). This implies that the interviews have yielded more addi-

tional ~formation about the course of events in those plants where the descriptions in the accident reports were less detailed.

Results of the safety rounds

The content analysis of the safety-round protocols showed that these inspection tours chiefly functioned as a trouble-shooting system with regard to certain types of deviations, namely technical deviations, environmental factors and deviations in stationary guards. An interesting question here is whether or not the safety rounds have resulted in measures intended to prevent these types of deviations from recurring. The remedial actions in that case would take the form of changing the determining factors. The safety-round protocols were devised as decision reports, which made it possible to study this aspect of the problem.

Out of a total of 536 observations in the safety-round protocols dealing with accident risks, 92% (493) pertained to the deviations mentioned above (see Table 1). An analysis of the content of the safety-round protocols has shown that only 6% of these deviations were remedied by changing the determining factors and recording these changes in the protocols. The most common type of remedial action was to eliminate specific deviations, e.g. to repair a malfunctioning item of equipment or to set up a protective guard where there had been none before.

DISCUSSION

The accident reports, the near-accident reports (if any) and the safety- round protocols have been the safety committees’ main sources of informa- tion in planning accident prevention in the six firms studied. Information has also been obtained directly from committee members. The information from these three sources has not been integrated. Accident reporting, near- accident reporting and safety rounds can thus be considered as three separate information systems.

It has not been possible to determine the effect of the respective informa- tion systems on the number and severity of accidents. On the other hand, research has shown strengths and weaknesses in the different information systems that affect the possibility of systematic accident prevention.

Accident reporting

The results indicate the possibility that not all reportable accidents have been reported. However, the indications are that the data that go into the accident reporting system are more reliable than the data that, go into the other two information systems. There is an operationally defined criterion for what is a reportable accident, i.e. an accident resulting in at least one day’s absence from work. Furthermore, the injured person has an economic

283

incentive to notify the National Social Insurance Board which in turn is obliged to ensure that an investigation is carried out at the company;and that this investigation is documented on a special form. This means that the data from the accident reports are suitable for statistical analysis. The results can, for example, be used to consider the generality and severity of accident prob- lems as a basis for giving priority to preventive measures.

Accident reporting has, however, several weaknesses. Accidents are relatively rare events and are thus not suitable as a sole source of information for a planned control of the production system from the safety point of view.

Research has shown that the data documented in the accident reports have been incomplete. The latter part of the sequence of events which leads to bodily injury, i.e. beginning from the moment energy is released, has been fairly well described in the reports. On the other hand, the information given in the accident reports about the early part of the event sequence, and also about the underlying, dete~ining factors, has either been meagre or non- existent. That has been true both of the minor accidents, with perhaps a mere couple of days of absence from work, and of the severe mishaps. Supplementary interviews have shown that there has been latent information at the workplaces about accident occurrences which could underpin the taking of remedial action, yet which have not been documented in the accident reports.

One item of information provided by the initiatory phase of the event sequence is why the injury-inflicting energy has been released. It is therefore reasonable to assume that defective documentation of this phase of the sequence narrows the scope for taking measures to prevent the release of energy. This implies that the accident counter-measures will chiefly focus on stopping the flow of energy before it leads to bodily injury, e.g. through setting up a stationary guard.

There are several plausible explanations to account for incomplete filling-in of the accident reports: l The time that is spent on the inquiry (Kletz, 1979). Many of the inquiries

undertaken at the firms have been summary in nature, typically consisting of the joint filling-in of an accident report form by the supervisor and the injured worker.

l What the investigator knows about the nature of accidents and the possibilities of taking measures (Benner, 1980). Supervisors who have filled in the reports have had at most an elementary training in accident ~v~tigation.

4 The investigator’s ,motivation. One factor which ought to have some importance, is the usefulness of the inquiry as perceived by the investigator (Adams & Hartwell, 1977). A consistent finding was the faulty feedback of information to supervisors and safety stewards telling them how the accident reports have been used in the safety work. The fact that the supervisors have a legal responsib~ity for safety in the workshop must also influence motivation for making exhaustive investigations.

284

l Delaying the implementation of the inquiry. As a consequence of such delay, witnesses tend to forget important information and to have rationalized what they remember (Swain, 1974; FijrsBkringsbranschens Trafiksgkerhetskommitte, 1978). The present study found that in-plant inquiries into accidents were completed several weeks after the accidents occurred. Run-throughs of accident statistics by the safety committees have seldom

resulted in recorded decisions to take remedial action. This implies that the statistics have not been systematically exploited for planning preventive measures. One reason for this may be that the statistics produced at the firms on a routine basis have been descriptive and not suited for the identification of important accident problems or for allocating priorities to safety measures.

Furthermore, the research results have made it evident that the firms have lacked routines to follow up the implementation of measures taken at the workplace in response to accidents that have occurred. This means that, in the normal case, it has been completely left up to supervisors and safety stewards to take remedial action or not.

The firms have also, with one exception, lacked routines for disseminating the results of accident investigations to the workers at the workplace, as well as to the line and staff personnel concerned.

Near-accident reporting

Different researchers note that between 300 and 700 near accidents occur for every accident which results in absence from work (Gustafsson et al., 1970; Heinrich et al., 1980; O’Shell and Bird, 1969). Accident research has also shown that routine reporting of near accidents as well as reporting of such narrow escapes in the form of campaigns can provide valuable inputs for the planning of safety work (Gustafsson et al., 1970; Kjellkn and Baneryd, 1976; O’Shell and Bird, 1969; Swain, 1974). Near accidents, rather than accidents, would thus provide a more suitable source of information for the planned control of production from a safety point of view.

The Swedish Work Environment Law neither requires nor recommends near-accident reporting on a routine basis, whereas the law requires accidents to be reported and recommends safety rounds on a routine basis. This may be the most important reason why only two out of the six firms investigated near accidents on a routine basis.

The results show that the near-accident reporting by these two firms has mainly been used for the collection of data on technical deviations and technical determining factors. It is also indicated that routines for the follow-up of near accidents with preventive measures and the dissemination of information about them worked better than the corresponding routines for accident reporting in these two firms. This may be because the persons involved have found it easier to talk about “close calls” and to discuss ways

285

and means of preventing them than to talk about accidents (Tar-rants, 1963). Near accidents were reported on a scale equivalent to between a quarter

and twice the number of accidents which occurred in a corresponding period. This indicates that substantial information has been lost in reporting near accidents at the workplaces. This also holds true if it is taken into account that practitioners may have a different view from researchers of what is a near accident, Although one might expect a frequent occurrence of near accidents, it has thus not been possible to take advantage of this in the safety work at the two firms because of considerable under-reporting.

The fact that mainly “technical” near accidents have been reported is in contrast to the results of another study on near accidents. This has shown that the wholly dorn~~t type of information that emerged from near- accident reporting had to do with human error (Cheaney, 1980). An explanation of the differences can be that the anonymity of the reporters was not guaranteed, since in those two firms reports were channelled through supervisors. Other investigations have also found that fear of disciplinary action and a negative attitude among supervisors vitiate the chances of getting a system of near-accident reporting to work (Lager, 1914; Swain, 1974).

There are also other factors that may influence willingness to report dif- ferent types of near accident: o Technical failures’are regarded as something to which the individual is

exposed and whose prevention comes under the purview of the line. By contrast, there is a tendency both among supervisors and among blue- collar workers to regard an accident attributable to the “human factor” as something for which the individual is responsible and which cannot be prevented or is hard to prevent (Larsson, 1980b).

* Technical failures are often expensive for the company, and there arc economic reasons for reporting and preventing these failures.

Safety munds

All the firms under study had safety rounds as a matter of routine. In con- trast to accident and near-accident reporting the safety rounds provided data for planned control of the production system from a safety point of view. Most firms had developed routines to ensure the follow-up of measures decided or suggested during safety rounds.

As in the case of near-accident reporting, the safety rounds were selective. The results of the study suggest that the safety rounds chiefly functioned as a system for identification and correction of deviations in the form of hazardous conditions in the production system (chiefly defective safety guards, technical failures, and inadequate cleaning). Only exception~ly were deviations identified which manifested themselves solely in the course of operation, e.g. temporary undermanning or conflicts between work crews, or determining factors such as shortcomings ia design and routine of the pro-

286

duction system. Moreover, all the firms studied lacked routines for analyzing the data from the safety rounds in order to identify the most common faults and to provide against their recurrence.

The total safety information system

According to the results there have been gaps in the information about accident risks to which the safety committees have had access via accident and near-accident reports and safety-round protocols. Both the results of earlier research and of interviews at the workplaces studied indicate that omitting certain types of information about accident risks may explain the deficiencies in the information that has been documented. Further research is needed to clarify more fully which type of information is being suppressed and what consequcences this has for the prevention of accidents.

Several studies have shown that workers command a body of knowledge and experience of great value for the prevention of accidents (Gustafsson et al., 1970; Tar-rants, 1963; Nilsson, 1976). The research results suggest that none of the information sources used by the safety committees has entailed a systematic utilization of workers’ knowledge and experience.

To judge from some of the company studies, moreover, there are various reporting systems operating in-house with information about hazardous deviations that the safety committees have not tapped (Carlsson, 1980a; Carlsson, 1980b; Kjelldn, 1980). Examples of such reporting systems are down-time reports, waste and rework reports, repair orders and personnel reports.

CONCLUSIONS

The study reported here covered workplaces at six medium-sized and large firms, none of them selected at random. It has elucidated a number of general deficiencies in accident-reporting, near-accident reporting and safety-rounds at these firms. The deficiencies are of such significance that the systematic pursuit of accident prevention on the basis of information on accident risks from these systems has not been possible.

The deficiencies can be attributed to economic, methodological, psychological and organizational obstacles.

Operations at the firms’ workplaces have been characterized by a higher accident risk than the average for Swedish industry. Furthermore, the firms have run fully developed safety organizations. It is therefore likely that the safety information systems at these six firms were of a higher standard than in Swedish industry in general. Thus these deficiencies probably occur in many other firms too.

The results from the study imply in particular the following criteria for planning change-oriented research inside firms in the field of safety informa- tion systems:

287

l Accident reporting, near-accident reporting and safety rounds have dif- ferent unique characteristics and potentialities. Information from all three sources should be integrated in the total safety information system.

l The development of methods for data-collection and for analysis, sum- marization and presentation of information should be based on an analysis of the different receivers’ needs for information about accident risks.

l Remedies to remove obstacles to a systematic utilization of workers’ knowledge and experience on accident risks should be introduced. These include improved training in and aids for data-collection, guarantees for the reporter’s anonymity in near-accident reporting and improved feed- back of results from investigations.

REFERENCES

Adams, N. and Hartwell, N., 1977. Accident-reporting systems: a basic problem area in industrial society. J. Occup. Psychol., 50: 285-298.

Benner, L., 1980. Five accident perceptions: their implications for accident investigators.

Hazard Prevention, 16 (11) 16-20. Carlsson, J., 1980a. Accident risks and safety work. 1: a steel mill (in Swedish). Royal

Inst. of Technology, Report No. Trita AOG-0005, Stockholm. Carlsson, J., 1980b. Accident risks and safety work. 2: a wire-rolling mill (in Swedish).

Royal Inst. of Technology, Report No. Trita AOG-0006, Stockholm. Cheaney, E., 1980. Aviation safety reporting system. A progress report. Hazard Preven-

tion, 16 (8) 20-26. Cohen, A., Smith, M. and Cohen, H., 1975. Safety program practices in high vs. low

accident rate companies - an interim report. National Inst. for Occupational Safety and Health, Cincinnati.

Edwards, M., 1981. The design of an accident investigation procedure. Applied

Ergonomics, 12 (2) 111-115. Ellis, L., 1975. A review of research to promote occupational safety. J. of Safety

Research, 7 (4) 180-189. F%Jkringsbranschens Trafiks3ikerhetskommitt8, 1978. TRK:s haverikommission.

Redogiirelse fSr en fiirsiiksverksamhet. TRK rapport No. 1, Stockholm. Gustafsson, L., LagerlSf, E. and Pettersson, B., 1970. Analys av olyckstillbud vid

huggning. Institutionen fiir skogsteknik, Skogshijgskolan, Report No. 37, Garpenberg. Harms-Ringdahl, L., 1980. Accident risks and safety work. 5: mobile repairmen (in

Swedish). Royal Inst. of Technology, Report No. Trita AOG-0009, Stockholm. Heinrich, H., Petersen, D. and Roos, N., 1980. Industrial Accident Prevention. McGraw-

Hill, New York. KjellBn, U., 1980. Accident risks and safety work. 4: house construction (in Swedish).

Royal Inst. of Technology, Report No. Trita AOG-0008, Stockholm. KjellBn, U. and Baneryd, K., 1976. Undersijkning av stijrningar vid tillverkning av

explosiva”mnen - inrapportering, analys och %tgiirder. National Defence Research Inst., Report No. A 20020-Dl, Stockholm.

KjellBn, U. and Larsson, T., 1981. Investigating accidents and reducing risks - a dynamic approach. J. of Occupational Accidents, 3: 129-140.

Kletz, T., 1979. ‘Layered’ accident investigation. Hydrocarbon Processing, November, 373-382.

Lager, C., 1974. Pilot Reliability. Doctoral dissertation, Royal Inst. of Technology, Stockholn Larsson, T., 1980a. Accident risks and safety work. 6 : railway work (in Swedish). Royal

Inst. of Technology, Report No. Trita AOG-0010, Stockholm.

288

Larsson, T., 1980b. Personal communication. Royal Inst. of Technology, Stockholm. Nilsson, B.-C., 1976. Att fijrebygga olycksfall. Sociologiska instituionen, Uppsala

Universitet, Uppsala. Nilsson, B.-C., 1980. Accident risks and safety work. 3: falling stones and scaling in

mining (in Swedish). Royal Inst. of Technology, Report No. Trita AOG-0007, Stockholm.

O’Shell, H. and Bird, F., 1969. Incident recall. National Safety News, October, 58-63. Simonds, R. and Shafai-Sahrai, Y., 1977. Factors apparently affecting injury frequency in

eleven matched pairs of companies. J. of Safety Research, 9 (3) 120-127. Smith, M., Cohen, H., Cohen, A. and Cleveland, R., 1978. Characteristics of successful

safety programs. J. of Safety Research, 10 (1) 5-15. Swain, A., 1974. The Human Element in Systems Safety. Industrial and Commercial

Techniques Ltd., London. Tarrants, W., 1963. An Evaluation of the Critical Incident Technique as a Method for

Identifying Industrial Accident Causal Factors. Unpublished doctoral dissertation, New

York University, New York. Tarrants, W., 1980. The Measurement of Safety Performance. Garland STPM Press,

New York.