Workplace Safety and Health Research Paper

2.    Short History Of Occupational Safety  And Health

The history of occupational safety and health starts in the fourth  century  BC, when Hippocrates described the occurrence of lead poisoning among miners. Later, another Roman physician, Galen, wrote about various occupational  diseases  (Fraser   1989).  However,  the first written  source  to  mention  a disease as actually being  occupational in  origin  was  Agricola’s  De Re Metallica (1556), which ends with a note on diseases and  accidents  among  miners.  At that  time a woman could be married to as many as seven miners, one after the other.  Eleven years later  (1567) Paracelsus  published Von der Bergsucht und anderen Bergkrankheiten, describing the diseases of smelter workers and metallurgists (Hunter  1955).

Finally,   in  1700,  Ramazzini   published   a  book entitled  De Morbis  Artificum  Diatriba,  in which  he examined  the  health  conditions  of over  50 occupations. Because of his success, he is called the Father of Occupational Medicine.

Observations on occupational diseases did not lead to legislative action  until 1802, when the Health  and Morals of Apprentices Act was confirmed in England. This act was then improved by the Factory Acts in the middle of the nineteenth century. The first compulsory social security system for wage-earners was also set up about  that time, in Germany  in 1883. The Workmen’s Compensation Act  followed  in  Britain  in  1897.  It provided  for  automatic compensation to  an  injured worker.  In  1919 occupational safety  and  health  became  a  truly  worldwide  concern  when  the  International  Labour  Organisation (ILO)  was  founded  in Geneva  to  establish  International Conventions and International Recommendations for the development of standards and procedures  for occupational health and accident prevention  (Fraser  1989).

3.    Theories On Occupational Accidents

A feature common to different definitions of an occupational accident  is the  unexpected  and  unintentional  nature of the incident. When physical injury is added, an occupational accident can be defined as an unexpected  and unintentional series of events leading to the physical injury of a person at work. The other precondition for  an  occupational accident  is a contractual  relationship between  the  employer  and  the employee (Figlio 1985).

Until  the  end  of  the  nineteenth   century,  people thought that accidents were brought on by their sins or were ‘God’s will.’ This idea was based on the fact that most hazards were caused by natural phenomena. People had very limited means with which to resist the forces of nature.  The onset of industrialization in the twentieth century meant that production moved from homes to factories. Since people had built these environments, most  of  the  hazards  were also  manmade.  Thus  the  idea  of accident  prevention  fell on fruitful ground.

3.1    Theories On Accident Causes

The first scientific theory attempting to explain accidents was the theory of accident proneness. Greenwood and Woods (1919) insisted that ‘accidents occur to a limited number  of individuals who have a special susceptibility  to accidents’ (preface). Accident proneness was presumed to be a stable personality trait. The theory assumed that some people have an inherent tendency to be involved in accidents (Shaw and Sichel 1971).

The theory of accident proneness has been criticized on both theoretical and empirical grounds. The studies published  have displayed conceptual  confusion,  how- ever, surrounding the meaning of accident proneness. Accident proneness is understood as a unitary trait, as a  general  characteristic, and  as  an  innate  and  un-modifiable  characteristic, and  it  is thus  used  as  an explanatory concept.  The  statistical  criticism  of the accident proneness  theory  culminates  in the question of whether the entire population has the same risk to be involved in an accident or not. Because the accident proneness theory is based on an incorrect assumption of homogeneous exposure  to risk, it should  be substituted  by the concept  of accident  liability, which is based on the assumption of the inequality of exposure (McKenna 1983).

According to the danger factor theory,  an accident happens  when a worker  and a danger  factor  meet so that the worker injures him/herself. The accident is a stochastic  event,  the  probability of  which  is determined  by  the  number  of  danger  factors.  The  most critical danger factors are those with the highest energy content  (Skiba 1973). This theory  is mostly based on practical  experience.

A more sophisticated version of the danger  factor theory  is the  ergonomic  approach to  the  causes  of accidents. The ergonomic approach assumes that disturbances  in the flow of information increase  the risk  of accident  occurrence.  The  exchange  of information between a worker and his or her environment, or between  a worker  and  co-workers  or supervisors are preconditions to accident avoidance  (Saari 1984).

3.2    Theories On Risk Taking

The recent theories  related  to accidents  are those on risk taking.  They have been developed in research on traffic safety. Most  of the authors  have thought that their theories can be applied to risk-taking situations in work-life as well.

The risk compensation theory assumes that  people have a constant  level of risk  that  they are  ready  to accept. If the risk level decreases due to safety measures (e.g., seat  belts in cars),  people  tend  to  adjust  their goals (e.g., try to arrive sooner at their destination) so that  the risk level returns  to exactly the same level as before. The effects of safety measures are thus eliminated in the long run (O’Neill 1977). Because the assumption of  exact  compensation is criticized,  the concept  of  risk  compensation is often  replaced  by the concept of behavioral  adaptation.

According to the risk homeostasis theory, the risk of traffic accidents is determined  by the homeostatic system, which is analogous  to the room  temperature system. The main assumption of the theory is that the target level of risk is the only factor  determining accident  rates.  Perceived  benefits  and  costs  of  risk taking  and  cautious   behavior   are  the  factors  controlling  the level of target  risk. The risk homeostasis theory   argues  that   the  effect  of  a  safety  measure depends on its motivational nature for the people who use it. The effect of a measure  (e.g., mandatory seat belt  use) which  does  not  motivate  its users  will be eliminated  in the  long run  as the  drivers  take  more risks. However,  safety can be enhanced  by measures that  motivate  people to desire safety (Wilde 1982). If such  motivation can  be  achieved,  the  effect  of  the measure  may  be greater  than  technical  calculations show.

A complementary theory  for the risk homeostasis theory  is the risk motivation theory  (Trimpop  1994). This model assumes that  both  personality  and situational factors influence one’s risk perception, which is divided into physiological, emotional, and cognitive components. Each one has its own target level of risk, which  contributes to  the  total  target  level of  risk. Based on the rational calculation of costs and benefits, a risk taker is motivated to act. These acts influence the environment, which then gives either positive or negative  feedback.  The  actor  can  then  perceive  the risks in a new and better way.

The theory  of zero risk received its name from the assumption that drivers of motor vehicles adapt to the risks involved in driving to the point that they do not generally  feel any  risk  in a traffic  situation  or  their subjective  risk  assessments  approach  zero.  People avoid the feeling of risk just as they avoid pain. There is a risk threshold, however, above which risk is experienced as aversive. A driver tends to satisfy his or her motives in traffic, as well as in other  areas of life. This need to satisfy motives pushes him or her to drive faster  and  more  hazardously. Both  excitatory   and inhibitory  motives influence the decision making of a driver.  The  most  hazardous excitatory  motives  are ‘extra motives’ (i.e., those outside  traffic, such as the saving of time and effort) which prompt  the driver to increase  speed.  The  increased  risk  of an  accident  is related   to   the   strength    of   these   extra   motives (Naatanen and Summala  1976).

If  the  psychological  portrayal of  a  human  being with respect to accident theories is analyzed, a change can be seen. The first theories assumed that  a human being is a passive accident victim who has an inherent tendency  to  be  involved  in  accidents.  In  the  more recent theories, a human being is an actor who actively searches for information from the environment. The theories,  however,  accept that  a human  being is not perfect and does make mistakes and errors. As technology becomes more reliable, a human being will be the greatest risk factor for him  herself.

4.    The Behavior Modification Approach

The approach of behavior  modification is based  on operant   learning   behavior.   Behavior   is  seen  as  a function  of its relation  to various  prior  and  current events.  The preceding  events are called antecedents, and  the  events  following  the  performance are  consequences.  If  an  act  has  increased  in  rate  after  a particular consequence  has  reliably  followed  it,  the consequence  is reinforcing.  But  if the  rate  has  decreased, the consequence is aversive. Last, if a behavior disappears  when  a prior  reinforcing  consequence  is continued,  the  situation  is called extinction  (SulzerAzaroff et al. 1994).

4.1    Interventions

The  training  of workers  is usually  the  first  form  of intervention included in a behavior  modification program. The aim of training  is to  teach  safe work habits. The training must be as illustrative as possible, for example,  showing  photographs of workers  using safety glasses. It is also possible to show unsafe scenes and ask workers to point  out the unsafe actions.  The important point in training is to repeat the safe habits in different ways for the workers. Several studies, however,   have   shown   that   training   alone   is  not sufficient to improve safety performance (SulzerAzaroff et al. 1994).

The alternative method for traditional classroom teaching is an informational safety campaign. The first prerequisite for an effective informational campaign is to set a clear objective with a specific message that  is positive,  simple, plausible,  and  comprehensible. The campaign material should be designed to attract attention. Posters should be placed at crucial points of action,  for example,  on the wall of the canteen  at a construction site. The effect of a campaign  is usually short-lived.  The types of intervention used in informational safety campaigns have not influenced the number  or the seriousness of accidents (Saarela et al. 1989).

The  other  alternative   method   is  to  utilize  small groups  in accident  prevention. This idea is based on experiences from  quality  circles, which are problem- solving  groups   largely  used  in  the  manufacturing industry.   As  in  the  case  of  quality   circles,  small accident prevention  groups consist of members of the same organization meeting regularly  to solve safety- related  problems.   In  a  Finnish   shipyard   the  small groups had from 4 to 13 members, and they met 5 to 11 times during  one year.  The two tasks  of the groups were  to  eliminate  the  obstacles  to  good  order  and safety  and  establish  new  housekeeping   practices— behavior  modification. The  program was successful both in improving housekeeping  and in reducing accidents (Saarela 1990).

Goal setting is the other main method with which to improve  performance. The first requirement for successful goal setting is that  the workers  who are to be the  subjects  of  the  intervention  participate  in  the setting  of  the  goals.  Difficult   goals  inspire  better performance than  easy goals do.  The best goals are difficult,  but  achievable.  The utmost  goal should  be divided into subgoals that are easy to understand. The first  goals  should  be  rather   concrete,  such  as  ‘we should  use hard  hats all the time at the construction site’ (Locke and Latham 1990).

4.2    Positive Reinforcement

Positive reinforcement refers to an increase in the frequency of a response following the presentation of a positive reinforcer. Whether or not a particular event is a positive reinforcer  is determined  empirically. The effectiveness  of  the  reinforcer   depends   on  several factors.  Reinforcement given  immediately  after  the response is more effective than feedback given after a long delay. The greater the amount  of a reinforcer, the more   frequent   the  response.   Reinforcers   that   are highly  valued  lead  to  greater  performance. A  continuous schedule of reinforcement, where a response is reinforced  each  time  it occurs,  is the  most  effective (Kazdin  1975).

Positive  reinforcement  should  be given soon  after the observation of a desired behavior.  It ought  to be given  regularly,   for   example,   once   a  week.  The feedback should be visible and given to all workers at the same time. One way to fulfill these requirements is to place a special information board  on the wall so that  a curve illustrates  the development  of the safety score on the experimental  area from one observation to another. One example of such an information board is  the  TUTTAVA-board developed  at  the  Finnish Institute  of Occupational Health (Fig. 1), TUTTAVA being  a  Finnish   acronym   for  a  program  to  help improve production, quality,  and safety.

One form of positive reinforcement  is the use of a token economy. A token is a tangible object, for example,  a lottery  ticket,  which can  become  a conditioned reinforcer. The use of tokens may shorten the delay between the behavior and the consequence, reinforce  the  behavior  any  time and  any  place, and provide a visible record for evaluation. The weakness of the token economy is that it pictures behavior  as a simple series of stimuli and reactions. The token economy  is,  however,  rather   inexpensive  for  companies to operate  (Brown 1978).

4.3    Measurement  Of Safety  Behavior

It is useful to start  a behavior  modification program with  accident   analysis.   The  analysis   of  accidents during  the previous  three  to five years pinpoints  the hazardous sites  of  the  workplace.   It  is worthwhile classifying  accidents  as  unpreventable and  preventable, that is, as whether or not workers could avoid a similar accident  if they changed  their  behavior.  The researcher  should  then discuss the actions  that  could prevent accidents with the on-site personnel. Using the results  of this analysis,  the researcher  collects 10–20 safety  practices  which  should  be phrased  positively (Komaki  1986).

The  next  step  is to  prepare  the  measurement of safety performance. Generally an observation method is used for this purpose because the object of measurement  is the  external  behavior  of  the  subjects.  The researchers  should  select observation periods  at random (Tarrants 1980). When the observation is done at different times of the day, the workers behave normally and  do  not  avoid  the  most  hazardous tasks.  The observation can  be done  from  a single observation point or by walking through the observation area; the latter method  is more common.

The observation sheet includes at a minimum a list of the desired safety practices. Usually it is constructed with only a dichotomized classification. The behavior of  the  observed  worker  is either  safe or  unsafe,  as defined by the criterion  of behavior  included  on the observation sheet. The third possible category  is ‘not observed,’  under  which the  practice  does  not  occur (Komaki  1986).

One example of an observation method is ELMERI (Fig. 2), ELMERI being the Finnish  acronym  for a workplace  safety and  health  observation method.  It has been developed  through cooperation with safety authorities, safety  managers,  and  safety  representatives   from   selected   workplaces.   The   observation method consists of 26 items, divided into 7 groups: (a) safety  behavior,  (b) order  and  tidiness,  (c) machine safety,  (d)  industrial   hygiene,  (e)  ergonomics,   (f ) walkways,  and  (g) first aid and  fire safety. For  each item, the observer makes a decision as to whether the item is correct according  to the safety criteria defined in  the  handbook of  the  observation  method.   The alternatives for the observer are ‘correct,’ ‘not correct,’ or  ‘not  observed.’  The  ‘not  observed’  alternative  is used, for example, when there are no machines  with permanent means of access in the workplace.  It takes about 15 minutes to observe one work station with the ELMERI method,  and  only a few hours  of training are needed to reach an acceptable level of interobserver reliability. It has been shown that a high safety index from  ELMERI  is  related   to  a  low  accident   rate (Laitinen  et al. 1998).

The walkthrough observation rounds  should be repeated  once a week during  five to ten weeks. The suggested time for each round  is 30–60 minutes.  The minimum  number  of observations during  one observation  round  is 100. The results  of the observations are calculated  as a safety score as follows:

safety   score = correct/(correct + not   correct) × 100. The higher the value of the safety score, the better the safety situation  within the workplace  (Komaki  1986, Tarrants 1980).

The  reliability  of the  observation is based  on  inter-observer agreement.  Therefore,  two observers should independently assess the behavior of the same workers. It is desirable to conduct  reliability checks when both the observers and the observed  workers  are unaware that  checks  are  being  conducted. The  reliability  is calculated  by dividing  the  smaller  frequency  by the bigger  frequency  and  multiplying  by  100. The  reliability should  be at least 80 percent  to be acceptable (Kazdin  1975).

4.4    Experimental  Designs

The effect of the intervention is demonstrated by the experimental  design. It has two tasks. First,  it should show  that   the  change   in  behavior   is  permanent. Second, it should prove that the change in behavior is due to the intervention. There are three main types of experimental  designs.

4.4.1    Reversal Or ABAB Design. The reversal design demonstrates the effect of the intervention by alternating  the presentation and  removal  of the program over  time.  The  purpose  of the  design  is to  demonstrate  a  functional   relationship between  the  target behavior  and the intervention. The first task  in a reversal  design  is  to   measure   the  baseline   rate   of behavior,  which describes  the  behavior  in a normal situation  before the intervention. The baseline period (referred to as phase A) is continued  until the rate of the response  becomes stable.

Then,  in  the  experimental   phase  (referred  to  as phase B), the intervention is carried  out.  This phase continues  until the behavior  reaches a stable level or diverges  clearly  from  the  baseline  level.  Now  the change in behavior is evident, but the cause of change is unclear.

In the reversal phase (phase A) the intervention is withdrawn. The target  behavior  usually returns  to or near  the  original  baseline  level. The  purpose  of the reversal  phase  is to determine  whether  the behavior would  have remained  unchanged if the intervention had not been introduced. When the behavior reverts to the baseline, it is possible to reinstate the intervention (phase B). The design is called the ABAB design because  the phases  A and  B are alternated (Kazdin 1975).

4.4.2    Control  Group Design. Another   way  to  demonstrate  the effect of intervention is the classical experimental   design   with   experimental   and   control groups. The intervention is done with one group (the experimental   group)  and  not  done  with  the  other group  (the  control  group).  Immediately  before  and after  the intervention the behavior  of the workers  is assessed. The basic requirement of the study  design is the similarity of the groups.  The best procedure  to control  for systematic differences between the groups is  to  allocate   the  workers   to  groups   at   random (Kazdin  1975).

At  the  shopfloor level a random classification  of workers is not always possible for production reasons. Then   two   departments,  which   are   as  similar   as possible, are selected as experimental and control groups. In addition to nonrandomness, overlapping of the  information between  the  departments may  be a problem  in the comparison of departments.

4.4.3    Multiple-Baseline   Design.  The   third   experimental  design to  show  the  effect of interventions is the multiple-baseline  design. The aim of this design is to demonstrate that  the change  in behavior  is associated  with the introduction of the contingency  at different  time points.  This design requires  the use of two or more interventions. After the baseline reaches a stable  rate,  one of the interventions is introduced, while baseline conditions  are continued  for other interventions. When  the rates  are stable  once again, the second  intervention is introduced. The effects of each  intervention are  established   only  after  it  has been introduced.

One problem that can occur with the multiplebaseline design is that changes in behavior  take place too early. If behavior  changes before the intervention is introduced, it is not clear whether the intervention is responsible for the changes or not (Kazdin 1975). Another  problem  is that  the  effects of the  different interventions could be confused.

4.5    Analysis Of Intervention Data

Intervention studies often require the use of statistical tests appropriate for single-case experimental  designs. Conventional t and F tests can be used in the statistical analysis of intervention data if the data are not serially dependent. This practice requires that the autocorrelation (i.e., correlation between data points separated by different time intervals) not be significant (Kazdin  1984). The other  statistical  method  is to use tests for time series. The best for a short  time series (n     10) is Kendall’s  tau,  which  analyses  the  mean trend.

It is possible to complement the quantitative data of the  effects  of  the  intervention study  by  qualitative data. Qualitative methods are the most useful for analysing the meanings and symbols of the culture or subculture. In an intervention study, the most useful qualitative   method  is perhaps  a  discourse  analysis. With this method, a researcher can analyse the possible change in the attitudes of the workers participating in the intervention.

In  the  reports   of  behavior   modification  studies, graphs  should  be  used  to  illustrate  both  the  study design and the results. It is easy to show the development of a safety score with a line graph (Gillan et al. 1998). In some cases it is possible to use the same graph to give feedback to workers and to illustrate the results  in the  report,  as the  TUTTAVA curve  does (Fig. 1).

High-quality intervention studies help to avoid the wasting  of time,  money,  and  effort.  Shannon  et al. (1999) have presented eight criteria for an intervention study of high quality. First, the objectives of the study or its underlying  hypotheses  are clearly presented.  In the experimental design the subjects are randomly allocated  to  experimental   and  control  groups.  The study has external validity and therefore its results can be generalized to other workplaces. The outcome measurement with acceptable  reliability  and  validity should be appropriate to the objective of the intervention. Qualitative data can be used to supplement quantitative data.  Diffusion  of intervention manipulation  from  the  experimental   group  to  the  control group  (overlapping)  or improper  randomization are the main threats to internal validity. Appropriate statistical analysis should be applied in the analysis of the  data   set.  The  conclusions   should   address   the program objectives  and  the  limitations  of the  study and  be supported by the analysis.  These criteria  are useful both  in the assessment  of other  studies and in the planning  of one’s own study.

5.    Future Recommendations  For Workplace Safety  And Health

The fatality rate in Finland has fallen from 7.2 to 3.2 deaths  per 100,000 workers  from 1975 to 1994 (Statistics Finland 1997). Most of this positive development is  due   to   technical   solutions   to   prevention    and improved  safety inspections.  Nowadays  most  of the technical solutions are in use and simple accidents can be prevented.  Only the more complex accidents  need to be tackled nowadays; they are due to the interaction between technical failure and human  error.

Occupational accidents  cause disturbances  in production.  When a serious accident occurs, all workers stop their work and go to help. According to Finnish calculations  of accident costs, this participation forms the highest costs to the employer. Some customers therefore demand that the number of occupational accidents should stay under the acceptable limit.

Traffic  research  in Sweden  has  produced  an  idea called  Vision  Zero.  According  to  this  idea,  no  one should be killed or sustain an injury resulting in permanent impairment (Tingvall 1997). Some workplaces, at least in Finland, have already  accepted the idea of Vision Zero.  In other  words,  they follow the principle  that  no accident  is acceptable  and  that  all accidents are preventable.

These trends increase the expectations for safety modification behavior.  Workplaces  require more effective ways of changing the attitudes of workers to a more positive direction and activating them towards work safety. Alternatively,  a deeper understanding of human  behavior  based on the advances of social and behavioral  sciences makes  it possible  to  respond  to these increased requirements.

Bibliography:

1. Brown R L  1978  Adapting   token   economy   systems  in  occupational safety. Accidient Analysis and Prevention 10: 51–60
2. Figlio K 1985 What is an accident? In: Weindling  P (ed.) The Social History of Occupational Health. Croom Helm, London, pp. 180–206
3. Fraser T M  1989 The  Worker  at    Taylor  and  Francis, London
4. Gillan D J, Wickens  C D,  Hollands  J G,  Carswell  C M  1998 Guidelines  for  presenting  quantitative data  in  HFES  Human Factors 40: 28–41
5. Greenwood M, Woods  H M  1919 The  Incidence of Industrial Accidents upon Individuals with Special Reference to Multiple Accidents.  Medical  Research  Committee,  Industrial Fatigue Research  Board,  Report     4.  Her  Majesty’s  Stationery Office, London
6. Hunter D 1955 Diseases of Occupations. Hodder and Stoughton, London
7. Kazdin A E  1975  Behavior Modification  in Applied  Dorsey Press, Homewood, IL
8. Kazdin A E 1984 Statistical analyses for single-case experimental designs. In:   Barlow   D H,   Hersey   M   (eds.)  Single  Case Experimental  Pergamon, New York
9. Komaki J L 1986 Promoting job safety and accident prevention. In:  Cataldo  M F,  Coates  T J  (eds.)  Health  and  A Behavioral Medicine Perspective. Wiley, New York
10. Laitinen H,  Rasa  P-L,  Rasanen   T,  Nykyri  E  1998  An  observation  method  as a  tool  for  workplace  risk  Technology, Law and Insurance 3: 57–61
11. Locke E A, Latham G P 1990 A Theory of Goal Setting & Task Performance. Prentice Hall, Englewood Cliffs, NJ
12. McKenna F P 1983 Accident proneness: A conceptual Accidient Analysis and Prevention 15: 65–71
13. Naatanen R, Summala H 1976 Road-user Behavior and Traffic Accidents. North-Holland, Amsterdam
14. O’Neill B 1977 A  decision-theory model  of  danger    Accidient Analysis and Prevention 9: 157–65
15. Saarela K L  1990  An  intervention  program  utilizing  small groups:  A  comparative study.  Journal  of  Safety   Research 21: 149–56
16. Saarela K L,  Saari  J,  Aaltonen   M  1989  The  effects  of  an informational safety campaign  in the shipbuilding  Journal of Occupational Accidents 10: 255–66
17. Saari J  1984  Accidents,   and   disturbances   in  the   flow  of information. Journal of Occupational Accidents 6: 91–105
18. Shannon H S, Robson L S, Guastello  S J 1999 Methodological criteria  for  evaluating  occupational safety  intervention  Safety  Science 31: 161–79
19. Shaw L, Sichel H S 1971 Accident Proneness. Pergamon Press, Oxford,  UK
20. Skiba R 1973 Die Gefahrentragertheorie. Forschungsbericht Nr. 106. Bundesanstalt fur Arbeitsschutz and  Unfallforschung, Dortmund, Germany
21. Statistics Finland 1997 Accidents  at Work.  1994 SVT Labour market  1997:1. Tilastokeskus, Helsinki, Finland
22. Sulzer-Azaroff B,  Harris   T C,  McCann  K B  1994  Beyond training: Organizational performance management techniques.  Occupational  Medicine:  State   of  the  Art   Reviews 9: 321–39
23. Tarrants W E 1980 The Measurement  of Safety  Garland STPM Press, New York
24. Tingvall C 1997 The Zero Vision. A road transport system free from serious health losses. In: von Holst H, Nygren A, Thord R (eds.) Transportation, Traffic Safety  and Health. Springer, Berlin
25. Trimpop R M 1994 The Psychology  of Risk  Taking  North-Holland, Amsterdam
26. Wilde G J S 1982 The theory of risk homeostasis:  Implications for safety and health.  Risk Analysis 2: 209–25