On the other hand, in certain heavy jobs with high numbers of persons reporting long-term illnesses it is rare for individuals to blame their health problems on ...
Mats Thorslund, Bo Warneryd and Piroska Ostlin The work-relatedness of disease: workers' own assessment
Abstract In the routine compilation of statistics on the relation between occupation and ill-health it is usually possible to demonstrate links to certain physically demanding occupations. However, studies of the causal links between work and ill-health present problems. Besides the difficulties that often occur in collecting information on present and previous occupations, exposure periods, latency, duration, etc, the analysis and interpretations themselves are complicated. The aim of this paper is to study the interviewees' own views on whether or not any ill-health they may suffer could be caused by any particular job they may hold or have held in the past. Does one get the same picture of the relationship between occupation and ill-health as when routinely compiling statistics? The study population comprised 12,664 men and 12,942 women between 25 and 75 years of age. For all respondents, detailed occupational histories were recorded The results indicated that health problems are more common among persons who have or have had physically demanding work. On the other hand, in certain heavy jobs with high numbers of persons reporting long-term illnesses it is rare for individuals to blame their health problems on working conditions. Conversely, other occupations have relatively low numbers of persons with long-term illnesses, yet high proportions of these attribute their health problems to their work.
Introduction
In Sweden, information about people's occupations is regularly collected by means of the Population and Housing Census (FOB). This occupational information can then be combined with other registers containing Sociology of Health & Illness Vol. 14 No. 1 1992 ISSN 0141-9889
58 Mats Thorslund, Bo Warneryd and Piroska Ostlin information on morbidity and mortality. For example, the individual's occupation must always be given when work injuries are registered, while information on jobs is also held by the social insurance offices which deal with the payment of sick leave. Thus, there is extensive information held on the links between occupation and ill-health. The results clearly demonstrate that people in certain occupations have more accidents than others (Larsson 1989). Certain illnesses are more common than others among people in particular occupations (Ostlin, Lindberg and Thorslund 1985). Sick leave varies from one occupational field to another (Wikman 1989). Studies of the relationship between occupation and ill-health are often based on present occupation and present ill-health. This is open to serious criticism since it does not control for previous exposure to a harmful work environment. However, studies of the relationship between previous occupation and present state of ill-health have also been carried out. For example, persons who, according to census data, worked in certain types of occupation at a certain point in time may have a higher mortality over a subsequent 10-year period than persons who at the same point in time were occupied in other types of work. There is thus also a demonstrated connection between present health and previous occupation. Even if the connections between previous occupation and present state of health are, statistically, indisputable, it is often far less easy to draw conclusions about causal links. One point of departure for studies of the link between occupation and ill-health is that this link is a reflection of that between work environment and ill-health. This means that information about previous and present employment gives at least indirect information about previous and present work environments. However, occupation is not only an indicator of work environment. Persons all holding the same type of job at a particular point in time also frequently have a similar educational background, approximately the same income, and sometimes also typical consumption and behavioural patterns in terms of food preferences, lifestyle, and alcohol and tobacco consumption (Lundberg 1990). In other words, they resemble each other with respect to variables which in various ways are usually linked with state of health. A further difficulty encountered in the interpretation of possible links between present occupation and state of health is that typical 'negative career patterns' can appear. In other words, after a long period in a physically demanding job a person may no longer be able to meet these physical demands and may be 'forced' to switch to less demanding work. Such 'lighter' retreat jobs can thus demonstrate an 'undeservedly' high incidence of certain illnesses (see, eg. Fox and Collier 1976, Olsen and Sabroe 1976 and Vinni and Hakama 1980). Further, certain health problems may only manifest themselves after a long delay, often of many years. Another, so-called healthy worker, effect concerns health-related selection into jobs. In the literature there are many examples of studies
The work-relatedness of disease 59 which have attempted to examine these selection effects (For an overview, see eg Ostlin 1989a). A further difficulty in interpreting possible connections between previous occupation and present state of health is that the information on occupation is sometimes an unreliable indicator of the type of work environment to which the individual has been exposed. Not everyone with the same type of work has had the same exposure to various work environment factors. For a number of years we have been running a project to study and analyse the relationship between occupation (previous and at time of survey) and state of health for a representative sample of just over 25,000 persons in Sweden. We have tried, among other things, to study the consequences of various selection effects into and out of jobs over the course of a working life (Ostlin 1988, Ostlin and Thorslund 1988, Ostlin 1989b). We have also tried to see how the source of the occupational codes - here either census data or retrospective survey data - can influence the picture of the relationship between previous occupation and health (Ostlin, Warneryd and Thorslund 1990, Warneryd, Thorslund and Ostlin 1990). Also, mortality has been followed up (Ostlin 1990). Studies of the causal links between work and ill-health present certain problems. When analysing the relationship between earlier occupational exposure and present health, other variables have also to be taken into consideration. Latency, the period between the onset of the exposure and disease detection, is only one example. Another variable that has to be considered is the duration of the disease studied. And quite apart from the fact that collecting information on previous occupation, exposure period, latency, duration etc, is not entirely problem-free, the analyses and interpretations themselves are complicated. The aim of the present paper is to look at the interviewees' own views on whether or not any ill-health they might suffer might have been caused by any particular job they may hold or have held in the past. Does one get the same picture of the relationship between occupation and ill-health as when routinely compiling statistics? Even if certain types of illness cannot be pinpointed by the persons themselves, and even if they are not always able clearly to analyse cause and effect, it can nevertheless be of interest to identify which illnesses workers themselves most often consider to be caused by working conditions. It is interesting, equally, to ascertain whether certain occupational groups attribute their illnesses to their work more often than others. Material and methods
The study was based on data collected from the Statistics Sweden Survey of Living Conditions, which takes the form of interviews with about 8,000
60 Mats Thorslund, Bo Warneryd and Piroska Ostlin people every year. The subjects are chosen randomly and constitute a representative sample of the Swedish population. They are asked a great number of questions about their circumstances (eg housing, health, occupation, work environment, leisure time, etc). The development of these surveys as well as issues of quality control are discussed elsewhere (Thorslund and Warneryd 1985, Vogel et al 1988). During the years 1977 and 1979-81, a detailed recording of the occupational histories of 25,000 people was carried out. All jobs that were held for at least two years were recorded, together with detailed accounts of the respondents' current working conditions. Furthermore, the respondents were asked to report the length of each employment. The occupations were coded according to the Nordic Occupational Classification (NYK), which follows the recommendations of the 3-digit Intemational Classifications (ISCO). NYK comprise three levels. The most general level, the 1-digit level, contains nine occupational groups (Table 1). On the 2-digit level there are 58 groups and on the 3-digit level 282 groups. For example, a dental nurse is classified on the 3-digit level as 044, on the 2-digit level in group 04, 'Health and nursing work' and on the 1-digit level she is classified into group 0, 'Professional, technical and related work'. Table 1 Classification of occupations. 1-digit level 0 1 2 3 4 5 6 7-8 9
Professional, technical and related work Administration and managerial work Bookkeeping and clerical work Sales work Agriculture, forestry andfishingwork Mining and quarrying work Transport and communications work Production work Service work
State of health was measured by means of self-reported morbidity. People were classified as 'ill' if they answered 'yes' to either of the questions 'Do you suffer from any long-term illness, the after-effects of an accident, any disability or other ailment?' or 'Do you regularly take medicine for anything (else)?' If the answer was yes, the interviewee was asked to report the kind of illness he/she was suffering from as accurately as possible. There were also follow-up questions, with which it was possible to determine the amount of pain and suffering caused by the illness. The reported illnesses were coded by a group of trained office coders of Statistics Sweden. The symptoms are matched with the International Classification of Disease, 8th revision (ICD).
The work-relatedness of disease 61 For each illness reported, the interviewee was asked to say whether he/ she considered that the condition was due to any particular working conditions they had experienced. Those who felt that there was a causal link were then asked to describe the work in question. The occupations given by the interviewees were then classified according to the NYK. Results
Of the entire study population, 40.2 per cent of the men and 41.1 per cent of the women said that they suffered from a long-term illness. TTie most common groups of illnesses were diseases of the musculoskeletal system, the circulatory system, the nervous system and the sense organs. Not all groups of illnesses are discussed in the present article; those focused on here are the ones which may be work-related and which are relatively common in the population. It should be noted that illnesses which were seen by the interviewees themselves as presenting only minor problems are not included. Tables 2 and 3 show the prevalence of the diagnostic groups included for analysis. Almost one quarter of both men and women said that they had one or more of these illnesses. Most common were diseases of the musculoskeletal system and of the circulatory system.
Table 2 Long-term illness and identification of working conditions (%) Men (n=I2644) Suffering from the illness b. a. Total Identifies working conditions
Diseases of the respiratory system Diseases of the circulatory system Skin diseases Diseases of the nervous system and sensory organs Diseases of the musculo-skeletal system At least one of the above illnesses
*±. 100 a
c. Percentage of those with illness who identify working conditions*
2.8 6.9 2.1 5.4
0.9 1.9 0.5 2.1
32 28 26 39
11.2
7.4
66
23.3
11.8
51
62 Mats Thorslund, Bo Warneryd and Piroska Ostlin Table 3 Long-term illness and identification of working conditions (%) Women (n=12942) Suffering from the illness
Diseases of the respiratory system Diseases of the circulatory system Skin diseases Diseases of the nervous system and sensory organs Diseases of the musculo-skeletal system At least one of the above illnesses
b. a. Total Identifies working conditions
c. Percentage of those with illness who identify working conditions*
2.6 8.1 2.8 4.8
0.3 1.3 0.4 0.5
13 16 15 11
11.9
6.2
52
24.4
8.3
34
_b 100 a
A relatively large proportion of all those who reported some illness about half of the men and one third of the women - felt that their illness could be attributed to previous or present working conditions. This was particularly the case for diseases of the musculoskeletal system, where twothirds of the men and about half of the women saw their health complaints as caused by working conditions. As to the NYK-classification, one may note that at the 1-digit level, the men reporting health problems mainly gave occupations in the group 'Production work' (33.9 per cent of all musculoskeletai morbidity), followed by those belonging to the group 'Agricultural, forestry and fishing work' (19 per cent). At the 2-digit level, jobs most frequently given were those in the group 'Forestry Worker', followed by 'Engineering and building metal work'. At the 3-digit level, the largest group was among 'Forest workers and log drivers'. Among women, jobs belonging to the 1-digit level category 'Service work' (14.3 per cent) were most frequently named, with those in the group 'Health and nursing work' being most common at 2-digit level (7.5 per cent). Most of the latter were practical nurses and hospital orderlies, who also had the second highest frequency at the 3-digit level. The most frequently named occupation at 3-digit level was 'Agricultural and livestock workers'. Cleaning was also a job frequently identified by the women with health problems. Nearly 40 per cent of the men who reported illnesses belonging to the
The work-relatedness of disease 63 diagnostic group 'Diseases of the nervous system and sensory organs' attributed their health complaint to their work. This diagnostic group includes diseases of the eye and ear. Jobs within the 1-digit level group 'Production work', (23.8 per cent of all nervous system morbidity) were particularly frequently seen as having caused the illness in question. Next came 'Agricultural, forestry and fishing work' (5.4 per cent). At the 2-digit level, occupations within the group 'Engineering and building metal work' were named most often. At the 3-digit level, all frequencies were low - in no case higher than 20 per cent. Relatively few women with health problems in this diagnostic group made such a causal link. The following analyses used the occupations given in the occupational histories and the NYK classifications at the 2-digit level. Everyone who had worked at least two years in the occupation and/or was engaged in such work at the time of interview was put into a particular group. An individual can therefore appear in more than one occupational group. In the analyses presented here, when or for how long (in excess of the minimum two years) the person held a particular job has not been taken into account. The 2-digit level occupational groups were used for two reasons: partly because of the uneven quality of the NYK classification system, partly to give a sufficient number of observations in the various groups. Control studies of both the census data and the quality of the occupational histories demonstrate that the NYK has an inherent uncertainty or variability (i.e. lack of agreement between two independent but comparable classifications of the same individuals), which is naturally greatest at the 3-digit level and least at the 1-digit level (Warneryd, Thorslund and Ostlin (1991). It also appears that certain groups of occupations are more difficult to classify uniformly than others. Occupations in the NYK groups 1: 'Administrative and managerial work' and 2: 'Book-keeping and clerical work' seem, for example, to pose problems. For Group 1 there is no great difference between the 3-, 2-, and 1-digit levels, while the problems seem manifestly to decrease at the 2- and 1-digit levels in Group 2. A similar tendency can be observed with Group 7-8, 'Production work'. Fig. 1 and 2 give for each viable occupational group (those comprising at least 100 individuals) the percentage of persons who reported any kind of long-term illness at the time of interview, the percentage of those with work-related illnesses, the percentage of those who said that their illness was caused by working conditions, and how common it was for that particular occupation to be seen as having caused illness. Firstly, some general comments about Figs. 1 and 2. Since the occupational groups are based on the occupational histories and present occupational information, the same individual can appear in several places. Each man appears on average 1.8 times, each women 1.5. Another factor to be taken into account in studies of the relationship between occupation and ill-health is that the sex and age distributions in the various occupational groups often vary. The results presented here are
64 Mats Thorslund, Bo Wameryd and Piroska Ostlin broken down according to sex. The fact that the age distribution varies between the groups has not been taken into account here. However, in an earlier report from the project, the total proportion of those with longterm illnesses in the different occupations was presented by an age-atinterview-standardised index (Ostlin, Lindberg and Thorslund 1985). Even if the prevalence varies depending on whether or not standardisation is used, the main structure of the ranking order of the occupations in terms of the proportion of persons reporting some long-term illness is retained. In Fig. 1 one finds, not unexpectedly, 'heavy' occupations at the top. For the men, 'Mining and quarrying workers' and 'Forestry workers' reported most cases of illness belonging to the five chosen diagnostic groups. The proportion of those who blamed working conditions for their illnesses was also highest among these men (just over two-thirds of those who reported any of the illnesses). It can also be noted that the groups 'Agricultural, horticultural and livestock workers' and 'Ships' deck and engine room workers' contained a relatively high proportion of people who blamed working conditions, but in occupations other than their own. 'Agricultural workers', for example, attributed their long-term illnesses to previous employment in the forestry industry, while 'Ships' deck and engine room workers' instead identified a number of occupations on land. For the women (Fig. 2) it is evident that, apart from the agricultural groups, jobs belonging to the 1-digit level group 7-8: 'Production work', feature strongly, both in terms of total morbidity, morbidity from the illnesses looked at here, and the numbers of individuals attributing their health problems to working conditions. However, the total number of those who blamed their working conditions is lower for the women than for the men. Further, certain jobs within the group 'Service work' (codes 91-94) are identified. For both the men and the women, the occupations with the highest proportion of persons reporting long-term illness also contain the highest proportion of persons who attribute their illness to working conditions. In the upper part of both Figs. 1 and 2 approximately two-thirds of the longterm illnesses are categorised as work-related. Further down, where the number of long-term illnesses overall is lower, work-related illnesses account for less than half of the long-term illnesses in some occupations. The number of persons in a particular occupation reporting a long-term illness is proportionally linked to the number of work-related illnesses in that occupation. The innermost bar - giving the proportion of interviewees who attributed their illness to a particular occupation they have held - is also longer for occupations with a high proportion of persons reporting illnesses which are work-related. However, Figs. 1 and 2 show that the relationship between the proportion of those with work-related illnesses and those who name a particular occupation as the cause of their illness is not entirely unequivocal. In certain jobs, for example, illnesses are relatively uncommon overall, yet among those (few) persons in these jobs
The work-relatedness of disease 65 Mining & quarry UnsiciHad manuai work Matai procaswng Forvstry woHc Agric, hort. & iivastocic woric IMeUi work, angin. & building Painting work Dack & angina, craw WFOOO work
Food procassing work Build. & constfuct. work Monitoring ft mat. handling Padcing ft freight work Train drivars/assists. Traffic suparvision Ciaricai work, etc Printing work Litarary ft artistic work Book kaap. ft casliiar work Govt. iagtei. ft admin. Elactric ft aiactron. work Businass admin. Sni|}s ofncars OthcH' salas work Haatth ft nursing Enginaaring work Othar prof., tachnicai, ate Educational work
Fig. 1 Percentage of those with long-term illness and of those who blame their illness/injury on their occupation. Men.
The total peroentags reporting long-tenn Mness Percentage wtth tong^enn wofk-related Mlness
Peioentage identilying thsa particular occupation Percentage who reply that their Wnees is caused by wortt conditions
Legend to Figs. 1 and 2.
66 Mats Thorslund, Bo Wameryd and Piroska Ostlin Textile work Agric., hort. & livestock work Printing work Agric., hort. & forest, management Sewing work Food processing work Waiters Metal work, engin. ft building Housekeeping, etc Caretaking ft cleaning work Train drivers/assists. Wholes, ft retail proprietors Electric, ft electron, work Packing ft freight work Other sales work Other service work Po^ ft tele, communication work Literary and artistic work Book keep, ft cashier, work Health ft nursing Clerical work, etc Engineering work Other prof., technical, etc Educational work Chemical ft physical work
30
40
50
60
Fig. 2. Percentage of those with long-term illness and of those who blame their illness/injury on their occupation, Women.
who do report an illness, it is nevertheless common in certain cases to give their occupation as a cause. Tables 4 and 5 show those occupations where it was either common or uncommon for persons with work-related illnesses also to identify that particular occupation as the cause of their illness. Relatively many of those who at interview reported a long-term illness (i.e. a work-related illness) attributed to a present or previous occupation (Table 4). The 'traditional heavy' jobs such as mining and quarrying, building, painting, forestry, food processing, were identified most by the men. However, as we establish below, there are other physically demanding jobs which are seldom named - in spite of the fact that many people who are or have been employed in such occupations report long-term illnesses. Looking at the different illnesses separately one finds much the same picture for diseases of the musculoskeletal system as for work-related illnesses taken together. As for the other less frequent illnesses one may note that neurological diseases are commonly reported by occupational groups 79 and 75. With regard to diseases of the circulatory system no occupational group commonly reports their illness as being caused by their work.
The work-relatedness of disease
67
Table 4 Occupations where it is common/uncommon to identify occupation as cause of illness. Males Occupation code
Occupations often identified^
50 79 78 44 82 11 75 60 40 77 87 64 63 73 86
Mining & quarry Build. & construction work Painting work Forestry work Food processing work Business administration Metalwork, engin. & build. Ships' officers Agric, hort. & forest, manag. Wood work Monitoring & mat. handling Traffic supervision Train drivers/assists Metal processing work Unskilled manual work Occupations seldom identified'^
04 74 61 05 20 29 08 41 10
.
Health & nursing Precision - tool manufact. work Deck & engine crew Educational work Bookkeep. & cashier work Clerical work, etc Literature & artistic Agric, hort. & livestock work Gov. legist. & administration
' Occupations where more than one in four persons with a work-related illness identifies that particular occupation. ^ Occupations where less than one in ten persons with a work-related illness identifies that particular occupation.
Table 5 reveals that a corresponding tendency also holds for women. A number of physically demanding occupations are named as the cause of long-term illnesses or health problems. The same pattern is found when we look at the diseases of the musculoskeletal system separately. It is interesting to note, however, that the occupational field blamed most frequently - health and nursing work - has a relatively low overall percentage of persons with long-term illness, yet that of the latter persons, a high proportion attribute their illness to their work. It is also worth noting that certain occupations seldom identified by men as causing illness are named by women. This is particularly the case for
68 Mats Thorslund, Bo Wameryd and Piroska Ostlin Table 5 Occupations where it is common/uncommon to identify occupation as cause of illness. Females Occupation code
Occupations often identified*
04 40 41 63 75 93
Health & Nursing Agric, hort. & forest, manag. Agric, hort. & livestock work Train drivers/assists Metalwork, engin. & build. Caretaking and cleaning work Occupations seldom identified^
30 01 20
Wholes. & retail proprietors Chemical & physical work Bookkeep. & cashier work
' Occupations where more than one in four persons with a work-related illness identifies that particular occupation. ~ Occupations where less than one in ten persons with a work-related illness identifies that particular occupation.
health and nursing workers and agricultural workers. In all likelihood, these differences reflect the fact that men and women hold different types of jobs within these 'heavy' occupations. Discussion The occupation/ill-health link is well documented in the literature. This is so both for the link between present occupation and ill-health, and for that between earlier occupation and present state of health. On the other hand, causal links are more difficult to study and analyse. Illnesses vary with regard to selection mechanisms, duration, latency, etc (Pearce, Checkoway and Shy 1986). Furthermore, exposure to various work environment factors often varies over time in one and the same job (Koskela et al 1986). Neither can one say that exposure is always the same for all persons who, at a particular time, hold similar jobs in different workplaces - or even in the same workplace. Equally, a possible link between occupation and ill-health is not always a sign of a link between work environment factors and ill-health. The link may be explained by the fact that persons holding the same types of job often also have a similar level of education, similar eating habits and alcohol and tobacco consumption patterns. Analyses, to the extent that it has been possible to gain access to all the relevant variables, soon become very complicated. As an altemative or a complement to these difficult analyses one can also question people who
The work-relatedness of disease 69 themselves work/have worked in various occupations and who possibly have health problems. What do they consider the cause of their ill-health? The results indicate that many persons who report some long-term illness, handicap, after-effects of an accident, etc. attribute their illness/ injury to present or previous working conditions. This is particularly the case for diseases/injuries of the musculoskeletal system. In studies of the relationship between occupation and illness previous and/or present occupation is often related to present ill-health. Physically demanding jobs usually come top. Or to be more precise, health problems are more common among persons who have or have had physically demanding work. This is particularly the case for men. Our research into respondents' own views on which occupation has caused their illness or injury gives a slightly less uniform picture. Among persons who have worked in certain heavy occupations, such as mining and quarrying, it is common both to have a long-term illness and to attribute this to working conditions. On the other hand, in certain other heavy jobs with high numbers of persons reporting long-term illnesses/injuries it is rare for individuals to blame their health problems on working conditions. Examples of such occupations are agricultural, horticultural and livestock workers and ships' deck and engine room workers. Thus, in a routine compilation of statistics on the connections between occupation and illhealth it is possible to demonstrate strong links for these occupations, even though most of their present or past employees do not consider that such a link exists. People tend rather to point to other occupations in which they have been employed. Conversely, other occupations have relatively low numbers of persons with long-term illnesses, yet a higher proportion than usual of these attribute their health problems to their work. Health and nursing work is one such example for women. Relating to other studies, one may note that individuals are not necessarily consistent in their ideas about the causes of illness (Linn, Linn and Stein 1982). In the British Health and Lifestyle Survey it was found that those occupational groups whose work was probably unfavourable to their health were unlikely to offer a perception that this was so (Cox, Blaxter, Buckle, Fenner et al 1987). Most people thought unhealthy habits (smoking, unwise diet, lack of exercise) to be the main cause of bad health. It is possible that Swedish workers have a greater awareness about workrelated health risks compared to British workers. Ouestions regarding working conditions have been increasingly discussed in Sweden during the last 20 years by trade unions, employers, politicians, researchers and in the media. Workers' demands as regards a safe and satisfactory working environment have risen, and have in various ways contributed to better knowledge about risk factors in the work environment. When analysing the results it is important to discuss the quality of the basic data. There are, for example, certain weaknesses in the quality of the
70 Mats Thorslund, Bo Wameryd and Piroska Ostlin occupational histories, the occupational classifications, the self-reported morbidity and the classification of the illnesses. The combined effect of these weaknesses is hard to assess. The interview questions are also open to criticism. One cannot, for example, discount the possibility of the method seeming to be 'leading' - in other words, that it invites respondents to identify certain working conditions as the cause of a health problem. Here it is a question, for example, of working conditions which in the public debate and elsewhere are widely seen as being linked to ill-health. It often seems to be taken for granted that a physically demanding job is often the direct cause of a back complaint. Since back pain is very common this can easily lead people seeking a 'popular' or well-known explanation for their pain - such as a present or former physically demanding job. The extent to which the form of the question can infiuence the tendency to give known or established links ('collective awareness') can be ascertained only through an experimental study in which various question methods are tested. Another factor to be considered is that it can be difficult to maintain a physically demanding job once one has contracted an illness or sustained an injury. For example, it is easier for a university lecturer to continue to work with back pain than it is for a miner. Indeed, in some cases the individual may even answer, when questioned, that he does not have a long-term illness or injury, or that the problem is 'insignificant' and is therefore not included in our analyses. We have not, however, paid special attention to this factor. In certain occupations workers are sometimes concerned that their health may be affected by factors in the workplace. This concern can be caused by earlier studies and observations, yet in many cases there is no factual evidence. In Sweden there have been many cases of workers in a particular workplace complaining about the effects of their working environment on their health, with it not being until much later - in certain cases decades that research findings and the authorities have confirmed their views/ suspicions or demonstrated that they were unfounded. In our opinion, notice should nevertheless be taken of the views of employees on the 'dangers' present in occupations and workplaces. The ideas which long-standing employees have formed about causal links should be used to complement the more traditional analyses of the relationship between occupation and ill-health. Workers' opinions could also be put to greater use as an early warning system for specific action, at least as a starting point for further research and investigations. However, this demands further research and development work in order, among other things, to develop better methods for measurement and analysis. University Hospital, Uppsala; Statistics Sweden, Stockholm; and Karolinska Hospital, Stockholm, Sweden
The work-relatedness of disease 71 References Cox, B.D., Blaxter, M., Buckle, A.L.J., Fenner, J.F. et al (1987) The Health and Lifestyle Survey. London: Health Promotion Research Trust. Fox, A.J. and Collier, P.F. (1976) Low mortality rates in industrial cohort studies due to selection for work and survival in the industry, British J. Preventive and Social Medicine. 30, 225-30. Koskela, R.-S., Korhonen, H., Jarvinen, E., Kolari, J.P., Mutanen, P. (1984) Different stages of disease, changes in heaviness of work and life cycle, Scandinavian J. Work Environment and Health, 10, 451-4. Larsson, J.T. (1989) Allvarliga arbetsolycksfallsskador, (Serious injuries due to accidents at work). Arbeten utsatta for sarskilda halsorisker. Rapport iom arbetsmiljokommissionens kartlaggning. Stockholm: Bilagedel B. Linn. M.W., Linn, B.S. and Stein, S.R. (1982) Beliefs about causes of cancer in cancer patients. Social Science and Medicine. 16, 835. Lundberg, O. (1990) Den ojamlika ohalsan (Inequality in Health), Swedish Institute of Social Research. Thesis No 11, Stockholm. Olsen, J. and Sabroe, S. (1976) Health selection among members of a Danish trade union, Interrmtional J. Epidemiology. 8, 155-9. Pearce, N., Checkoway, H., Shy, C. (1986) Time-related factors as potential confounders and effect modifiers in studies based on an occupational cohort, Scandinavian J. Work Environment and Health. 12, 97-107. Thorslund, M., Warneryd, B. (1985) Methodological research in the Swedish surveys of living conditions. Problems of measurement and data collection. Social Indicators Research, 16, 77-95. Vinni, K., Hakama, M. (1980) Healthy worker effect in the total Finnish population, British J. Industrial Medicine. 37, 180-4. Wikman, A. (1989) Sjukfrinvaro och yrkessjuklighet, (Sick leave and occupational diseases). Arbeten utsatta for sarskilda halsorisker. Rapport inom Arbetsmiljokommissionens kartlaggning. Stockholm: Bilagedel A. Vogel, J., Andersson, L.-G., Davidsson, U. and Hall, L. (1988) Inequality in Sweden. Trends and Current Situation. Living conditions 1975-1985, Stockholm: Statistics Sweden. Warneryd, B., Thorslund, M. and Ostlin, P. (1991) The quality of retrospective questions on occupational history - a comparison between survey and census data, Scandinavian J. Social Medicine. 19, 7-13. Ostlin, P. (1988) Negative health selections into physically light occupations', J. Epidemiology and Community Health, 42, 152-6. Ostlin, P. (1989a) The 'health-related selection effect' on occupational morbidity rates, Scandinavian J. Social Medicine. 17, 365-70. Ostlin, P. (1989b) Occupational Career and Health. Methodological considerations on the healthy worker effects. Acta Universitatis Upsaliensis 224, Dissertation. Uppsala. Ostlin, P. (1990) Occupational history, self-reported chronic illness and mortility: a follow-up of 25,586 Swedish men and women, /. Epidemiology and Community Health. 44, 12-6. Ostlin, P., Lindberg, G. and Thorslund, M. (1985) Yrke och halsa. En studie av sambandet mellan yrkeserfarenhet och sjukdom (Occupation and ill health. A
72 Mats Thorslund, Bo Warneryd and Piroska Ostlin study on the association between occupational experience and disease). Department of Social Medicine, University of Uppsala. Ostlin, P. and Thorslund, M. (1988). Problems with cross-sectional data in research on working environment and health, Scandinavian J. Social Medicine. 16, 13943. Ostlin, P., Warneryd, B. and Thorslund, M. (1990) Should occupational codes be obtained from census or from retrospective survey data in studies on occupational health? Social Indicators Research. 23, 231-46.
