Occupational and Environmental Medicine 1997;54:367-375 367

REVIEW Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from

Health and safety problems associated with long working hours: a review of the current position

Anne Spurgeon, J Malcolm Harrington, Cary L Cooper

Abstract and other aspects of the organisational The European Community Directive on and cultural climate. It is concluded that , which should have been there is currently sufficient evidence to implemented in member states of the raise concerns about the risks to health European Community by November 1996, and safety of long working hours. How- contains several requirements related to ever, much more work is required to working hours, including the right of define the level and nature ofthose risks. employees to refuse to work more than 48 hours a week. The United Kingdom gov- (Occup Environ Med 1997;54:367-375) ernment attempted to oppose the Direc- tive, arguing that there is no convincing Keywords: long working hours; health and performance evidence that hours of work should be limited on health and safety grounds. Much of the research in this area has By November 1996, the European Community focused on the problems of shiftworking Directive on Working Time' should have been and previous reviews have therefore implemented by all member states ofthe Euro- tended to emphasise this aspect of work- pean Community. This measure has, as its legal ing hours. However, there is much less basis, Article 11 8A of the Treaty of Rome,

information about the effects of which is intended to encourage proposals for http://oem.bmj.com/ work, which is a central element of the "improvements especially in the working envi- terms ofthe Directive. This paper reviews ronment as regards the health and safety of its the current evidence relating to the poten- workers". The United Kingdom government tial effects on health and performance of attempted to oppose the introduction of the extensions to the normal working day. Directive and challenged the legal basis of the Several gaps in the literature are identi- original proposal arguing that it should be fied. Research to date has been restricted implemented under Employment Law rather to a limited range of health outcomes- than Health and Safety Legislation. The on October 2, 2021 by guest. Protected copyright. namely, mental health and cardiovascular significance of this is that employment law disorders. Other potential effects which requires unanimous agreement for its imple- are normally associated with stress-for mentation, unlike health and safety legislation example, gastrointestinal disorders, which may be passed on majority voting. Institute of musculoskeletal disorders, and problems The United Kingdom's challenge was, how- Occupational Health, associated with of the immune ever, overruled by the European Court of Jus- University of which means that the Directive should Birmingham system, have received little attention. tice, A Spurgeon Also, there have been few systematic now be implemented in the United Kingdom. J M Harrington investigations of performance effects, and However, the British government continues to little consideration of the implications for voice its objections. The basis for such School of occupational exposure limits ofextensions objections seems to be primarily an economic Management, to the working day. Existing data relate one-namely, that certain aspects of the Direc- University of Manchester Institute largely to situations where working hours tive would impose unacceptable restrictions on of Science and exceed 50 a week and there is a lack of employers and hence reduce the competitive- Technology information on hours below this level, ness of British industry. In particular there is C L Cooper which is of direct relevance to the Euro- concern about the section of the Directive pean Community proposal. Finally, it is which gives employees the legal right to refuse Correspondence to: to work more than 48 hours a week. Associated Dr Anne Spurgeon, Institute clear from investigations relating to shift- of Occupational Health, The work that a range ofmodifying factors are with this are the requirements for a daily rest University of Birmingham, likely to influence the level and nature of period of 11 consecutive hours in each 24 hour Edgbaston, Birmingham health and performance outcomes. These period and a minimum weekly rest period of B15 2TT. include the attitudes and motivation ofthe one day. In support of their argument, the Accepted 26 February 1997 people concerned, the job requirements, United Kingdom government maintains that 368 Spurgeon, Harrington, Cooper

there is no convincing evidence to support the European Union who work more than 48 view that long working hours are damaging to hours in a normal week. In a recent survey car- Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from either the health or the safety of employees. ried out by Austin Knight9 which sampled What, therefore, is the current status of the employees from 22 large United Kingdom scientific and medical evidence relating to the organisations, two thirds reported regularly effects ofworking long hours? Is there adequate working more than 40 hours a week and one information available in this area? If so does quarter more than 50 hours a week, despite this support the United Kingdom position, or having contracted hours of between 35 and 37 is it actually the case that there are insufficient hours. A survey carried out by the United Road data on which to base firm conclusions? The Transport Union of the hours and conditions overwhelming emphasis ofresearch in this field of 640 professional lorry drivers in the United to date has been on the effects of shiftworking Kingdom found that the average working week and not surprisingly earlier reviews have was tended to emphasise this.2 'However, although of those sampled 62 hours.10 In the United clearly of relevance and importance to the States the average contracted working week Directive as a whole, most of the evidence was 37.5 hours in 1990, although with derived from the field of shiftwork research is overtime the working week of the average male not strictly applicable to the question of long American worker was nearer 41 hours. During hours or overtime. Shiftworking, or any work the same period the average Japanese worker which involves phase shifting, invariably results worked a contracted 41 hour week, but, in in the disruption of circadian rhythms. This addition, averaged nearly 36 hours of overtime presents a more complex situation than that a month.'1 Japan tends to be regarded as an where stress and fatigue result from an example of the extreme in terms of long hours, extension ofthe working day. Similarly, 12 hour but some of the data recently gathered in the shifts which might, in one sense, be viewed as United Kingdom suggest that conditions for at an extension of the working day should also be least some groups of British workers are not so considered in the category ofshiftworking. The different from those of the average Japanese. systematic and often rotational nature of the A growing trend which is not always work pattern in 12 hour shiftworking, together reflected in official statistics is the tendency for with the different motivational and attitudinal professionals and managers in particular to factors involved, merit separate consideration. work unpaid overtime to deal with excessive (These reviews2 include such shiftwork.) Our workloads. Data to support this are generally review therefore is concerned exclusively with collected and published by Unions. A survey the information available on the health and performance effects of overtime work. This is carried out by the National Association of Teachers in Further and Higher Education, for defined as extending the "normal" work day- that is, beyond the traditional eight hours example, showed that lecturers in further edu- which are usually worked between 9 00 am and cation worked averages of 46.7 to 55.9 hours a

5 00 pm or between 8 00 am and 4 00 pm for week on and off site."2 Another survey by the http://oem.bmj.com/ five or six days a week. It includes overtime Manufacturing, Science and Finance Union of working in the evenings, at weekends, and staff in charities and non-profit making organi- occasionally during holiday periods. sations found that most were working up to 10 Relevant papers for this review were identi- hours a week beyond their contractual obliga- fied from seven databases-namely, ABI/ tions without remuneration.'2 INFORM, CINAHL, CISDOC, HSELINE, Many such surveys also highlight the stress

medline, NIOSHTIC, and PsycLIT covering experienced by workers who reported long on October 2, 2021 by guest. Protected copyright. the period from 1975. These databases had hours. It is arguable that the relation between been identified as being those which were hours of work and ill health is mediated by relevant in the preparation of a previous review stress, in that long hours act both directly as a by JMH.' Only those papers which referred to stressor, in increasing the demands on a person long hours, as defined above, were considered. who attempts to maintain performance levels However, papers concerned with shiftworking in the face of increasing fatigue, and indirectly were also examined for any additional refer- by increasing the time that a worker is exposed ence to long hours in the course ofthe study in to other sources of workplace stress. A high question. We had previously aquired certain level of stress has long been regarded as a con- other papers-namely, those relating to some tributory factor in the development of certain very early studies.4`6 types of psychiatric problems," coronary heart Current working hours disease,' musculoskeletal problems,'5 and There is considerable international variation in symptoms associated with gastrointestinal attitudes and practices relating to working disorders.' All of these have been considered hours, even within the developed world. In the in relation to shiftwork. However, examination United Kingdom the tendency is for employees of the literature relating exclusively to long to work longer hours than comparable groups hours shows that investigations have so far in other European states, averaging a total of largely been limited to mental health and 44.7 working hours a week as opposed, for cardiovascular effects. There is also a limited example, to 39.9 in Germany and 39 in but growing literature on the consequences for Denmark and the Netherlands.7 Recent data8 performance and safety,'7-22 although little show that the United Kingdom has about twice information on the implications for control of as many employees as any other country in the chemical and other exposures.2" Health and safety problems associated with long working hours 369

Mental health disorders Finally a study of British accountants32 iden- Despite evidence of an association between tified a factor of quantitative workload, of shiftworking and increased psychiatric which working long hours was a major compo- Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from morbidity,24 25 particularly among shiftworkers nent, which was significantly related to psycho- who regularly work more than 48 hours a logical wellbeing as measured by two standard week,26 there is a dearth of studies which focus measures.33 34 Two other important factors exclusively on the relations between regular identified in this group were administrative overtime working and mental health status. problems and the level of social support. However, several studies have discussed long Perhaps the country which has most obvi- hours in the context of a range of other poten- ously developed a culture of overtime working tial stressors and their effects on psychological in recent years, and hence concern about its wellbeing. In these studies, psychological potential problems is Japan. A study which health is defined and measured in various ways, examined the determinants of mental health in ranging from specific measures of psychiatric a large sample of factory workers,35 again with status to the incidence of non-specific symp- a standard measure,3 found that working time toms or maladaptive behaviour. of more than nine hours a day alongside In one of the earlier studies, Schmitt et al lifestyle factors such as physical exercise and collected data on unexplained physical symp- nutritional balance was significantly related to toms in over 800 subjects from eight American the overall level ofpsychological distress. Inter- organisations where outbreaks of mass psycho- estingly there were some differences between genic illness had occurred.27 Small but signifi- the sexes in that adverse effects of long hours cant correlations were found between the were more noticeable in women than men, number of hours of overtime worked and the presumably because of the additional domestic number of symptoms reported during an stressors experienced by women. A study in outbreak of illness in the organisation. How- Canada36 highlighted the importance of taking ever, working hours was only one of a range of into account the domestic load ofwomen when variables included, many of which showed a calculating weekly working hours. Female hos- similar relation with symptom reporting. Not pital workers in this study had an average surprisingly the source of the problem was professional working week of just over 32 judged to be multifactorial. Also, the authors hours, but also a domestic working week of 19 note the atypical nature ofthe respondents and hours. Reports of exhaustion and insomnia the lack of comparison with workers in were associated primarily with the duration of non-affected companies. the domestic working week. Hall and More recently several studies have used co-workers37 also noted that the interactive direct measures of mental health with standard effects between "home work" and "paid work" measurement tools. A study ofmusic therapists on psychosomatic symptoms were more pro- working with mentally handicapped children28 nounced for women than for men. examined the relation between several predic- Three studies have examined the prevalence tor variables, including hours worked and the of what are termed "maladaptive behaviours" http://oem.bmj.com/ concept of "burn out", a condition character- in response to stress. This refers to types of ised in the psychological literature as a state of behaviour frequently used as coping strategies mental and physical exhaustion.29 There was which are primarily aimed at avoidance of the no significant association between the number problem. Such strategies are unlikely to result ofhours worked a week and the various indica- in long term solutions and frequently lead to tors of burn out. The response rate was further health problems. Typical examples are misuse. low however-less than 50%-raising the increased smoking and substance on October 2, 2021 by guest. Protected copyright. possibility ofunderreporting, as those failing to Raggatt38 found that in long distance coach cope may have been reluctant to put this on drivers in long driving hours provided record. Also, few of those who replied worked the single best predictor of stimulant use and more than 50 hours a week, most working sleep disturbance. Further, as might be ex- between 40 and 49 hours. It should perhaps be pected, these behaviours were predictive of the noted that nearly 80% of the respondents had level of symptom reporting, particularly stress worked for less than five years as therapists and related problems such as and depres- that indicators of exhaustion were more sion, musculoskeletal symptoms, and of visits pronounced in those who had worked longer, to doctors. Average weekly hours in this sample suggesting the possibility of a cumulative effect were under 47, although the range was 30 to 70 in a job which was clearly emotionally demand- hours. There was no relation between mal- ing. adaptive behaviour and the number of hours A study of the mental health of urban bus driving at night or the number of years driving drivers30 also examined, with a standard meas- coaches, suggesting that the actual time spent ure of psychiatric status,' a range of possible behind the wheel was the important factor, and sources of . Mental health that, in this occupational group, the effect was scores were poor compared with those of a immediate rather than cumulative. Some tenta- male normative sample from general practice, tive support for these findings are provided by and "family problems associated with long the results of a survey commissioned by a large working hours" was identified as a significant United Kingdom telecommunications factor related to occupational stress. Again, company.39 Over the three years of the survey, however, many other work stressors were noted most respondents were working in excess of which seemed to be of equal or greater their weekly contracted hours of around 42 importance for this group. with nearly 18% working more than 50 hours a 370 Spurgeon, Harrington, Cooper

week. In this group, 28% reported heavy excessive as to seriously restrict the time avail- able for sleep. The issue of doctors' hours is drinking, smoking, or drug misuse compared Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from with 8% of those working their contracted clearly one of major importance, particularly in hours only, although these behaviours were not the light of suggestions that doctors should be further defined and may have been interpreted excluded from the terms of the Directive. differently by different respondents. Those However, data from this atypical group are of working longer hours, however, also reported a perhaps limited relevance in examining specifi- higher frequency of a range of stress related cally the effect of overtime worked by employ- symptoms-for example headaches, sleep dis- ees in other sectors, and they will not therefore turbance, and relationship problems. be discussed in detail here. A study carried out in a very different setting Several questions arise when one considers among male members of an Israeli kibbutzim40 only those studies which have been concerned focused particularly on smoking behaviour. with "normal" overtime. Firstly, the relation More than eight hours ofwork a day, and what between hours worked and mental health is was termed "addiction to work", was positively never studied directly and in isolation, but related to smoking intensity and negatively rather within the context of the investigation of related to stopping smoking. Other studies numerous potential sources of occupational have shown that stress contributes to intensity stress, which may together lead to varying of smoking.4' In the Israeli study the main degrees of psychological disturbance. In this source of stress seemed to be long working context long hours have often been found to be hours, although there is perhaps some sugges- an important factor relative to some others tion that personality factors and also social fac- studied, but not necessarily the only or the tors in this highly self selected group, may have most important factor. Further, it is difficult to contributed both to long hours and smoking determine whether long hours are themselves a behaviour. Social and cultural influences may direct source of stress or whether they simply have a significant part to play both in serve to increase the impact of other stressors. determining the hours people choose to work In practice it is difficult to separate the effects and the form of resulting maladaptive behav- of long hours from those of occupational stress iour. Studies in Japan for example, have often in general, as by its very nature, overtime usu- focused on alcohol consumption,42 which ally occurs in situations of work overload. seems to have become an integral part of the Notwithstanding this, another important long hours culture. issue relates to the effects of individual Alongside those studies which have specifi- differences in motivation and working condi- cally assessed mental health status or maladap- tions. Most obviously, sex differences in tive behaviour are some which have examined response to overtime have been highlighted. the relative importance of various factors, Also, Bliese and Halverson4" have drawn atten- including long hours, in producing self re- tion to the importance ofchoice in determining ported stress in different populations. Here an a person's response to long hours. Effects on assumption must be made that the presence of wellbeing may be much less in those electing to http://oem.bmj.com/ stress equates to the presence of mental health work overtime, because of internal commit- problems, although mental health is not meas- ment or enjoyment ofwork, than in those com- ured directly. Most studies treat work overload pelled to do so because of work overload or and long hours as separate factors, although company pressure. They argue that investiga- they might be viewed as overlapping to some tions which rely on the summing of individual extent. Populations in which working long responses may not accurately reflect the

hours has been identified as a significant source relation between hours of work and wellbeing on October 2, 2021 by guest. Protected copyright. of stress include hospital doctors,4' police in the work groups as a whole, as the group will officers in the United Kingdom,4445 and contain members who both work more over- managerial46" and clerical personnel46 in time than others and have varying reasons for America and Japan. Working overtime was doing so. identified as a much greater source of stress by Watanabe et al, for example, examined the managerial and professional groups than by relation between health status and working clerical groups in the American study,46 al- conditions in operators of visual display though no figures for actual hours worked were terminals50 and found an association between given or mention made of home working. In self rated symptoms of depression and long the Japanese study of managerial staff,47 more working hours. However, long hours were also than 40% worked more than 10 hours a day, associated with personality factors (type A with more than 30% of section chiefs working characteristics) and with attitudes to comput- more than 11 hours a day. ers, with those describing themselves as "tech- In two studies of police officers,44 4' no nocentred" working longer than others. In this figures are given for average hours, and it case overtime workers may have been a self should perhaps be borne in mind that in these selected group, making it difficult to draw con- cases shiftworking was a complicating factor, clusions about the direction of any causal rela- which is outside the scope of this review. Simi- tion between long hours and mental health larly, many studies have shown stress problems problems. and poor mental health in junior hospital Work conditions during overtime may differ doctors,48 an occupational group in which considerably from those during normal work- extended hours have long been the norm. ing hours. Haugland,5' in a small scale study of Here, however, the situation is one which stress and performance effects in 26 academic includes night working and hours which are so personnel working beyond the end of the Health and safety problems associated with long working hours 371

normal day, found that stress levels decreased work, a factor which seemed to offer some pro- during overtime. The people themselves attrib- tection. Further evidence of an increased Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from uted this to a period of uninterrupted quiet cardiovascular mortality risk associated with when tasks could be completed quickly and long hours was provided by a United Kingdom efficiently. This raises a wider question of study of employees of the Bell Telephone whether the effects of long hours can be gener- Company during the 1960s who worked full alised across populations with different situa- time and also attended night school.56 tions, demands, and cultural norms. Further, it By contrast the results of a study in Sweden points to the need to investigate in more detail in the 1960s," which analysed a one year inci- the effects, particularly on performance and dence of admission to hospital for myocardial safety, of working overtime on any single day. infarction, showed lower age adjusted inci- Published studies to date have focused on the dence of admission for those men regularly cumulative effects of the total number of extra working at least 10 hours of overtime a week, hours in a working week. compared with other workers. Factors associ- Finally, if one accepts that long hours may ated with the nature of the work seemed to be contribute to occupational stress and its conse- more important than working hours in predict- quences, directly or indirectly, at what level of ing admission to hospital for women. The overtime do effects begin to occur? Overall, the explanation, as in the case of mental health existing data tend to support the view that problems, presumably lies at least in part, in weekly hours which exceed 50 are associated the dual role responsibilities of women. Fur- with increased occupational stress, both in ther support for these conclusions was pro- terms of subjective reports and behavioural vided by data collected 20 years later in responses. However, it should be noted that Sweden in the course of a study to examine most of the studies in which effects were found associations between several aspects of the contained a high percentage of people who work environment and admission to hospital worked well in excess of this. Further, many for .58 For men, long studies have been concerned with somewhat working hours were associated with a low risk unusual or highly self selected groups. The of admission, whereas for women the risk was adoption of 50 hours as a threshold on this increased. Similarly, analysis of morbidity and basis therefore might be viewed as somewhat mortality data in Sweden over a 20 year period arbitrary or perhaps cautious. up to 198359 indicated an association between cardiovascular mortality and overtime worked Cardiovascular disorders in women but not in men. If one assumes that long working hours are a Data collected in the United States in the major source of occupational stress, it is early 1980s60 failed to find any association reasonable to expect an association between between job experiences including working long hours and cardiovascular disorders. The hours and certain risk factors for coronary mechanisms underlying the association be- heart disease, namely blood pressure and tween stress and coronary heart disease are serum cholesterol concentrations. However, http://oem.bmj.com/ thought to relate to multiple and sustained work hours and deadlines were found to increases in heart rate and blood pressure as a contribute significantly to reported stress and result of neuroendocrine activation,52 although stress related behaviours, particularly in men. individual difference in cardiovascular reactiv- The relation between stress and factors such as ity in response to acute stress have long been blood pressure and cholesterol is by no means recognised." In the 1960s, increasing concern straightforward and is complicated by indi-

over the levels of coronary heart disease in vidual and time dependent variation. Lack of on October 2, 2021 by guest. Protected copyright. developed countries produced several studies any direct association is perhaps not surprising. primarily aimed at identifying important risk A more recent study of construction managers factors. Forty years ago the concept of occupa- who regularly worked overtime did in fact tional stress was in its infancy, but fortuitously identify increased cholesterol concentrations in this offers certain advantages in the present this group, who also reported poor psychologi- context. Two studies at that time elected to cal health. A range of other determining factors define and measure occupational stress solely are also likely to have been influential here. in terms of working hours, an approach which A study carried out in the early 1970S62 con- would be unlikely today. cerned specifically with psychosocial factors Russek and Zohman5 found that in 100 and the risk of myocardial infarction was, like coronary patients under the age of 40 the pres- those in the 1960s, somewhat limited in its ence of severe occupational strain-that is, very examination of occupational stressors. This long working hours-was more than four times time however, type of occupation, occupational that in a control group. Seventy one per cent of responsibility, and working hours were consid- the coronary group had for a prolonged period, ered. In a group of 50 patients, 46% worked either worked both days and evenings or had more than 60 hours a week compared with worked more than 60 hours a week. Subse- 26% of the controls. The number working quently Buell and Breslow" analysed Califor- between 50 and 60 hours or working on nian occupational mortality data for a three Sundays was slightly higher in the patient year period and concluded that the mortality group, although the difference here was less risk from coronary heart disease was greater in obvious. In this study, several behavioural and men under the age of 44 who worked more lifestyle factors were also considered which than 48 hours a week in certain occupations, suggested that those electing to work long notably those not involving heavy physical hours might be a self selected group in terms of 372 Spurgeon, Harringtoni, Cooper

personality and behaviour patterns. This ques- who carried out meta-analyses on 17 study tion, raised by Buell and Breslow,55 was also a samples derived from investigations on working concern of Freidman and Rosenman in the hours and health.' They concluded that Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from course of their comprehensive research pro- significant associations could be identified gramme on the type A behaviour pattern.' It between hours of work and both physiological has been suggested that those who often work and psychological health symptoms. Given the long hours might be especially competitive, or range of factors which may moderate the rela- may particularly enjoy their work4" both factors tion between working hours and health the which might in some way influence the risk of authors noted that the results of such analyses associated health consequences. Such issues are, if anything, likely to underestimate the tend to be assessed more often in studies con- strength of the relation. Evidence currently cerned with mental health rather than other available, therefore, is undoubtedly sufficient to types of disorder. However, they may be raise concerns about a possible link between equally relevant in this context where the long hours and the risk of significant health development ofthe physical problem is likely to outcomes, including , be mediated by stress. particularly when those hours exceed 50 a Recent studies of the relation between long week. However, much remains to be explored hours and cardiovascular disease are difficult to in terms of the nature and interactions of the find. Exceptions are those studies carried out in factors involved. Japan where the problem of karoshi, sudden unexplained death in relatively young workers, Performance effects has attracted widespread attention. Uehata' The onset of fatigue is perhaps the most obvi- noted that karoshi is a sociomedical term refer- ous direct result of working long hours. ring to a range of cardiovascular attacks such as However, the concept of fatigue is difficult to strokes, myocardial infarction, or acute cardiac define operationally and hence to measure, failure resulting from hypertensive or arterio- other than by self report. Most studies have sclerotic disease. It is widely thought that the therefore opted to consider fatigue within the long working hours which are so much a part of context of performance that is, to measure it the modern Japanese culture are a major indirectly in terms of its effect on performance. contributory factor. In 1988 it was estimated Human performance in an occupational set- that one in four of all male Japanese employees ting may itself be considered either directly in worked more than 60 hours a week.` Uehata'" terms of outcomes such as productivity and studied the behaviour and lifestyle of over 200 quality assessment, or indirectly in terms of the karoshi victims and concluded that the prob- speed and accuracy in tests which approximate lem resulted from a complex interaction of to the work task. These are a relatively recent work stress and lifestyle factors, with the fatal innovation and the first recorded studies of attack usually triggered by an acute stressor. hours of work and performance focused on Long working hours were judged to play a cen- direct measures of economic efficiency-such

tral part. Interpretation of these data in epide- as productivity costs and absenteeism. http://oem.bmj.com/ miological terms is difficult because of the An early experiment was carried out in a range of conditions encompassed by the single large Manchester engineering works at the end term karoshi. The data are perhaps best viewed of the last century.' After discussions with the therefore as a series of case studies, which sug- workers, the management took the somewhat gest the need for future epidemiological inves- radical step of eliminating early morning tigation. Further, it should be noted that 651Yo (before breakfast) working and also overtime,

of these cases regularly worked more than 60 effectively reducing weekly hours from 53 to on October 2, 2021 by guest. Protected copyright. hours a week, with some working more than 50 48. Production levels remained about the same hours overtime a month and half of their con- and the proportional rise in wages versus tracted holidays. These hours perhaps repre- output was compensated for by a reduction in sent the extreme in terms of working time fuel costs. An important additional benefit was although they are not of course unknown else- a considerable reduction in absenteeism. where among certain groups for example, Around the same time a similar approach was junior hospital doctors in Europe and the adopted by an optical works in Germany.' Here United States. daily hours were reduced from nine to eight, Consideration of the evidence on cardiovas- resulting in a 3% rise in production over a year. cular disease and long hours suggests that the The conclusions of these pioneering experi- relation is not a straightforward one. As with ments were supported by the results of a series the case of mental health disorders working of studies involving munitions workers during hours seem to play an interactive or perhaps an the first world war.' Despite being carried out exacerbating part with other occupational over 80 years ago, these studies have rarely stressors. The influence of sex, of individual been surpassed in terms of scientific method factors (motivation, choice, and disposition), and attention to detail, and remain one of the and of situational factors (work type, home- most important data sources in the field. Dur- work interface) are all important in determin- ing the first world war, munitions workers ing the perception of stress and its associated regularly worked well over 60 hours a week. physical responses. Moreover, these physical Reductions of between seven and 20 hours for responses themselves show wide individual various groups resulted in a working week of variation, which may in turn be linked to between 50 and 55 hours. The authors of the dispositional factors. These complicating fac- report noted that in no case did the reduction tors have also been discussed by Sparks et al in hours result in more than an "insignificant Health and safety problems associated with long working hours 373

diminution of total output" and "on average it gathered in the early part of this century and produced a distinct increase." Production confirmed nearly 50 years ago. increased gradually over the period of the Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from experiment and was further enhanced by the Safety introduction of 10 minute rest breaks in the Discussion of potential effects on safety mornings and afternoons. Interestingly this involves consideration of two separate issues. finding was supported more recently by the The first related to the problem of fatigue and results of a controlled laboratory experiment68 its influence on behaviour associated with safe in which student volunteers performed cogni- working practices-for example, the mainte- tive tests under different conditions. As well as nance of attention and the tendency to take showing a relation between the time spent risks. In this sense, the question of safety is one working on the tests and a deterioration in per- particular aspect of human performance. The formance, the results indicated that total work second issue relates to the problem of pro- output was 6% better with scheduled rest longed exposure to physical, chemical, and breaks than without such breaks. The muni- other hazards in the working environment. tions worker study was one of the first system- The question of fatigue and its effects have atic demonstrations that reducing working been extensively studied relative to shiftwork, hours does not necessarily reduce productivity in which disruption of normal sleep patterns is and in certain circumstances may increase it. clearly of major concern. The problem is a Since the first world war those findings have complex one involving consideration of a vari- been replicated many times in several field ety of physiological and social factors which studies throughout the world.'7 For example, a may influence the performance of the worker, major investigation was conducted in the and several variables which may influence the United States at the end of the second world nature of accident reporting. Most studies tend war to compare the effects of total hours to support the view that safety is more likely to worked in employees who had changed from be compromised during the night shift, par- wartime to peacetime schedules.'7 Although it ticularly where night working is coupled with was evident that effects were dependent on a extended hours.69 70 Little information on the range of factors such as type of work and likelihood of increased frequency of accidents degree of worker control, there was a general occurring-for example, during evening hours trend for longer hours to be associated with after an extended day-can be gleaned from lower productivity and higher absenteeism with these data. Information from studies of 12 hour the optimum schedule seeming to conform to shifts does not tend to support the view that an the standard five day, 40 hour week. extended day results in higher accident rates." In more recent years, there has been a shift in However, the situation is not directly compara- emphasis towards measurement through per- ble with that of the extended day as those formance tests designed to simulate work tasks working 12 hour shifts do so in the expectation or which involve skills regarded as essential to of an extended rest period. Further, studies on

such tasks, in particular the maintenance of accident rates tend to compare data with those http://oem.bmj.com/ attention. Fieldwork in this area, however, has from a previous rotating eight hour system tended to focus on specific occupational rather than a normal eight hour day. Data on groups such as professional drivers and junior accident rates associated specifically with long doctors who combine long hours with regular hours, excluding shiftwork, are scarce. There is and unpredictable work schedules or with shift the information gathered in the munitions work. Similarly, a study of automotive workers workers study, which showed a 2.5-fold reduc- in the United States, which showed cumulative tion in accidents in women, but not in men, negative effects of overtime on negative per- after the reduction in hours of work. However, on October 2, 2021 by guest. Protected copyright. formance, was also concerned with - it should be remembered that these workers ers many of whom regularly worked at night. also worked shifts, including night shifts, and Although strictly outside the scope of this that the hours in question were excessive by review,'8 48 data from these studies and from today's standards. numerous laboratory investigations point to A more recent study in Hong Kong72 was the complex relation between long periods of able to show a relation between severe hand work and impairment of performance. For injuries in factory workers and long working example, the nature of the task, whether it is hours, in this case an average working day of routine and monotonous or complex and nearly 11.5 hours. However, a range of other stimulating,'9 the motivation of the person20 factors such as lack of training and supervision and the presence of other stressors" are all and worker inexperience were also identified. likely to determine the extent of impairment of No analysis was carried out of the time of day performance in any given situation. Further- at which the accidents occurred and the more, the nature of the impairment-for authors concluded that the nature of the example, in terms of speed or accuracy20 or in relation between hours worked and accidents terms of the type of error22-is likely to vary was unclear. between situations and occupations. Although not directly applicable to the ques- To date, the methodology which has been tion of overtime work, data relating to groups developed to investigate these complex rela- working irregular or unsocial hours involving tions in the field of shiftwork does not seem to phase shifting-for example, junior doctors,73 have been applied to the study of overtime airline flight attendants,74 and long distance effects. Strangely the most reliable information drivers75-do provide pointers to the issues in this field remains that which was initially which remain to be explored in this area. These 374 Spurgeon, Harrington, Cooper

include the relevance of time of day, the tial effects ofmore moderate overtime, between importance of rest breaks, and the influence of 40 and 50 hours a week. the type of work on the nature of errors and A further requirement is for the investigation Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from behavioural change. For example, does fatigue of a wider range of potential health outcomes increase risk taking behaviour? Alternatively, as many have been virtually ignored. Numer- do workers opt for a fail safe strategy when they ous problems associated with stress may also be are tired or does this depend on factors such as associated with long hours. These include gas- the perceived importance of the task? Do trointestinal disorders, musculoskeletal disor- workers make more skill based or more knowl- ders, problems associated with depression of edge based errors at certain times of the day? the immune system, and psychosomatic com- These and many other questions remain to be plaints likely to reduce efficiency and increase explored, together with the most fundamental short term absenteeism. Also, the results of one one of whether hours beyond a certain range study which investigated the relation between do actually result in any reduction in safety as several work factors and fetal death76 indicated measured by accident rates. an increased rate of abortion in women The implications of overtime work for expo- working 46 or more hours a week. Studies of sure to workplace hazards has been similarly human performance in response to fatigue in neglected. For example, national and inter- overtime situations are surprisingly limited national exposure standards for toxic chemi- when the attention paid to this in other areas is cals, designed to limit exposure to levels below considered, particularly shiftwork research. those likely to cause harm, are based on an Associated with this there has been little eight hour day, 40 hour week. Only rarely2" has consideration of safety concerns either from the question of necessary adjustment to the legislative point of view or in terms of the accommodate working hours outside this changes in worker behaviour likely to result standard week been discussed. The issues from fatigue. Finally, despite anecdotal evi- involved are complicated and require detailed dence and frequent passing references to what consideration of factors such as the pharma- may be termed social effects-that is effects on cokinetics of the toxic effect of individual home, family, leisure, and general quality of chemicals. In this field as in many others life-these perhaps most obvious conse- adequate information is currently absent. quences oflong hours have not been systemati- cally investigated. Viewed from a health and safety perspective Conclusions and future needs it is clear that excessive overtime working is not This review has considered the weight of the without risk. Given the increasing trend evidence linking long working hours with towards such excess a proper assessment of the health problems in the light of the European nature and level of that risk is now urgently Community Directive on Working Time. What required. emerges most strongly is how little information

is currently available. The effects of long hours 1 Council of the European Communities. Concerning certain http://oem.bmj.com/ or overtime work seem to have been seriously aspects of the organisation of working time. Council Directive 9311041EC. Luxembourg: Official Journal of the European neglected in favour of an understandable Communities, 1993. (No. L 307, 13-12-93, 18-24.) preoccupation with the problems of shiftwork- 2 Waterhouse JM, Folkard S, Minors DS. Shiftywrk, health and safety: an overview of the scientific literature 1978-90. ing. The results of those studies which have London: HMSO, 1992. investigated overtime effects suggest grounds 3 Harrington JM. Shiftwork and health: a critical review ofthe literature on working hours. Ann AcadMed Singapore 1994; for concern but have also highlighted the com- 23:699-705. plex nature of the relation between long hours 4 Mather W. Theforty eight hour week: a years experiment and its and health. It is clear that the results. Manchester: Guardian Printing Works, 1984. on October 2, 2021 by guest. Protected copyright. type and severity 5 Abbe E. 1901. Cited in: J Goldmark. Fatigue, efficiency: a of any identified effects are unlikely to be uni- study in industry. New York: Russell Sage Foundation, versally 1912. applicable, but rather will depend on 6 Vernon HM. Industrial efficiency and fatigue. In: EL Collis, the characteristics and attitudes ofeach person, ed. The industrial dinic, London: John Bale, 1920:51-74. 7 Kogan H. Long hours culture: time for a change. Occup the job requirements, and other aspects of the Health 1996;48:48. organisational and cultural climate. Following 8 Bulletin ofEuropean studies on time: statistics and news. In: Wedderburn A, ed. Trends in general hours of work. the example of shiftwork research, carefully 1996;9: 17-29. controlled studies are required that pay atten- 9 Austin Knight London. The family friendly workplace: report tion to a presented to the conference of the institute of personnel and range of factors which may influence development. London: Austin Knight, 1995. the workers' responses, whether objectively or 10 Higginbottom D. Pay and hours. The essential equation. Report on the United Road Transport Union survey. Chorlton, subjectively assessed. Manchester: United Road Transport Union, 1995. Most of the studies quoted here which have 11 Fukushima T. Working hours. In: Yoichi I, Hiroshi I, Fuku- shown shima T, eds. International comparison of labour statistics. health effects have been concerned with Tokyo: Azuza, 1996. working time beyond 50 hours a week. It is dif- 12 Can long working hours kill? Labour Research 1995;84:21-2. ficult to 13 Kasl SV. Mental health and the work environment. J Occup escape the conclusion that schedules of Med 1973;15:509-18. this nature are detrimental to health and well- 14 Cooper CL, Marshall J. Occupational sources of stress: a being. It is also difficult to find review of the literature relating to coronary heart disease evidence that and mental ill health. Journal of Occupational Psychology long working hours are beneficial, either to 1976;49:11-28. employees or the efficiency of the 15 Leino P. Symptoms of stress predict musculoskeletal disor- organisation ders. YEpidemiol Community Health 1989;43:293-300. as a whole. However, currently available data 16 House JS, McMichael AJ, Wells JA. Occupational stress and are health among factory workers.J Health Soc Behav 1979;20: insufficient to determine exactly how many 139-60. hours people should be required to work ifthey 17 Alluisi EA, Morgan BB. Stress performance effectiveness. are to remain safe and healthy. In particular, In: EA Alluisi, EA Fleishman, eds. Human performance,pro- ductivity. Vol 3. New Jersey: Lawrence Erlbaun Associates, more emphasis should be placed on the poten- 1982. Health and safety problems associated with long working hours 375

18 Brown ID. Driver fatigue. Hum Factors 1994;36:298-314. 48 Spurgeon A, Harrington JM. Work performance and health 19 Dinges DF, Kribb NB. Performance while sleepy. Effects of of junior hospital doctors: a review of the literature. Work experimentally induced sleepiness. In: TH Monk, ed. Sleep, and Stress 1989;3:1 17-28. sleepiness and performance. Chichester: John Wiley, 1991. 49 Bliese PD, Halverson RR. Individual, nomothetic models of Occup Environ Med: first published as 10.1136/oem.54.6.367 on 1 June 1997. Downloaded from 20 Craig A. Human engineering: In: JS Warr, ed. The control of job stress. An examination of work hours, cohesion and vigilance in sustained attention and human performance. well-being. Journal of Applied Social Psychology. 1996;26: Chichester: John Wiley, 1984. 1171-81. 21 Craig A, Cooper RE. Symptoms of acute, chronic fatigue. 50 Watanabe S, Torii J, Shinkai S, Watanabe T. Relationships In: AP Smith, DM Jones, eds. Handbook of human perform- between health status and working conditions and person- ance. Vol 3. State and Trait. London: Academic Press, alities among VDT workers. Environ Res 1992:12. 1993;61:258-65. 22 Williamson AM, Feyer AM. Causes of accidents and time of 51 Haugland G. An investigation into the effects of working day. Work and Stress 1995;9:158-64. overtime on: a) mental health, b) human performance - a pilot 23 Goyal R, Krishnan K, Tardif R, Krishnan K, Tardiff R, study. [MSc dissertation]. Birmingham: University of Lapere S, Brodeur J. Assessment of occupational health Birmingham, 1996. risk during unusual work shifts: review of the needs and 52 Steptoe A, Fieldman C, Evans 0. An experimental study of solutions for modifying environmental and biological limit the effects of control over work pace on cardiovascular values for volatile organic solvents. Can J Public Health responsivity. Journal ofPsychophysiology 1993;7:290-300. 1992;183: 109-12. 53 Carroll C, Harris MG, Ross G. Haemodynamic adjust- 24 Bohle P, Tilley AJ. The impact of night work on psychologi- ments to mental stress in normotensives and subjects with cal well-being. Ergonomics 1989;32:1089-99. mildly elevated blood pressure. Psychophysiology 1991;28: 25 Vidacek S, Radosevic-Vidacek B, Kaliterna U. Predictive 438-46. validity of individual difference measures for sleep 54 Russek HI, Zohman BL. Relative significance of heredity, problems in shiftworkers: preliminary results. In: Oginski diet and occupational stress in coronary heart disease of A, Pokorski J, Rutenfranz J, eds. Contemporary advances in young adults. AmJMed 1958;325:266-75. shiftwork research. Krakow: Krakow Medical Academy, 55 Beull P. Breslow L. Mortality from coronary heart disease in 1987:277-85. Californian men who work long hours. J Chron Dis 26 Folkard S, Barton J. The scientific basis for the CEC's pro- 1960;11:615-26. posed directive on working time. Annals of the Polish Acad- 56 Hinkle LE. Whitney LH, Lehman EW, DunnJ, Benjamin B, emy of Sciences 1992;36:311-3. King R. Occupation, education, and coronary heart 27 Schmitt N, Colligan MJ, Fitzgerald M. Unexplained physi- disease. Science 1968;161:238-48. cal symptoms in eight organizations: individual and organi- 57 Pell S, d'Alonzo CA. Acute myocardial infarction in a large sational analyses. Journal of Occupational Psychology 1980; employed population. Report of six-year study of 1356 53:305-17. cases. JAMA 1963;185:831-41. 28 Oppenheim L. Factors related to occupational stress or 58 Alfredsson L, Spetz CL, Theorell T. Type of occupation and burnout among music therapists. Journal ofMusic Therapy. near-future hospitalisation for myocardial infarction and 1987;24:97-106. some other diagnosis. IntJEpidemiol 1985;14:378-88. 29 Maslach C. Burnout-the cost of caring. New York: Prentice 59 Starrin Larsson Brenner Hall, 1982. B, G, SO, Levi L, Pettersen IL. 30 Duffy CA, McGoldrick AE. Stress and the bus driver in the Structural changes, ill health and mortality in sweden. UK transport industry. Work and Stress 1990;4:17-27. 1963-83. A macroaggregated study. Int J Health Serv 31 Crown S, Crisp AH. Manual of the Crown-Crisp experiential 1990;20:27-42. index. London: Hodder and Stoughton, 1979. 60 Sorenson G, Pirie P, Folsom A, Luepker R. Jacobs D, 32 Daniels K, Guppy A. Stress, social support, and psychologi- Gillum R. Sex differences in the relationship between work cal well-being in British accountants. Work and Stress 1995; and health. The Minnesota heart survey. J Health Soc 9:432-47. Behav 1985;26:379-94. 33 Warr PB. The measurement of well-being and other aspects 61 Sutherland V, Davidson MJ. Using a stress audit: the of mental health. Journal of Occupational Psychology construction site manager experience in the UK. Work and 1990;63: 193-210. Stress 1993;7:273-86. 34 Goldberg D, Williams P. A users guide to the general health 62 Thiel HG, Parker D, Bruce TA. Stress factors and the risk of questionnaire, (GHQ). Windsor: Nelson, 1988. myocardial infarction. J Psychosom Res 1973;17:43-57. 35 Ezoe S, Morimoto K. Behavioural lifestyle and mental 63 Friedman M, Rosenman RH. Association with specific overt health status of Japanese factory workers. Prev Med behaviour pattern with blood and cardiovascular findings. 1994;23:98-105. JAAMA 1959;169:1286-97. 36 Galambos NL, Walters BJ. Work hours, schedule inflexibil- 64 Uehata T. Karoshi due to occupational stress-related ity and stress in dual-earner spouses. Can J Behav Sci cardiovascular injuries among middle-aged workers in 1992;24:290-302. Japan. Journal of Science of Labour 1991;67:20-8. 37 Hall E, Johnson Stewart W. Double exposure: the com- JV, 65 Ohta T. Work rules in Japan. International Labour Review http://oem.bmj.com/ bined impact of the home and work environments on psy- 1988;127:627-39. chosomatic strain in Swedish women and men. IntJ Health 66 Uehata T. Long working hours and occupational stress- Serv 1992;22:239-45. related cardiovascular attacks among middle-aged workers 38 Raggatt PTF. Work stress among long distance coach inJapan. J Hum Ergol (Tokyo) 1992;20:147-53. drivers: a survey and correlational study. Journal of Organi- 67 Sparks K, Cooper CL, Fried Y, Shirom A. The effects of sational Behaviour 1991;12:565-79. hours of work on health: a meta-analytic review. Journal of 39 Stressing performance IV. A report on peformance, pay, work- Occupational and Organisational Psychology 1997 (in press). loads and stress among managerial and professional employees 68 Okogbaa OG, Shell RL. The measurement of knowledge in BTpk. Report of Society of Telecom Executives. London: worker fatigue. STE Research c/o IPMS, 1995. HIE Transactions 1986;335-42. 40 Westman M, Eden D, Shirom A. Job stress, cigarette smok- 69 Gold DR, Rogocz S, Bock N, Tosteson TD Baum M, ing and cessation: the conditioning effects of peer support. Speizer FE. Rotating shift-work, sleep and accidents on October 2, 2021 by guest. Protected copyright. Soc Sci Med 1985;20:637-44. related to sleepiness in hospital nurses. Am J Public Health 41 Caplan RD, Cobb S, French RP. Relationships of cessation 1992;7:1011-4. of smoking with job stress, personality and personal 70 Smith L, Folkard S, Poole CJM. Increased injuries on the support. JAppl Psychol 1975;60:211-9. night shift. Lancet 1994;344: 1137-9. 42 Kawakami N, Araki S, Haratani T, Hemmi T. Relations of 71 Laundry BR, Lees REM. Industrial accident experience of work stress to alcohol use and drinking problems in male one company of eight and 12 hour shift systems. J Occup and female employees of a computer factory in Japan. Med 1991;33:903-6. Environ Res 1993;62:314-24. 72 Ong SG, Fung SC, Chow SP, Kleevens JWL. A study of 43 Ford CV, Wentz DK. Internship: what is stressful? South major factors associated with several occupational hand MedJ 1986;79:595-9. injury in Hong Kong island. Journal of Society of 44 Brown JM, Campbell EA. Sources of occupational stress in Occupational Medicine. 1982;32:82-8. the police. Work and Stress 1990;4:305-18. 73 McCall TB. The impact of long working hours on resident 45 Cooper CL, Davidson MA, Robinson P. Stress in the police physicians. NEnglJMed 1988;318:775-8. service. Jf Occup Med 1988;24:30-6. 74 Ono Y, Watanabe S, Kaneko S, Matsumoto K, Miyao M. 46 Turnage JJ, Spielberger CD. Job stress in managers, profes- Working hours and fatigue of Japanese flight attendants. J sional and clerical workers. Work and Stress 1991;5:165-76. Hum Ergol (Tokyo) 1991;20:155-64. 47 Maruyama S, Kohno K, Morimotot K. A study of 75 Feyer A-M, Williamson AM. Work and rest in the long dis- preventive medicine in relation to mental health among tance road transport industry in Australia. Work and Stress middle management employees (part 2). Effects of long 1995;9: 198-205. working hours on lifestyles, perceived stress and working 76 McDonald AD, McDonald JC, Armstrong B, Cherry NM, life satisfaction among white collar middle management Cote R, Lavoie J, et al. Fetal death and work in pregnancy. employees. Japanese Journal ofHygiene 1995;50:849-60. BrJInd Med 1988;45:148-57.