S Taimela, 1 A Malmivaara, 2 S Justén, 1 ELäärä, 3 H Sintonen, 4 J Tiekso, 1 T Aro 5. Original article

Transcription

1 Original artile See editorial, p Evalua International, Vantaa, Finland; 2 Finnish Offie for Health Tehnology Assessment, FinOHTA/Stakes, Helsinki, Finland; 3 University of Oulu, Department of Mathematial Sienes, Oulu, Finland; 4 University of Helsinki, Department of Publi Health, and FinOHTA, Helsinki, Finland; 5 Mutual Pension Insurane Company Ilmarinen, Helsinki, Finland Correspondene to: Dr S Taimela, Evalua International, PO Box 35, FIN Vantaa, Finland; simo.taimela@evalua.fi Aepted 16 July 2007 Published Online First 6 August 2007 This paper is freely available online under the BMJ Journals unloked sheme, see oem.bmj.om/info/unloked.dtl The effetiveness of two oupational health intervention programmes in reduing sikness absene among employees at risk. Two randomised ontrolled trials S Taimela, 1 A Malmivaara, 2 S Justén, 1 ELäärä, 3 H Sintonen, 4 J Tiekso, 1 T Aro 5 ABSTRACT Objetives: To evaluate the effetiveness of two oupational health intervention programmes, both ompared with usual are. Methods: Based on a health survey, 1341 employees (88% males) in onstrution, servie and maintenane work were lassified into three groups: low risk (n = 386), intermediate risk (n = 537) and high risk (n = 418) of sikness absene. Two separate randomised trials were performed in the groups high risk and intermediate risk, respetively. Those high risk subjets that were alloated to the intervention group (n = 209) were invited to oupational health servie for a onsultation. The intervention inluded, if appropriate, a referral to speialist treatment. Among the intermediate risk employees those in the intervention group (n = 268) were invited to all a phone advie entre. In both trials the ontrol group reeived usual oupational health are. The primary outome was sikness absene during a 12- month follow-up (register data). Results: The high risk group, representing 31% of the ohort, aounted for 62% of sikness absene days. In the trial for the high risk group the mean sikness absene was 30 days in the usual are group and 19 days in the intervention group; the mean differene was 11 days (95% CI 1 to 20 days). In the trial for the intermediate risk group the mean sikness absene was 7 days in both arms (95% CI of the mean differene 2.3 to 2.4 days). Conlusions: The identifiation of high risk of work disability was suessful. The oupational health intervention was effetive in ontrolling work loss to a degree that is likely to be eonomially advantageous within the high risk group. The phone advie intervention for the intermediate risk group was not effetive in ontrolling work loss primarily due to poor adherene. Sikness absene may result in onsiderable personal and publi finanial onsequenes. Long-term sikness absene also predits early retirement. 1 3 Biographial and soioeonomi fators, diagnosed diseases, poor self-rated health, hroni omplaints and poor work apaity predit sikness absene. 4 7 Some randomised ontrolled trials (RCTs) have been performed in oupational settings in order to intervene speifi diseases (musuloskeletal disorders or depression), or to advoate exerise Only a few studies have aimed at identifying employees at high risk of work disability 7 15 or at reduing sikness absene within a high risk subgroup. Telephone health ounselling has been marketed as a low-ost intervention, but its effiay in the oupational healthare setting has not been tested in a randomised trial. In this study we evaluated the effetiveness of two interventions for employees at high or intermediate risk of sikness absene, respetively. Subjets in the intervention for high risk were invited to a onsultation at the oupational health servies. Subjets in the intervention for intermediate risk were invited to all a telephone health advie entre. METHODS Study design and ethis The design was a longitudinal ohort study with two embedded randomised trials. The risk of work disability was lassified with self-administered questionnaires. 18 Table 1 shows the riteria for the risk lassifiation. Two separate randomised trials were performed in the subgroups of high risk (HR) and intermediate risk (IR) of sikness absene, respetively. The primary outome was sikness absene during the 12-month follow-up. The Helsinki University Researh Ethis Board approved the study, and it was performed aording to the Delaration of Helsinki. Partiipants The study was performed within one orporation in Finland. Inlusion riteria were permanent employment and age years. Questionnaires were sent to a ohort of 3115 employees in September The proposed study design, impliations of the trial, and alternative options were explained in the over letter. The letter also emphasised that taking part in the trial was voluntary, and employees would get the best treatment available and full attention of the oupational physiian even if they did not want to partiipate. In addition, it was explained that partiipants were free to withdraw from the trial at any point, and it would not prejudie their treatment. Of the target group, 49% were employed in the onstrution industry: ivil engineering, building ontrating, tehnial building servies and building materials industry. 51% were employed in the repair, servie and maintenane of buildings, industrial installations or ommuniations networks. Oup Environ Med: first published as /oem on 6 August Downloaded from on 31 August 2018 by guest. Proteted by opyright. 236 Oup Environ Med 2008;65: doi: /oem

2 Table 1 The riteria for lassifying the employees into high risk and intermediate risk groups Topi Criteria High risk for work disability At least one of the riteria fulfilled Impairment due to musuloskeletal >5 (sale 0 10) problems at work 19 Potential depression 20 DEPS sore >11 (sale 0 30) Distress 21 Very muh feeling tense, strained, nervous and/or anxious beause things are on one s mind all the time Fatigue 19 Very muh feeling of being squeezed empty beause of work Sleep disturbanes 22 Problems in falling asleep or night awakenings AND daytime tiredness daily or almost daily Future working ability 23 Unertain of own ability or quite sure of not being able to ontinue in the present job due to health problems Intermediate risk for work At least one of the riteria fulfilled, but none of disability the riteria for high risk fulfilled Impairment due to musuloskeletal 4 (sale 0 10) problems at work 19 Impairment due to musuloskeletal >5 (sale 0 10) problems at leisure time ativities 19 Pain (frequeny and intensity) At least moderate pain that affets working ability at minimum three times a week Weight problems BMI (body mass index) >30 or BMI (18.5 Exess alohol onsumption 24 Males >350 ml/week; Females >240 ml/ week* Mood disturbanes 20 DEPS sore >8, (sale 0 30) Sleep disturbanes 22 Problems in falling asleep or night awakenings AND daytime tiredness three times a week or more Daytime sleepiness 25 Epworth sleepiness sale (ESS) sore >8 (sale 0 24) Suspiion of sleep apnoea 22 Snoring and shortness of breath while asleep daily or almost daily Insuffiient sleep Differene between reported need and the realisation of sleep >2 h *Expressed as absolute alohol. Randomisation After olleting all responses and proessing the risk group lassifiation, a researh assistant randomised eah subjet in the HR and IR groups into one of the two subgroups, intervention and ontrol ( high risk : HR-IG and HR-CG; intermediate risk : IR-IG and IR-CG). First, to ensure a balaned distribution of subjets by age, sripted four-digit identifiation odes (ID) were sorted by age within both RCTs and then all other items but the ID odes were removed from the list of subjets. An IT expert did this first step. After that the researh assistant performed the randomisation in bloks of 10. A biostatistiian had prepared the order from a random number table. The researh assistant and researhers were not aware of whih of the odes belonged to the intervention group and whih to the ontrol group in either trial. Neither were they able to identify the individuals based on the IDs, and ould not therefore predit the group assignments. The oding was opened only after the primary analysis of the follow-up data was ompleted. Intervention versus are as usual RCT 1: High risk group The employees own oupational nurses and physiians arried out the intervention for subjets at HR. Forty eight oupational health entres were involved in the study. The employees in the HR-IG reeived a letter with personal feedbak of their questionnaire results and invitation to a Original artile onsultation at the oupational health servies (OHS). At most, two reminders were sent. The main purpose of the onsultation was the onstrution of an ation plan, and if appropriate, referral to a further onsultation by a speialist, or psyhologist. The oupational nurse first started the onsultation, the planned duration of whih was 90 minutes, and an oupational physiian joined the meeting later if needed. The individual findings of the questionnaire were available for the OHS professionals during the onsultation. Key treatment proesses were defined in advane and the poliies and praties at the oupational health entres were not altered as a result of the study. To find out what ations were taken within the intervention, an oupational nurse wrote a personal file for eah employee in the HR-IG at the end of the follow-up. The personal files inluded information about the employee attending to the onsultation, the referrals to further evaluation or interventions, the health advie reeived at the OHS, the onsiderations of OHS professionals that no further ations were needed, and the refusals of some employees to take further ation. Additionally, the nurses reported if the employee had already reeived treatment at the OHS for the health issues that were the reason for the invitation of onsultation. The employees in the HR-CG reeived are as usual. They ould onsult their oupational nurse or physiian on request, but they were not invited for a onsultation and did not reeive feedbak of their results. RCT 2: Intermediate risk group The intervention for workers at IR onsisted of an aess to medial ounselling over the telephone from one phone advie entre. The employees in the IR-IG reeived a letter with personal feedbak of their results and invitation to all the phone advie entre in order to reeive respetive medial advie. Two reminders were sent. The swithboard was always open, and the ost for the telephone all was the same as for a loal all. All telephones were manned by trained nurses with several years of experiene and speifi training for their job. During the ounselling the individual findings of the questionnaire were available for the nurses who also had aess to relevant health databases while providing the health advie. The employees in the IR-CG reeived are as usual. Measurements Questionnaire data The baseline questionnaire inluded items on the following: lifestyle, anthropometris, sleep disturbanes, work-related stress and fatigue, depression, pain, disability due to musuloskeletal problems, and a predition of future working ability. The responses were interpreted on the basis of a priori defined ut-off limits (table 1). Sikness absene from work Sikness absene data were obtained, without medial diagnoses, from the employer s reords. The baseline overed the period from 1 Otober 2003 to 30 September 2004 and the follow-up overed the period from 1 Otober 2004 to 30 September Data privay was stritly followed. Reords were heked for inonsistenies. Maternity/paternity leave and absene from work to are for a sik hild are not inluded in the sikness absenes. Oup Environ Med: first published as /oem on 6 August Downloaded from on 31 August 2018 by guest. Proteted by opyright. Oup Environ Med 2008;65: doi: /oem

3 Original artile Sample size alulations RCT 1: High risk group The target sample size of 420 employees was based on the assumptions that 360 of them an be followed-up for one year, and that there will be a 15% differene between the groups in sikness absene with the mean baseline sikness absene estimated to be 20 (SD 9) days/year. Assuming a normal distribution for the outome variable this gave an alpha of 0.05 with 80% power. RCT 2: Intermediate risk group The original target sample size of 840 employees was based on the assumptions that 686 of them an be followed-up for one year, and that there will be an 11% differene between the groups in sikness absene with the mean baseline sikness absene estimated to be 11 (SD 5) days/year. This gave an alpha of 0.05 with 80% power. However, at the time of randomisation, there were only 537 subjets eligible for the IR group. We reviewed the power alulation: our sample size was suffiient to detet a 14% differene. Statistial methods We arried out an intention-to-treat analysis. The effetiveness of the interventions was estimated by the differene of mean number of sikness absene days between the randomised groups, and the onfidene interval was omputed based on t distribution. 26 We used Statistia data analysis software, version 6 (StatSoft, In, Tulsa, OK, USA; 2001). RESULTS At baseline, we reeived 1507 responses (48.4%) of whih 166 were exluded. Thus, the final study population onsisted of 1341 subjets (fig 1). The respondents were on average 44 years Figure 1 Study flow. ET, employment terminated during follow-up, SA, sikness absene. old (range years). Of them 12% were females and 62% were blue-ollar workers. Risk lassifiation 418 (31%) subjets belonged to the high risk group, 537 (40%) to the intermediate risk group, and 386 (29%) to the low risk group. In omparison to other partiipants, the subjets in the high risk group were on average older and a larger proportion of them was male and worked in physially demanding jobs (table 2). Effetiveness of the interventions RCT 1 The oupational health intervention for the HR-IG was effetive in ontrolling sikness absene. In the HR-CG, the mean, median and total sikness absene days inreased, and the proportion of subjets with zero absene dereased. No hange took plae in the mean, median and total sikness absene days, or in the proportion of subjets with zero absene in the HR-IG (table 3). The group differene between the means was 11 days (95% CI 1 to 20). RCT 2 The oupational health intervention by phone advie for the IR-IG was not effetive in reduing sikness absene. The mean, median and total sikness absene days, or the proportion of subjets with zero absene did not differ between the IR-CG and IR-IG (table 3). Adherene Of the subjets in the HR-IG in the RCT 1 (n = 209), 142 (68%) attended the onsultation at OHS (fig 2). Fifty did not attend for unknown reasons. The employment had terminated with 17 subjets during the follow-up. Of the attendees, five refused Oup Environ Med: first published as /oem on 6 August Downloaded from on 31 August 2018 by guest. Proteted by opyright. 238 Oup Environ Med 2008;65: doi: /oem

4 Original artile Table 2 Baseline harateristis, and one-year follow-up of sikness absene in risk groups of work disability Risk group lassifiation Low risk Intermediate risk High risk All subjets Baseline n Mean age (years) (range) 43 (19 60) 43 (19 61) 47 (23 60) 44 (19 61) Sex (female; %) Blue-ollar (%) Sikness absene None (% within group) Mean (days) Standard deviation Median (days) Upper quartile (days) Maximum (days) Sum (days) Employment terminated during follow-up (%) year follow-up n Sikness absene None (% within group) Mean (days) Standard deviation Median (days) Upper quartile (days) Maximum (days) Sum (days) The risk lassifiation predited sikness absene: 62% of sikness absene days during the 12-month follow-up took plae within the high risk group (table 2). Table 3 Baseline harateristis and sikness absene in the intervention and ontrol (usual are) groups in RCT 1 (oupational health intervention for the high risk group) and RCT 2 (telephone advie for the intermediate risk group) Group alloation Intermediate risk (IR) High risk (HR) Control (IR-CG) Intervention (IR-IG) Control (HR-CG) Intervention (HR-IG) Baseline n Mean age (years) Sex (female; %) Blue-ollar (%) Sikness absene None (% within group) Mean (days) Standard deviation Median (days) Upper quartile (days) Maximum (days) Sum (days) Employment terminated during follow-up (%) 1 year follow-up n Sikness absene None (% within group) Mean (days) Standard deviation Median (days) Upper quartile (days) Maximum (days) Sum (days) further examinations or interventions, and the OHS professionals had onsidered that eight subjets did not warrant further ations. 129 subjets ended up in the interventions: health advie (n = 106), referral to onsultation or hospital outpatient lini (n = 64), or a group intervention at the OHS (n = 6), in different ombinations. Of the 142 subjets who visited OHS, 72 (51%) had not reeived earlier treatment at OHS for the respetive reasons for belonging to the high risk group. Of the subjets in the IR-IG in the RCT 2 (n = 268), 57 (21%) had alled the phone advie entre during the follow-up. Adverse events No adverse events were reported during the interventions. DISCUSSION Main findings It was possible to identify a group of employees with high risk of work disability and subsequent sikness absene. Moreover, the oupational health intervention was effetive in ontrolling sikness absene within this group. The differene ompared with the usual are treatment arm was 11 days per year, whih is obviously of eonomi importane. The majority of subjets in the intermediate risk intervention did not use the telephone health advie and its effetiveness remains unertain. Strengths and weaknesses of the study The main strength of this study lies in the pragmati approah in the randomised ontrolled trial for employees at high risk. All permanent employees in the target ohort were offered the opportunity to partiipate. Although the response rate was somewhat low, it was in line with other studies in oupational Oup Environ Med: first published as /oem on 6 August Downloaded from on 31 August 2018 by guest. Proteted by opyright. Oup Environ Med 2008;65: doi: /oem

5 Original artile Figure 2 Adherene to the oupational health intervention: HR-IG (RCT 1). *Aording to the OHS professionals. {Many subjets reeived more than one intervention. populations in European ountries. 27 The employees own oupational nurses and physiians arried out the oupational health intervention. Adherene within the HR-IG was reasonably high, and the intervention sueeded in apturing many workers with underlying health problems that had not been properly attended to. Of the subjets who visited OHS, more than half had not reeived treatment at OHS before. New ways of treatment were not introdued and the intervention relied on existing praties. As there was no initial randomisation to getting a sreening questionnaire or not, our study annot genuinely answer the overall question of whether the sreening programme as a whole was effetive. Our researh question was formulated to study whether the interventions for the high risk and intermediate risk groups were effetive. However, it is possible to estimate whether the savings due to the redution of sikness absene in the high risk group exeed the investment in the whole sreening proess, but we leave the alulation to be done by the reader with the reader s own ost parameters. Control arm ontamination that is, subjets randomised to usual are reeiving the same treatment as the intervention subjets is a ommon soure of bias in randomised trials. In the present trial randomisation within the high risk and intermediate risk groups was performed individually, instead of lusters. Theoretially, ontrol arm ontamination within workplaes is a potential weakness in both trials in suh setting. However, the subjets randomised to the usual are at OHS did not reeive any feedbak of their own survey results that is, in whih group they belonged. Moreover, most of the subjets were working in small workplaes or workgroups and therefore Main messages It was possible to identify individuals at a high risk of sikness absene with a simple health questionnaire among employees predominantly engaged in physial work. An oupational health intervention, whih inluded an invitation to oupational health servie for a onsultation and, if appropriate, a referral to speialist treatment, was effetive in ontrolling sikness absene within the high risk group. The mean differene of 11 days between the intervention and usual are treatment arms in this randomised trial is likely to be eonomially advantageous. Poliy impliations Identifiation of individuals at high risk of sikness absene an be done with self-administered health questionnaires. Oupational health intervention an ontrol sikness absene within the high risk group. interation between treatment arms would have been limited anyway: there were 48 oupational health entres involved in the study and far more workplaes. Furthermore, the OHS personnel were not aware of the survey results of the usual are groups and thus ould not offer them the same servies as for the intervention group. For these reasons, we believe that it is unlikely that the results would be biased by ontamination. Our primary outome was based on reorded sikness absenes. This has several advantages: good overage, auray and onsisteny. 28 We were able to follow 92% of the subjets in both arms of the high risk trial. Despite the heavily skewed distribution of sikness absene, in suffiiently large samples linear regression models, inluding t test and ANCOVA, are valid for any distribution. 26 Some differenes ompared with previous studies The majority of previous randomised ontrolled trials in oupational health settings have been illness-related, or the fous has been on highly seleted groups of employees. Some RCTs for musuloskeletal disorders 8 10 and depression 11 have been reported. Few studies have dealt with developing a sreening instrument for employees at high risk of work disability and sikness absene Two RCTs have aimed at reduing sikness absene within a high-risk subgroup. Fleten and Johnsen reported a trial in Norway in with 990 onseutive newly sik-listed employees with musuloskeletal or mental disorders randomised to intervention and ontrol group. 16 The minimal postal intervention a general information letter redued the length of sik leave periods in subgroups with mental disorders, rheumati disorders and arthritis, but did not show effiay in the intention-to-treat analysis. A Duth trial 17 within one large ompany randomised 116 employees to an intervention (n = 61) and ontrol (n = 55) group. Subjets were older than 50 years and had reported that they would not be able to ontinue in the present job until retirement. The programme was exeuted by an oupational physiian and omprised at least three onsultations inluding an assessment interview. The proedure inluded the onstrution of a detailed ation plan, onsultation of the employee s Oup Environ Med: first published as /oem on 6 August Downloaded from on 31 August 2018 by guest. Proteted by opyright. 240 Oup Environ Med 2008;65: doi: /oem

6 supervisors and personnel managers and, if appropriate, referral to the general pratitioner, a medial speialist or psyhologist. The authors reported fewer retirements in the intervention group (11%) than in the ontrol group (28%). The total average number of sik leave days in two years was 82 for the intervention group and 108 for the ontrol group. CONCLUSIONS Employees at a high risk for sikness absene an be identified by a health survey and oupational health are an support the working ability of these individuals. The intervention showed a lear advantage in sikness absenes in omparison with usual are, but the ost onsequenes of intervention and usual are need to be onsidered in order to evaluate the ost effetiveness of oupational health intervention in the high risk group. Future researh should also address the question of whether the same intervention approah is effetive in different oupational settings and professional groups. Funding: Finnish Funding Ageny for Tehnology and Innovation (TEKES); the Finnish National Fund for Researh and Development (SITRA); Pfizer Oy. The authors work was independent of the funders. Competing interests: ST and JT are shareholders of and SJ is employed by Evalua International, whih was responsible for the sreening instrument used in the study. EL, AM, HS and TA have no ompeting interests to delare. Ethis approval: The Helsinki University Researh Ethis Board for Oupational Health reviewed the study plan and gave their approval in advane. Reord number (DNRO): 28/e2/04; date: All subjets reeived written information regarding the study aording to the priniples of the Delaration of Helsinki. Only subjets who gave their signed informed onsent were inluded in the study. The onsent letters are stored with other study material. REFERENCES 1. Salonen P, Arola H, Nygard CH, et al. Fators assoiated with premature departure from working life among ageing food industry employees. Oup Med (Lond) 2003;53: Szubert Z, Sobala W. Current determinants of early retirement among blue ollar workers in Poland. Int J Oup Med Environ Health 2005;18: Virtanen M, Kivimaki M, Vahtera J, et al. Sikness absene as a risk fator for job termination, unemployment, and disability pension among temporary and permanent employees. Oup Environ Med 2006;63: Reiso H, Nygard JF, Brage S, et al. Work ability and duration of ertified sikness absene. Sand J Publi Health 2001;29: Virtanen M, Kivimaki M, Elovainio M, et al. From inseure to seure employment: hanges in work, health, health related behaviours, and sikness absene. Oup Environ Med 2003;60: Vaananen A, Toppinen-Tanner S, Kalimo R, et al. Job harateristis, physial and psyhologial symptoms, and soial support as anteedents of sikness absene among men and women in the private industrial setor. So Si Med 2003;57: Original artile 7. Roelen CA, van der Pol TR, Koopmans PC, et al. Identifying workers at risk of sikness absene by questionnaire. Oup Med (Lond) Heymans MW, de Vet HC, Bongers PM, et al. The effetiveness of high-intensity versus low-intensity bak shools in an oupational setting: a pragmati randomized ontrolled trial. Spine 2006;31: Hlobil H, Staal JB, Twisk J, et al. The effets of a graded ativity intervention for low bak pain in oupational health on sik leave, funtional status and pain: 12-month results of a randomized ontrolled trial. J Oup Rehabil 2005;15: Arnetz BB, Sjogren B, Rydehn B, et al. Early workplae intervention for employees with musuloskeletal-related absenteeism: a prospetive ontrolled intervention study. J Oup Environ Med 2003;45: Lo Sasso AT, Rost K, Bek A. Modeling the impat of enhaned depression treatment on workplae funtioning and osts: a ost-benefit approah. Med Care 2006;44: Proper KI, de Bruyne MC, Hildebrandt VH, et al. Costs, benefits and effetiveness of worksite physial ativity ounseling from the employer s perspetive. Sand J Work Environ Health 2004;30: Eriksen HR, Ihlebaek C, Mikkelsen A, et al. Improving subjetive health at the worksite: a randomized ontrolled trial of stress management training, physial exerise and an integrated health programme. Oup Med (Lond) 2002;52: Brox JI, Froystein O. Health-related quality of life and sikness absene in ommunity nursing home employees: randomized ontrolled trial of physial exerise. Oup Med (Lond) 2005;55: Duijts SF, Kant IJ, Landeweerd JA, et al. Predition of sikness absene: development of a sreening instrument. Oup Environ Med 2006;63: Fleten N, Johnsen R. Reduing sik leave by minimal postal intervention; a randomized, ontrolled intervention study. Oup Environ Med 2006;63: de Boer AG, van Beek JC, Durink J, et al. An oupational health intervention programme for workers at risk for early retirement; a randomised ontrolled trial. Oup Environ Med 2004;61: Taimela S, Läärä E, Malmivaara A, et al. Self-reported health problems and sikness absene in different age groups predominantly engaged in physial work. Oup Environ Med 2007;64: National Publi Health Institute. Implementation and methods of the Health 2000 Survey. (Menetelmäraportti. Terveys 2000-tutkimuksen toteutus, aineisto ja menetelmät). Helsinki: National Publi Health Institute, Salokangas RK, Poutanen O, Stengard E. Sreening for depression in primary are. Development and validation of the Depression Sale, a sreening instrument for depression. Ata Psyhiatr Sand 1995;92: Kauppinen T, Hanhela R, Heikkilä P, et al. Work and Health in Finland in 2003 (Työ ja terveys Suomessa 2003). Helsinki: Finnish Institute of Oupational Health, Partinen M, Gislason T. Basi Nordi Sleep Questionnaire (BNSQ): a quantitated measure of subjetive sleep omplaints. J Sleep Res 1995;4: Tuomi K, Ilmarinen J, Jahkola A, et al. Work Ability Index (Työkykyindeksi). Helsinki: Finnish Institute of Oupational Health, Simpura J. Development of ommon instrument for alohol onsumption. In: Nasikov A, Guder C, eds. Eurohis. IOS Press, WHO Regional Offie for Europe, Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness sale. Sleep 1991;14: Lumley T, Diehr P, Emerson S, et al. The importane of the normality assumption in large publi health data sets. Annu Rev Publi Health 2002;23: Gimeno D, Benavides FG, Amik BC 3rd, et al. Psyhosoial fators and work related sikness absene among permanent and non-permanent employees. J Epidemiol Community Health 2004;58: Ferrie JE, Kivimaki M, Head J, et al. A omparison of self-reported sikness absene with absenes reorded in employers registers: evidene from the Whitehall II study. Oup Environ Med 2005;62:74 9. Oup Environ Med: first published as /oem on 6 August Downloaded from on 31 August 2018 by guest. Proteted by opyright. Oup Environ Med 2008;65: doi: /oem