Scandinavian http://sjp.sagepub.com/ Journal of Public Health
Do work-related mechanical and psychosocial factors contribute to the social gradient in long-term sick leave: A prospective study of the general working population in Norway Tom Sterud and Håkon A Johannessen Scand J Public Health 2014 42: 329 originally published online 30 January 2014 DOI: 10.1177/1403494814521506 The online version of this article can be found at: http://sjp.sagepub.com/content/42/3/329
Published by: http://www.sagepublications.com
Additional services and information for Scandinavian Journal of Public Health can be found at: Email Alerts: http://sjp.sagepub.com/cgi/alerts Subscriptions: http://sjp.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav
>> Version of Record - Apr 23, 2014 OnlineFirst Version of Record - Jan 30, 2014 What is This?
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
521506 research-article2014
SJP0010.1177/1403494814521506T. Sterud and H.A. JohannessenWork-related educational differences in sick leave
Scandinavian Journal of Public Health, 2014; 42: 329–334
Original Article
Do work-related mechanical and psychosocial factors contribute to the social gradient in long-term sick leave: A prospective study of the general working population in Norway
Tom Sterud & Håkon A Johannessen Department of Occupational Health Surveillance, National Institute of Occupational Health, Oslo, Norway
Abstract Aims: A social gradient in long-term sick leave (LTSL) rates is well established, but only a few studies have examined to what extent this gradient may be explained by mechanical and psychosocial work environment factors. Methods: A randomly drawn cohort from the general population in Norway, aged 18–69 years, was interviewed in the second half of 2009 (n=12,255, response at baseline 60.9%) and followed up in national registries to the end of 2010. Eligible respondents were registered with an active employee relationship of at least 100 actual working days in 2009 and 2010 (n=6758). Based on administrative register data, respondents were coded into five educational levels (university/college ≥4 years was set as the reference group). Eight work-related psychosocial factors and 10 mechanical exposures were measured. The outcome of interest was medically confirmed LTSL ≥40 working days during 2010. Results: In total, 9.4% (635 individuals) were classified with LTSL during 2010. There was a strong social gradient ranging from 12.4% (elementary) to 3.3% (university/ college ≥4 years) among men. The corresponding figures among women were 15.4 and 4.6%. Adjusting for work-related mechanical and psychosocial factors explained between 41 and 44% of the social gradient in men. Among women, the corresponding figures were 31 and 54%. Conclusions: Work-related mechanical and psychosocial factors contribute to the social gradient in LTSL. The work-related factors that accounted for this gradient were rather similar for men and women. Key Words: Educational status, mechanical processes, occupational, prospective study, psychosocial factors, risk factors, sick leave, socioeconomic status, work, work place
Introduction A social gradient in long-term sick leave (LTSL) rates is well established [1], but only a few studies have examined to what extent this gradient can be explained by work environment factors [2–4]. Inability to work because of sickness and disability generate considerable costs to the individual as well as to society and more knowledge is needed about the impact of specific risk factors at the work place. We therefore studied differences between educational level groups in the risk of medically confirmed LTSL and the degree to which these differences were
explained when taking into account a wide range of mechanical and psychosocial risk factors in a nationwide general working population sample. Participants and methods Data were provided from the nationwide Survey of Level of Living– Working Conditions, conducted by Statistics Norway (SSB). Eligible respondents were community-living Norwegian residents aged 18–69 years. In 2009, a gross sample of 20,136 was randomly
Correspondence: Tom Sterud, Department of Occupational Health Surveillance, National Institute of Occupational Health, PO Box 8149 Dep., N-0033 Oslo. Norway. E-mail: [email protected] (Accepted 7 January 2014) © 2014 the Nordic Societies of Public Health DOI: 10.1177/1403494814521506
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
330 T. Sterud and H.A. Johannessen drawn from this population. Of these, 12,255 (60.9%) persons were interviewed between 22 June 2009 and 9 January 2010. Data on sick leave was obtained from the Norwegian Labour and Welfare Administration’s sickness benefit register. The register includes all workers aged 16–69 living in Norway and registered with an active employee relationship. Respondents who were in paid work for at least 1 hour during the reference week, or were temporarily absent from such work, and were registered with an active employee relationship of at least 100 actual working days in each year (2009 and 2010) constituted the follow-up sample in the present paper (n=6758). Measurements Long-term sick leave (LTSL) was defined as medically confirmed sick leave for a period equalling 40 or more actual working days during 2010, the year after the initial survey data was collected. Educational level was based on administrative register data and was coded into five educational levels: Group 5: elementary; Group 4: incomplete upper secondary; Group 3: upper secondary; Group 2: university/college 4 years; Group 1: university/college ≥4 years. Gender and age were based on self-reported information. Region was based on administrative register data and was coded into seven regions based on the standard for economic regions developed by Statistics Norway. Perceived mechanical workload was measured with 10 items: neck flexion; hand/arm repetition; work with hands lifted above shoulder height; squatting/kneeling; standing; work with upper body forward bend; awkward lifting; heavy physical work; whole body vibration; and heavy lifting. Scores were coded on a scale from 1 (not exposed or exposed very little of the work day) to 4 (exposed three-quarters of the work day or more). The items have been described in greater detail elsewhere [5]. Perceived psychosocial factors at work included job demands (two items), low job control (four items), role conflict (three items), low leadership support (three items), monotonous work (one item), emotional demands (two items), and bullying/harassment (three items), corresponding to those used in previous studies [6,7]. All factors were used as continuous variables (range 1–4/5), and high score indicates unfavourable exposure, except for a dichotomous variable for bullying/harassment. Self-reported health and lifestyle factors Musculoskeletal complaints (neck/shoulder, low back pain, arm and leg pain) and psychological distress
(dejection or depression and nervousness, anxiety or restlessness) was used as continuous variables ranging from 1 (not afflicted) to 4 (severely afflicted). Chronic health problems (i.e. long-term illnesses or health problems that must have lasted, or be expected to last, at least 6 months) and smoking (i.e. regular smokers and nonsmokers/occasional smokers) was coded as dichotomous variables. Statistics The associations between educational level and LTSL were calculated as odds ratios with 95% confidence intervals. The analyses were stratified by sex and conducted stepwise with both separate and simultaneous adjustments. First, in the initial models, we adjusted for age and region (initial ORs). Next, we added each work-related mechanical factor, one at the time. Then, we adjusted for all the mechanical factors simultaneously. The same procedure was applied for the work-related psychosocial factors and for the health and lifestyle factors. Finally, we added all factors simultaneously (adjusted ORs). The impact (%) of each separate factor or set of factors on the educational gradient was estimated as the percentage change in OR when comparing the adjusted ORs to the initial ORs using the formula: (ORadjusted – ORinitial) / (ORinitial – 1)*100. Statistical analyses were conducted with SPSS Statistics for Windows version 20.0 (IBM Corporation, Armonk, NY, USA). Results In total, 9.4% (635 individuals) were classified with LTSL during 2010. There was a strong social gradient ranging from 12.4% (elementary) to 3.3% (university/college ≥4 years) for men and from 15.4% (elementary) to 4.6% (university/college ≥4 years) among women (Tables I and II). In men, educational groups 3–5 had a significantly higher risk of LTSL after adjustment for age and region. ORs ranged from 2.65 to 4.50. Compared to the age- and region-adjusted model, adjustment for mechanical factors reduced ORs by 24–32%. The most important factors were squatting/kneeling, heavy physical work, and awkward lifting. Adjusting for psychosocial factors reduced ORs by 21–32% in educational groups 3–5. The most important psychosocial factors were monotonous work, limited possibilities for development, low job control, and low levels of leadership support. The combined impact of workrelated factors ranged from 41–44%. Finally, adjusting for health and lifestyle variables, ORs were reduced by 31–41% in educational groups 3–5, the most important factor being musculoskeletal complaints. When all variables were entered simultaneously, the
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
−11 −7 −11 0 −15
1.25 (0.69−2.27) 1.28 (0.71−2.33) 1.26 (0.69−2.29) 1.24 (0.68−2.25) 1.24 (0.68−2.25) 1.28 (0.70−2.31) 1.24 (0.69−2.26) 1.25 (0.69−2.26) 1.24 (0.68−2.25) 1.27 (0.70−2.30) 1.23 (0.68−2.24) 1.29 (0.71−2.34) 1.25 (0.69−2.26) 1.27 (0.70−2.31) 1.23 (0.68−2.23) 1.29 (0.71−2.34) 1.25 (0.69−2.27) 1.24 (0.68−2.25) 1.21 (0.67−2.20) 1.18 (0.65−2.15) 1.14 (0.63−2.09)
1.24 (0.68−2.25) 1.35 (0.74−2.46) 1.24 (0.68−2.27) 1.22 (0.67−2.23) 1.21 (0.66−2.21) 1.16 (0.63−2.14)
Ref.
Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref.
Ref. Ref. Ref.
Ref. Ref. Ref.
Age and region Mechanical factors Neck flexion Hand/arm repetition Hands above shoulder Squatting/kneeling Standing Upper body forward bend Awkward lifting Heavy lifting Heavy physical work Whole body vibration All combined Psychosocial factors Job demands Job control Role conflict Emotional demands Leader support Bullying Monotonous work Possibilites for development All combined Mechanical and psychosocial factors combined Health and lifestyle Smoking Chronic health Musculoskeletal complaints Psychological distress All combined All variables combined
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
2.56 (1.48−4.41) 2.14 (1.22−3.73) 1.84 (1.03−3.29)
2.49 (1.44−4.31) 2.61 (1.51−4.51) 2.25 (1.29−3.90)
2.30 (1.32−4.00) 1.98 (1.12−3.50)
2.69 (1.56−4.63) 2.50 (1.45−4.31) 2.72 (1.58−4.68) 2.74 (1.59−4.71) 2.49 (1.44−4.30) 2.60 (1.51−4.48) 2.38 (1.38−4.12) 2.46 (1.42−4.25)
2.39 (1.38−4.13) 2.45 (1.42−4.24) 2.35 (1.36−4.06) 2.52 (1.46−4.34) 2.16 (1.24−3.78)
2.59 (1.50−4.46) 2.65 (1.54−4.55) 2.49 (1.44−4.30) 2.28 (1.32−3.95) 2.45 (1.41−4.26) 2.53 (1.47−4.37)
2.65 (1.54−4.56)
OR (95% CI)
1391 (7.90)
3
−5 −31 −49
−10 −2 −24
−21 −41
2 −9 4 5 −10 −3 −16 −12
−16 −12 −18 −8 −30
−4 0 −10 −22 −12 −7
Change (%)
2.47 (1.31−4.66) 2.01 (1.05−3.87) 1.74 (0.87−3.49)
2.44 (1.29−4.61) 2.59 (1.37−4.91) 2.21 (1.16−4.21)
2.17 (1.13−4.17) 1.96 (1.01−3.82)
2.82 (1.50−5.28) 2.46 (1.31−4.63) 2.82 (1.50−5.28) 2.79 (1.49−5.22) 2.40 (1.26−4.55) 2.64 (1.41−4.94) 2.42 (1.29−4.54) 2.46 (1.30−4.63)
2.47 (1.32−4.63) 2.44 (1.30−4.61) 2.40 (1.28−4.51) 2.60 (1.39−4.87) 2.30 (1.20−4.38)
2.67 (1.43−5.00) 2.72 (1.45−5.09) 2.55 (1.36−4.79) 2.40 (1.28−4.51) 2.43 (1.28−4.60) 2.56 (1.36−4.80)
2.71 (1.45−5.07)
OR (95% CI)
318 (10.10)
4
−14 −41 −57
−16 −7 −29
−32 −44
6 −15 6 5 −18 −4 −17 −15
−14 −16 −18 −6 −24
−2 1 −9 −18 −16 −9
Change (%)
4.28 (2.41−7.60) 3.39 (1.87−6.14) 2.81 (1.49−5.30)
4.10 (2.30−7.32) 4.43 (2.48−7.92) 3.61 (2.02−6.46)
3.74 (2.06−6.78) 3.06 (1.65−5.67)
4.73 (2.67−8.40) 4.24 (2.39−7.53) 4.84 (2.73−8.58) 4.77 (2.69−8.47) 4.26 (2.40−7.57) 4.40 (2.48−7.80) 3.71 (2.07−6.64) 4.06 (2.28−7.23)
3.80 (2.13−6.78) 4.05 (2.26−7.23) 3.83 (2.15−6.84) 4.21 (2.37−7.48) 3.38 (1.86−6.14)
4.28 (2.41−7.60) 4.41 (2.49−7.81) 4.16 (2.33−7.40) 3.77 (2.12−6.73) 4.04 (2.25−7.26) 4.18 (2.35−7.43)
4.50 (2.54−7.96)
OR (95% CI)
534 (12.40)
5
−6 −32 −48
−11 −2 −25
−22 −41
7 −7 10 8 −7 −3 −23 −13
−20 −13 −19 −8 −32
−6 −3 −10 −21 −13 −9
Change (%)
Group 1: university/college ≥4 years; Group 2: university/college 4 years.; Group 3: upper secondary; Group 4: incomplete upper secondary; Group 5: elementary. OR: odds ratio for LTSL for educational groups 2–5 compared with the reference respondents with ≥4 years of university/college education (group 1); Change: change in OR when comparing the initial OR to the adjusted OR. LTSL: long-term sick leave.
−19 −22 −41
−11 30 −11
−33 −48
7 −7 0 −15 7 −7 −11 −22
−7 4 −4 −11 −11 4
1.27 (0.70−2.31)
Ref.
Change (%)
956 (4.20)
484 (3.30)
No. of respondents (cases of LTSL, %) OR (95% CI)
2
Educational group
1
Explanatory variable
Table I. Logistic regression analyses: medically certified sick leave regressed on educational level groups: the effect of adjusting for mechanical and psychosocial working conditions and health and lifestyle variables: men.
Work-related educational differences in sick leave 331
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
1219 (9.80) OR (95% CI) 2.15 (1.22−3.82) 2.18 (1.23−3.86) 2.23 (1.26−3.96) 2.13 (1.20−3.78) 2.01 (1.13−3.56) 1.92 (1.08−3.42) 2.09 (1.18−3.70) 2.03 (1.15−3.60) 2.12 (1.19−3.75) 2.06 (1.16−3.65) 2.16 (1.22−3.83) 1.94 (1.09−3.46) 2.17 (1.23−3.85) 2.13 (1.20−3.77) 2.13 (1.20−3.77) 2.03 (1.14−3.60) 2.03 (1.15−3.61) 2.09 (1.18−3.71) 2.05 (1.16−3.64) 2.09 (1.18−3.70) 1.81 (1.01−3.22) 1.75 (0.98−3.15)
2.09 (1.18−3.71) 2.00 (1.13−3.56) 1.97 (1.11−3.50) 2.10 (1.19−3.73) 1.89 (1.06−3.36) 1.65 (0.92−2.98)
306 (4.60)
Ref.
Ref. Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref.
Ref. Ref. Ref.
Ref. Ref. Ref.
No. of respondents (cases of LTSL, %)
Age and region Neck flexion Hand/arm repetition Hands above shoulder Squatting/kneeling Standing Upper body forward bend Awkward lifting Heavy lifting Heavy physical work Whole body vibration All combined Psychosocial factors Job demands Job control Role conflict Emotional demands Leader support Bullying Monotonous work Possibilites for development All combined Mechanical and psychosocial factors combined Health and lifestyle Smoking Chronic health Musculoskeletal complaints Psychological distress All combined All variables combined −4 −23 −43
−5 −13 −16
−30 −35
2 −2 −2 −10 −10 −5 −9 −5
−10 −3 −8 1 −18
3 7 −2 −12 −20 −5
Change (%)
3.57 (2.01−6.34) 2.93 (1.64−5.24) 2.57 (1.41−4.67)
3.52 (1.98−6.24) 3.41 (1.92−6.07) 3.08 (1.73−5.49)
3.11 (1.74−5.59) 2.88 (1.59−5.19)
3.84 (2.17−6.82) 3.66 (2.07−6.49) 3.94 (2.22−6.99) 3.59 (2.02−6.36) 3.68 (2.07−6.52) 3.61 (2.04−6.40) 3.13 (1.76−5.58) 3.55 (2.00−6.30)
3.40 (1.91−6.03) 3.60 (2.03−6.38) 3.47 (1.95−6.15) 3.74 (2.11−6.62) 3.10 (1.73−5.55)
3.72 (2.10−6.59) 3.69 (2.08−6.54) 3.69 (2.08−6.53) 3.52 (1.98−6.24) 3.47 (1.95−6.15) 3.24 (1.82−5.77) 3.48 (1.96−6.18)
OR (95% CI)
841 (15.60)
3
−6 −29 −42
−7 −11 −24
−22 −31
4 −2 8 −5 −1 −4 −22 −6
−12 −4 −9 1 −23
−1 −1 −7 −9 −18 −9
Change (%)
3.42 (1.82−6.44) 2.76 (1.45−5.24) 2.20 (1.13−4.27)
3.34 (1.78−6.30) 3.21 (1.70−6.05) 2.96 (1.57−5.60)
2.74 (1.43−5.25) 2.44 (1.27−4.71)
3.74 (1.99−7.03) 3.47 (1.85−6.52) 3.87 (2.06−7.29) 3.51 (1.87−6.60) 3.36 (1.78−6.32) 3.43 (1.82−6.45) 2.82 (1.49−5.35) 3.29 (1.74−6.20)
3.19 (1.69−6.01) 3.45 (1.84−6.49) 3.26 (1.73−6.14) 3.53 (1.88−6.63) 2.80 (1.47−5.31)
3.51 (1.87−6.60) 3.48 (1.85−6.55) 3.45 (1.83−6.49) 3.35 (1.78−6.30) 3.26 (1.73−6.13) 3.11 (1.65−5.87) 3.33 (1.77−6.27)
OR (95% CI)
354 (14.10)
4
−4 −30 −52
−7 −12 −22
−31 −43
9 −2 14 0 −6 −3 −27 −9
−13 −2 −10 1 −28
−1 −2 −6 −10 −16 −7
Change (%)
3.46 (1.87−6.39) 2.61 (1.39−4.89) 1.85 (0.96−3.59)
3.41 (1.84−6.33) 3.26 (1.76−6.04) 2.86 (1.54−5.33)
2.77 (1.47−5.23) 2.23 (1.16−4.28)
3.75 (2.03−6.93) 3.51 (1.90−6.49) 3.92 (2.12−7.25) 3.67 (1.99−6.76) 3.58 (1.94−6.62) 3.54 (1.91−6.54) 2.77 (1.48−5.19) 3.39 (1.83−6.27)
3.15 (1.70−5.84) 3.46 (1.87−6.40) 3.09 (1.66−5.73) 3.66 (1.98−6.76) 2.51 (1.33−4.74)
3.65 (1.98−6.73) 3.51 (1.90−6.49) 3.49 (1.89−6.46) 3.37 (1.82−6.24) 3.26 (1.76−6.04) 2.89 (1.54−5.41) 3.22 (1.74−5.97)
OR (95% CI)
356 (15.40)
5
−7 −39 −68
−9 −15 −30
−33 −54
−19 −7 −21 0 −43 4 −5 10 1 −3 −4 −33 −10
−5 −6 −11 −15 −29 −16
Change (%)
Group 1: university/college ≥4 years; Group 2: university/college 4 years.; Group 3: upper secondary; Group 4: incomplete upper secondary; Group 5: elementary. OR: odds ratio for LTSL for educational groups 2–5 compared with the reference respondents with ≥4 years of university/college education (group 1); Change: change in OR when comparing the initial OR to the adjusted OR. LTSL: long-term sick leave.
2
Educational group
1
Explanatory variable
Table II. Logistic regression analyses: medically certified sick leave regressed on educational level groups: the effect of adjusting for mechanical and psychosocial working conditions and health and lifestyle variables: women.
332 T. Sterud and H.A. Johannessen
Work-related educational differences in sick leave 333 differences between group 1 and groups 3–5 was reduced by 48–57%, but a significantly unexplained increased risk for LTLS was still observed in group 3 and 5. In women, educational groups 2–5 had a significantly higher risk of LTSL, after adjustment for age and region, ORs ranged from 2.15 to 3.72. Compared to the age- and region-adjusted model, adjusting for mechanical factors reduced ORs by 18–43. The most important factor across all educational groups was standing. Other important factors were squatting/kneeling, awkward lifting, and heavy physical work. Adjusting for psychosocial factors reduced ORs by 22–33%. In educational groups 3–5, the most important factors were monotonous work, whereas in educational group 2 several factors were equally important (emotional demands, leader support, and monotonous work). The combined impact of work-related factors ranged from 31–54% in educational groups 2–5 and adjustment for health and lifestyle variables reduced the ORs by 23–39%. The most important factor for all educational groups was musculoskeletal complaints. When all variables were entered simultaneously, the differences between group 1 and groups 2–5 was reduced with 42–68%, but a significantly unexplained increased risk for LTLS was still observed in groups 3–5. Discussion We found a substantial social gradient in LTSL in both men and women. LTSL was roughly 3–4-times more common among persons with basic or upper secondary education level compared to the reference group (≥4 years of university/college). Among women, a 2-fold increased risk was also observed among respondents with
Do work-related mechanical and psychosocial factors contribute to the social gradient in long-term sick leave: A prospective study of the general working population in Norway Tom Sterud and Håkon A Johannessen Scand J Public Health 2014 42: 329 originally published online 30 January 2014 DOI: 10.1177/1403494814521506 The online version of this article can be found at: http://sjp.sagepub.com/content/42/3/329
Published by: http://www.sagepublications.com
Additional services and information for Scandinavian Journal of Public Health can be found at: Email Alerts: http://sjp.sagepub.com/cgi/alerts Subscriptions: http://sjp.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav
>> Version of Record - Apr 23, 2014 OnlineFirst Version of Record - Jan 30, 2014 What is This?
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
521506 research-article2014
SJP0010.1177/1403494814521506T. Sterud and H.A. JohannessenWork-related educational differences in sick leave
Scandinavian Journal of Public Health, 2014; 42: 329–334
Original Article
Do work-related mechanical and psychosocial factors contribute to the social gradient in long-term sick leave: A prospective study of the general working population in Norway
Tom Sterud & Håkon A Johannessen Department of Occupational Health Surveillance, National Institute of Occupational Health, Oslo, Norway
Abstract Aims: A social gradient in long-term sick leave (LTSL) rates is well established, but only a few studies have examined to what extent this gradient may be explained by mechanical and psychosocial work environment factors. Methods: A randomly drawn cohort from the general population in Norway, aged 18–69 years, was interviewed in the second half of 2009 (n=12,255, response at baseline 60.9%) and followed up in national registries to the end of 2010. Eligible respondents were registered with an active employee relationship of at least 100 actual working days in 2009 and 2010 (n=6758). Based on administrative register data, respondents were coded into five educational levels (university/college ≥4 years was set as the reference group). Eight work-related psychosocial factors and 10 mechanical exposures were measured. The outcome of interest was medically confirmed LTSL ≥40 working days during 2010. Results: In total, 9.4% (635 individuals) were classified with LTSL during 2010. There was a strong social gradient ranging from 12.4% (elementary) to 3.3% (university/ college ≥4 years) among men. The corresponding figures among women were 15.4 and 4.6%. Adjusting for work-related mechanical and psychosocial factors explained between 41 and 44% of the social gradient in men. Among women, the corresponding figures were 31 and 54%. Conclusions: Work-related mechanical and psychosocial factors contribute to the social gradient in LTSL. The work-related factors that accounted for this gradient were rather similar for men and women. Key Words: Educational status, mechanical processes, occupational, prospective study, psychosocial factors, risk factors, sick leave, socioeconomic status, work, work place
Introduction A social gradient in long-term sick leave (LTSL) rates is well established [1], but only a few studies have examined to what extent this gradient can be explained by work environment factors [2–4]. Inability to work because of sickness and disability generate considerable costs to the individual as well as to society and more knowledge is needed about the impact of specific risk factors at the work place. We therefore studied differences between educational level groups in the risk of medically confirmed LTSL and the degree to which these differences were
explained when taking into account a wide range of mechanical and psychosocial risk factors in a nationwide general working population sample. Participants and methods Data were provided from the nationwide Survey of Level of Living– Working Conditions, conducted by Statistics Norway (SSB). Eligible respondents were community-living Norwegian residents aged 18–69 years. In 2009, a gross sample of 20,136 was randomly
Correspondence: Tom Sterud, Department of Occupational Health Surveillance, National Institute of Occupational Health, PO Box 8149 Dep., N-0033 Oslo. Norway. E-mail: [email protected] (Accepted 7 January 2014) © 2014 the Nordic Societies of Public Health DOI: 10.1177/1403494814521506
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
330 T. Sterud and H.A. Johannessen drawn from this population. Of these, 12,255 (60.9%) persons were interviewed between 22 June 2009 and 9 January 2010. Data on sick leave was obtained from the Norwegian Labour and Welfare Administration’s sickness benefit register. The register includes all workers aged 16–69 living in Norway and registered with an active employee relationship. Respondents who were in paid work for at least 1 hour during the reference week, or were temporarily absent from such work, and were registered with an active employee relationship of at least 100 actual working days in each year (2009 and 2010) constituted the follow-up sample in the present paper (n=6758). Measurements Long-term sick leave (LTSL) was defined as medically confirmed sick leave for a period equalling 40 or more actual working days during 2010, the year after the initial survey data was collected. Educational level was based on administrative register data and was coded into five educational levels: Group 5: elementary; Group 4: incomplete upper secondary; Group 3: upper secondary; Group 2: university/college 4 years; Group 1: university/college ≥4 years. Gender and age were based on self-reported information. Region was based on administrative register data and was coded into seven regions based on the standard for economic regions developed by Statistics Norway. Perceived mechanical workload was measured with 10 items: neck flexion; hand/arm repetition; work with hands lifted above shoulder height; squatting/kneeling; standing; work with upper body forward bend; awkward lifting; heavy physical work; whole body vibration; and heavy lifting. Scores were coded on a scale from 1 (not exposed or exposed very little of the work day) to 4 (exposed three-quarters of the work day or more). The items have been described in greater detail elsewhere [5]. Perceived psychosocial factors at work included job demands (two items), low job control (four items), role conflict (three items), low leadership support (three items), monotonous work (one item), emotional demands (two items), and bullying/harassment (three items), corresponding to those used in previous studies [6,7]. All factors were used as continuous variables (range 1–4/5), and high score indicates unfavourable exposure, except for a dichotomous variable for bullying/harassment. Self-reported health and lifestyle factors Musculoskeletal complaints (neck/shoulder, low back pain, arm and leg pain) and psychological distress
(dejection or depression and nervousness, anxiety or restlessness) was used as continuous variables ranging from 1 (not afflicted) to 4 (severely afflicted). Chronic health problems (i.e. long-term illnesses or health problems that must have lasted, or be expected to last, at least 6 months) and smoking (i.e. regular smokers and nonsmokers/occasional smokers) was coded as dichotomous variables. Statistics The associations between educational level and LTSL were calculated as odds ratios with 95% confidence intervals. The analyses were stratified by sex and conducted stepwise with both separate and simultaneous adjustments. First, in the initial models, we adjusted for age and region (initial ORs). Next, we added each work-related mechanical factor, one at the time. Then, we adjusted for all the mechanical factors simultaneously. The same procedure was applied for the work-related psychosocial factors and for the health and lifestyle factors. Finally, we added all factors simultaneously (adjusted ORs). The impact (%) of each separate factor or set of factors on the educational gradient was estimated as the percentage change in OR when comparing the adjusted ORs to the initial ORs using the formula: (ORadjusted – ORinitial) / (ORinitial – 1)*100. Statistical analyses were conducted with SPSS Statistics for Windows version 20.0 (IBM Corporation, Armonk, NY, USA). Results In total, 9.4% (635 individuals) were classified with LTSL during 2010. There was a strong social gradient ranging from 12.4% (elementary) to 3.3% (university/college ≥4 years) for men and from 15.4% (elementary) to 4.6% (university/college ≥4 years) among women (Tables I and II). In men, educational groups 3–5 had a significantly higher risk of LTSL after adjustment for age and region. ORs ranged from 2.65 to 4.50. Compared to the age- and region-adjusted model, adjustment for mechanical factors reduced ORs by 24–32%. The most important factors were squatting/kneeling, heavy physical work, and awkward lifting. Adjusting for psychosocial factors reduced ORs by 21–32% in educational groups 3–5. The most important psychosocial factors were monotonous work, limited possibilities for development, low job control, and low levels of leadership support. The combined impact of workrelated factors ranged from 41–44%. Finally, adjusting for health and lifestyle variables, ORs were reduced by 31–41% in educational groups 3–5, the most important factor being musculoskeletal complaints. When all variables were entered simultaneously, the
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
−11 −7 −11 0 −15
1.25 (0.69−2.27) 1.28 (0.71−2.33) 1.26 (0.69−2.29) 1.24 (0.68−2.25) 1.24 (0.68−2.25) 1.28 (0.70−2.31) 1.24 (0.69−2.26) 1.25 (0.69−2.26) 1.24 (0.68−2.25) 1.27 (0.70−2.30) 1.23 (0.68−2.24) 1.29 (0.71−2.34) 1.25 (0.69−2.26) 1.27 (0.70−2.31) 1.23 (0.68−2.23) 1.29 (0.71−2.34) 1.25 (0.69−2.27) 1.24 (0.68−2.25) 1.21 (0.67−2.20) 1.18 (0.65−2.15) 1.14 (0.63−2.09)
1.24 (0.68−2.25) 1.35 (0.74−2.46) 1.24 (0.68−2.27) 1.22 (0.67−2.23) 1.21 (0.66−2.21) 1.16 (0.63−2.14)
Ref.
Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref.
Ref. Ref. Ref.
Ref. Ref. Ref.
Age and region Mechanical factors Neck flexion Hand/arm repetition Hands above shoulder Squatting/kneeling Standing Upper body forward bend Awkward lifting Heavy lifting Heavy physical work Whole body vibration All combined Psychosocial factors Job demands Job control Role conflict Emotional demands Leader support Bullying Monotonous work Possibilites for development All combined Mechanical and psychosocial factors combined Health and lifestyle Smoking Chronic health Musculoskeletal complaints Psychological distress All combined All variables combined
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
2.56 (1.48−4.41) 2.14 (1.22−3.73) 1.84 (1.03−3.29)
2.49 (1.44−4.31) 2.61 (1.51−4.51) 2.25 (1.29−3.90)
2.30 (1.32−4.00) 1.98 (1.12−3.50)
2.69 (1.56−4.63) 2.50 (1.45−4.31) 2.72 (1.58−4.68) 2.74 (1.59−4.71) 2.49 (1.44−4.30) 2.60 (1.51−4.48) 2.38 (1.38−4.12) 2.46 (1.42−4.25)
2.39 (1.38−4.13) 2.45 (1.42−4.24) 2.35 (1.36−4.06) 2.52 (1.46−4.34) 2.16 (1.24−3.78)
2.59 (1.50−4.46) 2.65 (1.54−4.55) 2.49 (1.44−4.30) 2.28 (1.32−3.95) 2.45 (1.41−4.26) 2.53 (1.47−4.37)
2.65 (1.54−4.56)
OR (95% CI)
1391 (7.90)
3
−5 −31 −49
−10 −2 −24
−21 −41
2 −9 4 5 −10 −3 −16 −12
−16 −12 −18 −8 −30
−4 0 −10 −22 −12 −7
Change (%)
2.47 (1.31−4.66) 2.01 (1.05−3.87) 1.74 (0.87−3.49)
2.44 (1.29−4.61) 2.59 (1.37−4.91) 2.21 (1.16−4.21)
2.17 (1.13−4.17) 1.96 (1.01−3.82)
2.82 (1.50−5.28) 2.46 (1.31−4.63) 2.82 (1.50−5.28) 2.79 (1.49−5.22) 2.40 (1.26−4.55) 2.64 (1.41−4.94) 2.42 (1.29−4.54) 2.46 (1.30−4.63)
2.47 (1.32−4.63) 2.44 (1.30−4.61) 2.40 (1.28−4.51) 2.60 (1.39−4.87) 2.30 (1.20−4.38)
2.67 (1.43−5.00) 2.72 (1.45−5.09) 2.55 (1.36−4.79) 2.40 (1.28−4.51) 2.43 (1.28−4.60) 2.56 (1.36−4.80)
2.71 (1.45−5.07)
OR (95% CI)
318 (10.10)
4
−14 −41 −57
−16 −7 −29
−32 −44
6 −15 6 5 −18 −4 −17 −15
−14 −16 −18 −6 −24
−2 1 −9 −18 −16 −9
Change (%)
4.28 (2.41−7.60) 3.39 (1.87−6.14) 2.81 (1.49−5.30)
4.10 (2.30−7.32) 4.43 (2.48−7.92) 3.61 (2.02−6.46)
3.74 (2.06−6.78) 3.06 (1.65−5.67)
4.73 (2.67−8.40) 4.24 (2.39−7.53) 4.84 (2.73−8.58) 4.77 (2.69−8.47) 4.26 (2.40−7.57) 4.40 (2.48−7.80) 3.71 (2.07−6.64) 4.06 (2.28−7.23)
3.80 (2.13−6.78) 4.05 (2.26−7.23) 3.83 (2.15−6.84) 4.21 (2.37−7.48) 3.38 (1.86−6.14)
4.28 (2.41−7.60) 4.41 (2.49−7.81) 4.16 (2.33−7.40) 3.77 (2.12−6.73) 4.04 (2.25−7.26) 4.18 (2.35−7.43)
4.50 (2.54−7.96)
OR (95% CI)
534 (12.40)
5
−6 −32 −48
−11 −2 −25
−22 −41
7 −7 10 8 −7 −3 −23 −13
−20 −13 −19 −8 −32
−6 −3 −10 −21 −13 −9
Change (%)
Group 1: university/college ≥4 years; Group 2: university/college 4 years.; Group 3: upper secondary; Group 4: incomplete upper secondary; Group 5: elementary. OR: odds ratio for LTSL for educational groups 2–5 compared with the reference respondents with ≥4 years of university/college education (group 1); Change: change in OR when comparing the initial OR to the adjusted OR. LTSL: long-term sick leave.
−19 −22 −41
−11 30 −11
−33 −48
7 −7 0 −15 7 −7 −11 −22
−7 4 −4 −11 −11 4
1.27 (0.70−2.31)
Ref.
Change (%)
956 (4.20)
484 (3.30)
No. of respondents (cases of LTSL, %) OR (95% CI)
2
Educational group
1
Explanatory variable
Table I. Logistic regression analyses: medically certified sick leave regressed on educational level groups: the effect of adjusting for mechanical and psychosocial working conditions and health and lifestyle variables: men.
Work-related educational differences in sick leave 331
Downloaded from sjp.sagepub.com at TEXAS SOUTHERN UNIVERSITY on December 10, 2014
1219 (9.80) OR (95% CI) 2.15 (1.22−3.82) 2.18 (1.23−3.86) 2.23 (1.26−3.96) 2.13 (1.20−3.78) 2.01 (1.13−3.56) 1.92 (1.08−3.42) 2.09 (1.18−3.70) 2.03 (1.15−3.60) 2.12 (1.19−3.75) 2.06 (1.16−3.65) 2.16 (1.22−3.83) 1.94 (1.09−3.46) 2.17 (1.23−3.85) 2.13 (1.20−3.77) 2.13 (1.20−3.77) 2.03 (1.14−3.60) 2.03 (1.15−3.61) 2.09 (1.18−3.71) 2.05 (1.16−3.64) 2.09 (1.18−3.70) 1.81 (1.01−3.22) 1.75 (0.98−3.15)
2.09 (1.18−3.71) 2.00 (1.13−3.56) 1.97 (1.11−3.50) 2.10 (1.19−3.73) 1.89 (1.06−3.36) 1.65 (0.92−2.98)
306 (4.60)
Ref.
Ref. Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref.
Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref.
Ref. Ref.
Ref. Ref. Ref.
Ref. Ref. Ref.
No. of respondents (cases of LTSL, %)
Age and region Neck flexion Hand/arm repetition Hands above shoulder Squatting/kneeling Standing Upper body forward bend Awkward lifting Heavy lifting Heavy physical work Whole body vibration All combined Psychosocial factors Job demands Job control Role conflict Emotional demands Leader support Bullying Monotonous work Possibilites for development All combined Mechanical and psychosocial factors combined Health and lifestyle Smoking Chronic health Musculoskeletal complaints Psychological distress All combined All variables combined −4 −23 −43
−5 −13 −16
−30 −35
2 −2 −2 −10 −10 −5 −9 −5
−10 −3 −8 1 −18
3 7 −2 −12 −20 −5
Change (%)
3.57 (2.01−6.34) 2.93 (1.64−5.24) 2.57 (1.41−4.67)
3.52 (1.98−6.24) 3.41 (1.92−6.07) 3.08 (1.73−5.49)
3.11 (1.74−5.59) 2.88 (1.59−5.19)
3.84 (2.17−6.82) 3.66 (2.07−6.49) 3.94 (2.22−6.99) 3.59 (2.02−6.36) 3.68 (2.07−6.52) 3.61 (2.04−6.40) 3.13 (1.76−5.58) 3.55 (2.00−6.30)
3.40 (1.91−6.03) 3.60 (2.03−6.38) 3.47 (1.95−6.15) 3.74 (2.11−6.62) 3.10 (1.73−5.55)
3.72 (2.10−6.59) 3.69 (2.08−6.54) 3.69 (2.08−6.53) 3.52 (1.98−6.24) 3.47 (1.95−6.15) 3.24 (1.82−5.77) 3.48 (1.96−6.18)
OR (95% CI)
841 (15.60)
3
−6 −29 −42
−7 −11 −24
−22 −31
4 −2 8 −5 −1 −4 −22 −6
−12 −4 −9 1 −23
−1 −1 −7 −9 −18 −9
Change (%)
3.42 (1.82−6.44) 2.76 (1.45−5.24) 2.20 (1.13−4.27)
3.34 (1.78−6.30) 3.21 (1.70−6.05) 2.96 (1.57−5.60)
2.74 (1.43−5.25) 2.44 (1.27−4.71)
3.74 (1.99−7.03) 3.47 (1.85−6.52) 3.87 (2.06−7.29) 3.51 (1.87−6.60) 3.36 (1.78−6.32) 3.43 (1.82−6.45) 2.82 (1.49−5.35) 3.29 (1.74−6.20)
3.19 (1.69−6.01) 3.45 (1.84−6.49) 3.26 (1.73−6.14) 3.53 (1.88−6.63) 2.80 (1.47−5.31)
3.51 (1.87−6.60) 3.48 (1.85−6.55) 3.45 (1.83−6.49) 3.35 (1.78−6.30) 3.26 (1.73−6.13) 3.11 (1.65−5.87) 3.33 (1.77−6.27)
OR (95% CI)
354 (14.10)
4
−4 −30 −52
−7 −12 −22
−31 −43
9 −2 14 0 −6 −3 −27 −9
−13 −2 −10 1 −28
−1 −2 −6 −10 −16 −7
Change (%)
3.46 (1.87−6.39) 2.61 (1.39−4.89) 1.85 (0.96−3.59)
3.41 (1.84−6.33) 3.26 (1.76−6.04) 2.86 (1.54−5.33)
2.77 (1.47−5.23) 2.23 (1.16−4.28)
3.75 (2.03−6.93) 3.51 (1.90−6.49) 3.92 (2.12−7.25) 3.67 (1.99−6.76) 3.58 (1.94−6.62) 3.54 (1.91−6.54) 2.77 (1.48−5.19) 3.39 (1.83−6.27)
3.15 (1.70−5.84) 3.46 (1.87−6.40) 3.09 (1.66−5.73) 3.66 (1.98−6.76) 2.51 (1.33−4.74)
3.65 (1.98−6.73) 3.51 (1.90−6.49) 3.49 (1.89−6.46) 3.37 (1.82−6.24) 3.26 (1.76−6.04) 2.89 (1.54−5.41) 3.22 (1.74−5.97)
OR (95% CI)
356 (15.40)
5
−7 −39 −68
−9 −15 −30
−33 −54
−19 −7 −21 0 −43 4 −5 10 1 −3 −4 −33 −10
−5 −6 −11 −15 −29 −16
Change (%)
Group 1: university/college ≥4 years; Group 2: university/college 4 years.; Group 3: upper secondary; Group 4: incomplete upper secondary; Group 5: elementary. OR: odds ratio for LTSL for educational groups 2–5 compared with the reference respondents with ≥4 years of university/college education (group 1); Change: change in OR when comparing the initial OR to the adjusted OR. LTSL: long-term sick leave.
2
Educational group
1
Explanatory variable
Table II. Logistic regression analyses: medically certified sick leave regressed on educational level groups: the effect of adjusting for mechanical and psychosocial working conditions and health and lifestyle variables: women.
332 T. Sterud and H.A. Johannessen
Work-related educational differences in sick leave 333 differences between group 1 and groups 3–5 was reduced by 48–57%, but a significantly unexplained increased risk for LTLS was still observed in group 3 and 5. In women, educational groups 2–5 had a significantly higher risk of LTSL, after adjustment for age and region, ORs ranged from 2.15 to 3.72. Compared to the age- and region-adjusted model, adjusting for mechanical factors reduced ORs by 18–43. The most important factor across all educational groups was standing. Other important factors were squatting/kneeling, awkward lifting, and heavy physical work. Adjusting for psychosocial factors reduced ORs by 22–33%. In educational groups 3–5, the most important factors were monotonous work, whereas in educational group 2 several factors were equally important (emotional demands, leader support, and monotonous work). The combined impact of work-related factors ranged from 31–54% in educational groups 2–5 and adjustment for health and lifestyle variables reduced the ORs by 23–39%. The most important factor for all educational groups was musculoskeletal complaints. When all variables were entered simultaneously, the differences between group 1 and groups 2–5 was reduced with 42–68%, but a significantly unexplained increased risk for LTLS was still observed in groups 3–5. Discussion We found a substantial social gradient in LTSL in both men and women. LTSL was roughly 3–4-times more common among persons with basic or upper secondary education level compared to the reference group (≥4 years of university/college). Among women, a 2-fold increased risk was also observed among respondents with
