Content uploaded by Ute Bültmann
Author content
All content in this area was uploaded by Ute Bültmann on Apr 12, 2016
Content may be subject to copyright.
Int Arch Occup Environ Health (2009) 82:1133–1138
DOI 10.1007/s00420-009-0417-6
123
ORIGINAL ARTICLE
Predicting long-term sickness absence and early retirement
pension from self-reported work ability
Lea Sell
Received: 15 August 2008 / Accepted: 11 March 2009 / Published online: 14 April 2009
© Springer-Verlag 2009
Abstract
Objective The aim of this paper is to examine the rela-
tionship between self-reported work ability and long-term
term of sickness absence or early retirement from the
labour market.
Methods Data on work ability were retrieved from a rep-
resentative cohort study of Danish wage earners and linked
with a register of social payment transfers. In all, 4.743
individuals were followed from 2001 to 2005. Cox regres-
sion was used for the analyses.
Results A one point decrease in perceived work ability, on a
10 point scale, was associated with an increased risk of long-
term sickness absence (LTSA) of 15.1% (95% CI 12–19%,
P< 0.0001) and an increased risk of early retirement from the
labour market of 33% (95% CI 20–48%, P< 0.0001).
Conclusions Reports of reduced work ability were signiW-
cantly associated with both an increased risk of onset of
LTSA and early retirement from the labour market, after
adjustment for socio-demographic characteristics and life-
style variables.
Keywords Work ability · Long-term sickness absence ·
Early retirement pension · Work environment ·
Labour market
Introduction
In the ageing societies in both Europe and some Asian
countries, such as Japan and Thailand (UN Department of
Economic and Social AVairs PD2 2007), the increasing
fractions of the populations not being able to support them
selves is becoming a threat to the welfare and the capability
of the societies to maintain economic growth. In the health
care sector and branches characterized by high physical
demands in general, workers with reduced work ability,
increased long-term sickness absence (LTSA) and many
cases of early retirement (Lund et al. 2007; Fjell et al. 2007;
Hasselhorn et al. 2007) are already causing great concerns.
Due to these circumstances there is a growing awareness to
measure work ability and to identify possible causes of
reduced work ability as well as consequences. Earlier stud-
ies suggest that workers above 45 years of age are very het-
erogeneous in regard to work ability (Ilmarinen and Tuomi
1992) and that interventions to prevent reduced work abil-
ity can have signiWcant positive impacts (Nurminen et al.
2002).
Work ability has most frequently been assessed with the
work ability index (WAI) (Tuomi et al. 1998), which apart
from being used in research, also has been used in health
examinations and workplace surveys performed by health
professionals. However, the WAI is a relatively complex
measure, which is based on seven composite items and is
therefore seldom applied in larger surveys. For the purposes
of large surveys, a brief, single-item measure of work abil-
ity is more likely to be used. A recent Finnish study has
shown good results, predicting sickness absence from a
question asking the employees on a production work site to
evaluate their own work ability on a 10 point Likert scale
(Nygard et al. 2005). In the present paper, we use a similar
single-item measure based on the respondents’ view of cur-
rent work ability compared to when it is at its best.
Earlier studies on LTSA have identiWed a number of risk
factors, like psychosocial risk factors in the work environ-
ment (Lund et al. 2005; Melchior et al. 2003) and physical
risk factors (Vingard et al. 2005; Lund et al. 2006). Life
L. Sell (&)
National Research Centre for the Working Environment,
Copenhagen, Denmark
e-mail: lse@nrcwe.dk
1134 Int Arch Occup Environ Health (2009) 82:1133–1138
123
style factors as well have been shown to be associated with
LTSA (Christensen et al. 2007). As for risk factors for early
retirement pension (ERP) likewise psychosocial and the
physical work environment factors have been established as
risk factors (Lund et al. 2005). Work ability can be charac-
terized as an intermediate variable (MacKinnon et al.
2007), which transmits the eVects from work environment
and life style to sickness absence and labour market partici-
pation variables, like early retirement. This stresses the fact
that assessing work ability is important because it might be
an indicator of future more serious outcomes like LTSA or
early retirement from the labour market.
The aim of this paper is to examine the relationships
between self-reported work ability and later onset of LTSA
or ERP. The study population is a representative sample of
the Danish wage earners followed in the period 2001–2005.
Earlier case studies have already established links between
reduced work ability and an increased risk of episodes of
LTSA (Reiso et al. 2001; Nygard et al. 2005; Kujala et al.
2006) as well as early retirement from the labour market
(Hopsu et al. 2005; Liira et al. 2000; Salonen et al. 2003).
But no studies so far have to our knowledge tested if self-
reported work ability can predict LTSA and early retire-
ment for the general working population.
Methods
Data and the population
Information on lifestyle, demographic details and work
ability in the working population were retrieved from the
Danish Work Environment Cohort Study (DWECS) (Burr
et al. 2003). Data on LTSA and ERP were obtained from
the Danish national register of social transfer payments
(DREAM). The cohort in DWECS started out with a panel
of a simple random sample drawn in 1990 from the central
population register consisting of 9.653 individuals, aged
18–59 years per October 1, 1990 of which 8,664 participated
(90%). People in this panel were reinterviewed in 1995,
with a response rate of 80%, and in 2000, with a response
rate 75%, and the panel is continuously adjusted for age
and immigration by including new subjects. In the present
study the baseline population consists of the participants in
the 2000 survey who met three criteria: they were wage
earners in 2000, had not reached the pensioning age in 2005
and were not emigrated by 2005. After we excluded those
participants not meeting the criteria or with missing values
on any of the analysed items, the dataset consisted of 4.743
participants. Table 1 summarizes the baseline character-
istics for the study population divided into major job
groups.
Measurements
The question on work ability in 2000 was formulated as fol-
lows: “Imagine that your work ability is worth 10 points
when it is at its best. How many points would you give your
present work ability?” The scale was reversed so that 1 rep-
resents the highest and 10 the lowest level of work ability in
order to be able to interpret results directly in terms of
increase in risk for a one point decrease in work ability.
Age may pose a bias to the answers to this question since
when a 20 years old compares his or her present work abil-
ity with the work ability when it is at its best, the score
could have completely diVerent meaning than when a per-
son of 50 years of age reports the same score. For control
purposes, identical answers for diVerent age groups are
related to the answers to another question concerning work
ability, given in the same questionnaire. This question
reads: “Is your work ability reduced due to diseases, acci-
dents or toil?” With answers falling in four categories:
“Yes, indeed”, “yes, to some extent”, “no, not much” or
“no”.
Long-term sickness absence is deWned as receiving sick-
ness absence compensation for two consecutive weeks or
more during the follow-up period from January 1, 2001 to
December 31, 2005. An individual was censored as early
Table 1 Summary of socio-economic variables in baseline population (2000) divided on job groups (n= 4.743)
White collar
workers n=903
Teachers/shop
workers n=536
Blue collar
workers n=872
Non-skilled
workers n= 1.844
Care workers
n=588
Age in years (mean, SD) 40.3 (9.5) 39.2 (11.09 39.1 (10.4) 37.0 (11.3) 40.0 (9.9)
Education in years (mean, SD) 13.9 (2.3) 13.9 (2.2) 11.4 (2.1) 12.6 (2.5) 13.4 (2.4)
Income: DKK (mean, SD) 14.010 12.762 11.570 11.986 10.330
Male (%) 42.5 44.2 77.2 56.0 10.4
Female (%) 57.5 55.8 22.8 44.0 90.6
Civil status (%)
Cohabiting 78.9 78.4 78.4 71.6 76.5
Children 54.6 49.6 51.0 44.1 40.3
Int Arch Occup Environ Health (2009) 82:1133–1138 1135
123
retired from the labour market if he or she received ERP
during the same follow-up period, from January 1, 2001 to
December 31, 2005.
Covariates included in the analysis were gender, age in
years, monthly wage after tax, body mass index (BMI),
civil status and the sum of years of education. Life style
factors were smoking (smoking more than 20 cigarettes a
day or less), alcohol consumption (drinks a day) and physi-
cal activity level in leisure time (Level 1: almost com-
pletely physically inactive or light physical activity for
<2 h/week. Level 2: light physical activity for 2–4 h/week,
Level 3: light physical activity for more than 4 h/week or
more strenuous physical activity for 2–4 h/week. Level 4:
more vigorous physical activity for more than 4 h a week or
regular hard workout, and perhaps competitions, several
times per week.
Statistical analyses
The conceptual model
Earlier studies on the same cohort have established pro-
spective relationships between risk factors in the work
environment and LTSA (Lund et al. 2005, 2006), and
between work environmental risk factors and self-reported
work ability (Sell et al. 2008). Unlike if, for example, we
were analysing short terms of sickness absence represent-
ing less serious illnesses, having ERP and LTSA as out-
come variables and work ability as an independent variable,
the impacts from the work environment on these outcomes
can be assumed to be mediated by reductions in work abil-
ity. Work ability can thus be justiWed to function as a medi-
ating variable in the analyses (MacKinnon et al. 2007) and
the work environmental factors can be omitted from the
analysis. Since life style both aVects work ability and
LTSA and ERP these are included in the analysis as con-
founders as not doing so could lead to incorrect estimates of
the relation between work ability and LTSA and ERP
although the number of covariates may seem a lot. More-
over, basic socio-economic variables which are important
for the validity of the results are included in the model as
covariates.
The statistical model
The data on LTSA and ERP correspond to survival times
which in most cases are censored as it is only possible to
follow the subjects until 2005 in DREAM at this moment.
After 2005 the status of the individuals in the cohort is
unknown. In the case an individual has an onset of LTSA or
ERP in the period 2001–2005, the survival times were non-
censored and referred to as event times. The Cox propor-
tional hazard model (Cox 1972) is used for modelling the
probability of LTSA and ERP, in the period 2001–2005.
Self-reported work ability, lifestyle and demographic fac-
tors were all reported in year 2000. Initially, correlation
analysis was performed on all explanatory variables. The
correlation coeYcient was below 0.30 for all pairs of covar-
iates. Data were analysed with the statistical computer pro-
gram SAS 8.2, using the PHREG procedure. Results are
presented in hazard ratios, expressing the estimated change
of risk, for a one point change in the explanatory variable.
Results
During the 5-year period, 34.8% of the study cohort
(n= 1.648) experienced at least one spell of sickness
absence of 2 weeks or more and 1.5% of the study cohort
(n= 72) entered an early retirement scheme. Ten per cent
reported 7 points or less on the original work ability scale,
20.2 reported 8 points, 24.8% reported 9 points and 45.0
reported 10 points. Tables 2 and 3 summarize the results for
the main analyses. As can be seen from Table 2, a one point
decrease in perceived work ability is associated with an
increased risk of LTSA of 15%. Age is associated with a
lower risk of LTSA, with a just about 1% decrease in risk
of LTSA for each additional year of age. Both higher wage
and higher education are associated with a lower risk of
LTSA; for each 1,000 DKR increase in wage level, the risk
of LTSA decreases by 5% and for each additional year of
education, the risk of LTSA decreases by 6%. As for life
style factors, the risk of LTSA increased with as much as
38% when smoking more than 20 cigarettes a day, and a
Table 2 Results when estimating the risk of a spell of 2 weeks sick-
ness absence or more (LTSA)
aDichotomous variables
LTSA (n= 4.743)
Hazard ratio 95% CI P value
Work ability points 2000 1.15 1.11–1.19 <0.0001
Gender (male = 1,
female = 2)a1.07 0.95–1.19 0.2486
Age in years 0.99 0.99–1.00 0.0014
Monthly wages after
tax in 1,000 kr.
0.95 0.94–0.97 <0.0001
Children 1.02 0.97–1.07 0.4072
Cohabitinga0.90 0.80–1.02 0.1093
Educational level in years 0.94 0.92–0.96 <0.0001
20 cigarettes or more a daya1.38 1.10–1.72 0.0054
BMI in 2000 1.02 1.00–1.03 0.0059
Drinks a day 1.02 0.99–1.06 0.1947
Physical activity
level in leisure
1.00 0.95–1.06 0.8831
1136 Int Arch Occup Environ Health (2009) 82:1133–1138
123
one point increase in BMI increases the risk of LTSA with
about 2%. Finally, civil status, alcohol consumption and
level of physical activity did not show signiWcance in this
analysis.
Before wages was entered in the analysis, gender was
signiWcant, being a female increasing the risk of LTSA 22%
(OR 1.22, CI 1.10–1.36, P< 0.01), leaving the other covar-
iates including work ability more or less the same. Whereas
leaving out the work ability measure only changes the haz-
ard ratio for age, which then becomes insigniWcant (OR
0.995, CI 0.991–1.00, P= 0.65). No changes are seen in the
size and signiWcance of the other covariates. These results
are presented in Appendix 1.
The analyses on LTSA were initially completed on all
sub populations according to Table 1 with no major diVer-
ences found, except for care workers for which work ability
was still signiWcantly related to LTSA but with the size of
the coeYcient being smaller than for the other sub popula-
tions. Doing the analysis on ERP divided on job groups was
not possible due to lack of power.
With regard to ERP, a one point decrease in the per-
ceived work ability increases the risk of early retirement by
33%. SigniWcant covariates are age, where the risk is
increased by just about 10% for each year. Low wage has a
larger association with ERP than was the case for LTSA;
for each 1,000 DKR increase in wage, the risk of early
retirement decreases by 20%. Unlike the results for LTSA,
BMI, smoking and education did not seem to have any
impact on risk for early retirement. For physical activity
though, for each increase in the level of physical activity in
leisure, the risk of ERP is 44% lower.
Before wages was entered in the analysis, a one point
decrease in the perceived work ability increases the risk of
early retirement by 46% (OR 1.46, CI 1.33–1.60). Omitting
the work ability measure from the analysis increased the
eVect from wages from 20 to 24% (OR 0.76, CI 0.70–0.82).
This gives an impact on work ability of 33% from leaving
out wages and an eVect from leaving out work ability on
wages, of 5%. Other covariates are not inXuenced by the
exclusion of wages or work ability (Appendix 1).
In order to test the potential bias from age on reports of
work ability on the 10 points scale, the frequency of reports
of reductions in work ability due to diseases, accidents or
toil, when giving identical answers on the 10 points work
ability scale, is compared for the workers in the age group
20–30 years to that of workers in the age group 50–
60 years. In average, 15.6% of the respondents in their 20s
reported reduced work ability in the degree, “yes, indeed”
or “yes, to some extent” when reporting the value “8” on
the workability scale. Whereas 17.8% of the respondents in
their 50s reported “yes, indeed” or “yes, to some extent”,
when at the same time reporting 8 point on the work ability
scale. This was before the scale was reversed meaning 10 is
the highest level of work ability.
Also a general comparison of the answers to the two ques-
tions was performed: when answering “no” to reductions in
work ability due to diseases, accidents or toil, 49.8% answer
10 point on the work ability scale. But when answering “yes,
indeed” or “yes, to some extent” to reductions in work ability
94.8% at the same time add 10 points to their scaled work
ability. Thus, since 50.2% answer to have less than their full
work ability on the 10 point scale, while still answering to
have no reductions due to diseases, accidents or toil, this may
reXect that the four-category measure is not a general work
ability measure, contrary to the 10 point scale, but primary
covers physical attrition, and that reductions in work ability
due to mental strain should be considered too.
Discussion
In this representative sample of the Danish working popula-
tion, self-reported work ability was shown able to predict
both cases of LTSA and ERP during a 5-year follow up.
This Wnding shows that self-reported work ability is a feasi-
ble method to measure peoples’ ability to remain an active
part of the workforce. The estimated eVect from work abil-
ity was substantial with a one point decrease in perceived
work ability, associated with an increased risk of LTSA of
15.1% and a one point decrease in the perceived work abil-
ity followed by an increase in risk of early retirement of
33%. The larger size of the hazard ratio for ERP compared
to LTSA was expected since reduced work ability is
regarded as the main reason for entering ERP.
Table 3 Results when estimating the risk of early retirement pension
(ERP)
aDichotomous variables
ERP (n= 4.743)
Hazard ratio 95% CI P value
Work ability points 2000 1.33 1.20–1.48 <0.0001
Gender (male = 1,
female = 2)a0.64 0.37–1.10 0.1065
Age in years 1.10 1.07–1.12 <0.0001
Monthly wages after
tax in 1,000 kr.
0.80 0.74–0.86 <0.0001
BMI in 2000 1.01 0.96–1.07 0.6505
Children 1.07 0.82–1.40 0.6101
Cohabitinga2.12 1.27–3.53 0.0041
Educational level in years 1.01 0.93–1.11 0.7499
20 cigarettes a day or morea1.88 0.82–4.33 0.1346
Drinks a day 0.85 0.71–1.01 0.0707
Physical activity
level in leisure
0.56 0.41–0.77 0.0003
Int Arch Occup Environ Health (2009) 82:1133–1138 1137
123
Results for demographic and life style factors included in
the analyses are by and large similar to those of other studies.
Women have been shown to be at higher risk of illness than
men (Dahlberg et al. 2007). However, the gender covariate is
non-signiWcant in both analyses presented here. Before wage
is entered, however, the gender covariate was signiWcant,
being a woman increasing the risk of LTSA by 22.3%, This
stresses the relevance of including wage in the analyses since
it reXects other aspects of the living and working conditions
of people than measured in merely life style or single job
characteristics (Schochet and Rangarajan 2004).
Leaving out the work ability measure from the analysis
changed the hazard ratio for age, which then became insig-
niWcant at a 0.05% level. This suggests a general associa-
tion between increasing age and a decreasing number of
spells of sickness absence, when not considering those indi-
viduals with reduced work ability. This may be partly due
to an healthy worker eVect but it still in itself is a promising
result for future interventions in the elderly age group in
regard to sustain their work ability. For the small group of
workers who retire early (1.5%) not controlling for work
ability did not change the negative eVect of age and other
variables as well as the living situation thus may be the
inXuencing factors here. Life style factors like high BMI
and smoking increased the risk of LTSA but not of ERP.
This is in line with the results of (Christensen et al. 2007) in
regard to LTSA. A low level of physical activity in leisure
was associated with an increased risk of ERP.
There are both methodological advantages and draw-
backs using a simple scale of work ability as in this study.
No bias from age was found in regard to the answers when
holding the scale values from the 10 point scale up against a
more speciWc question on whether the respondent felt their
work ability reduced due to disease, accident or toil. But
since this question was placed just before the 10 points
evaluation it may have the impact the younger respondents
somehow compared their reported work ability to this more
general measure and not to more brief experiences of
reductions in their work ability. The question used for com-
parison, however, tend to lack information on reductions in
work ability due to mental strain. Thus, self-assessed ques-
tions on work ability should be used with consideration. In
deciding when to step in and what preventive measures to
take, the WAI index may oVer a more detailed picture since
it places employees in distinct groups with clear indications
of the need for intervention.
In conclusion, work ability seems to be a relevant factor
to include in studies of LTSA and early retirement from the
labour market. To this conclusion can be added the per-
spective that earlier studies have already identiWed a num-
ber of general risk factors for development of reduced work
ability (van der Berg et al. 2008; Gamperiene et al. 2008;
Sell et al. 2008; Lindberg et al. 2006). Self-reported work
ability is a simple and convenient early indicator that can be
used in large surveys of the working population.
Acknowledgments We would like to thank Thomas Lund, Senior
researcher, PhD in medical science, The Danish National Centre for
Social Research, Herluf Trolles Gade 11, DK-1052 Copenhagen, for
his pioneering work on the DWECS and DREAM-database. The study
was supported by a grant from the Danish Working Environment Re-
search Foundatio n and elaborated within the Wnale program for prev en-
tion of reduced work ability.
Appendix 1
Table 4
Table 4 LTSA and ERP without wages or work ability
Hazard
ratio
95% CI P value
LTSA without wages
Work ability points 2000 1.17 1.13–1.21 <0.0001
Gender (male = 1,
female = 2)
1.22 1.10–1.36 0.0002
Age in years 0.99 0.98–0.99 <0.0001
Children 1.00 0.95–1.06 0.8806
Cohabiting 0.94 0.83–1.06 0.3370
Educational level in years 0.91 0.89–0.93 <0.0001
20 cigarettes or more a day 1.37 1.09–1.71 0.0059
BMI in 2000 1.02 1.00–1.03 0.0086
Drinks a day 1.02 0.99–1.06 0.2347
Physical activity level
in leisure
1.01 0.95–1.07 0.7353
LTSA without work ability
Gender (male = 1,
female = 2)
1.08 0.96–1.20 0.1952
Age in years 1.00 0.99–1.00 0.0648
Monthly wages after
tax in 1,000 kr.
0.95 0.93–0.96 <0.0001
Children 1.02 0.97–1.08 0.3489
Cohabiting 0.91 0.81–1.03 0.1586
Educational level in years 0.93 0.91–0.95 <0.0001
20 cigarettes or more a day 1.40 1.12–1.76 0.0031
BMI in 2000 1.02 1.01–1.04 0.0007
Drinks a day 1.03 1.00–1.07 0.0828
Physical activity level
in leisure
0.99 0.94–1.05 0.7986
ERP without wages
Work ability points 2000 1.46 1.33–1.60 <0.0001
Gender (male = 1,
female = 2)
1.14 0.70–1.88 0.5951
Age in years 1.08 1.05–1.11 <0.0001
BMI in 2000 1.01 0.96–1.07 0.6885
Children 0.99 0.75–1.30 0.9399
Cohabiting 2.43 0.47–4.01 0.0005
1138 Int Arch Occup Environ Health (2009) 82:1133–1138
123
References
Burr H, Bjorner JB, Kristensen TS, Tuchsen F, Bach E (2003) Trends
in the Danish work environment in 1990–2000 and their associa-
tions with labor-force changes. Scand J Work Environ Health
29:270–279
Christensen KB, Lund T, Labriola M, Bültmann U, Villadsen E (2007)
The impact of health behaviour on long term sickness absence:
results from DWECS/DREAM. Ind Health 45:348–351
Cox DR (1972) Regression models and life-tables. J Roy Stat Soc
34:187–220
Dahlberg R, Bildt C, Vingard E, Karlqvist L (2007) Educational
background: diVerent processes and consequences on health and
physical and mental exposures among women and men. Work
28:57–66
Fjell Y, Alexanderson K, Karlqvist L, Bildt C (2007) Self-reported
musculoskeletal pain and working conditions among employees
in the Swedish public sector. Work 28:33–46
Gamperiene M, Nygard JF, Sandanger I, Lau B, Bruusgaard D (2008)
Self-reported work ability of Norwegian women in relation to
physical and mental health, and to the work environment. J Occup
Med Toxicol 3:8. doi:10.1186/1745-6673-3-8
Hasselhorn H-M, Tackenberg P, Müller BH (2007) Working Condi-
tions and intent to leave the profession among nursing staV in Eu-
rope. Report No 7: 2003, University of Wuppertal
Hopsu L, Leppanen A, Ranta R, Louhevaara V (2005) Perceived work
ability and individual characteristics as predictors for early exit
from working life in professional cleaners. Int Congr Ser
1280:84–88. doi:10.1016/j.ics.2005.03.062
Ilmarinen J, Tuomi K (1992) Work ability of aging workers 24. Scand
J Work Environ Health 18(Suppl 2):8–10
Kujala V, Tammelin T, Remes J, Vammavaara E, Ek E, Laitinen J
(2006) Work ability index of young employees and their sickness
absence during the following year. Scand J Work Environ Health
32:75–84
Liira J, Matikainen E, Leino-Arjas P, Malmivaara A, Mutanen P,
Rytkonen H, Juntunen J (2000) Work ability of middle-aged Finn-
ish construction workers—a follow-up study in 1991–1995. Int J
Ind Ergon 25:477–481. doi:10.1016/S0169-8141(99)00032-3
Lindberg P, Vingard E, Josephson M, Alfredsson L (2006) Retaining
the ability to work—associated factors at work. Eur J Public
Health 16:470–475. doi:10.1093/eurpub/cki190
Lund T, Labriola M, Christensen KB, Bultmann U, Villadsen E, Burr
H (2005) Psychosocial work environment exposures as risk
factors for long term sickness absence among Danish employees:
results from DWECS/DREAM. JOEM 47:1141–1147
Lund T, Labriola M, Christensen KB, Bultmann U, Villadsen E (2006)
Physical work environment risk factors for long term sickness
absence: prospective Wndings among a cohort of 5357 employees in
Denmark. BMJ 332:449–452. doi:10.1136/bmj.38731.622975.3A
Lund T, Labriola M, Villadsen E (2007) Who is at risk for long-term
sickness absence? A prospective cohort study of Danish employ-
ees. Work 28:225–230
MacKinnon DP, Fairchild AJ, Fritz MS (2007) Mediation analysis.
Annu Rev Psychol 58:593–614. doi:10.1146/annurev.psych.
58.110405.085542
Melchior M, Niedhammer I, Berkman LF, Goldberg M (2003) Do
psychosocial work factors and social relations exert independent
eVects on sickness absence? A six year prospective study of the
GAZEL cohort. J Epidemiol Community Health 57:285–293.
doi:10.1136/jech.57.4.285
Nurminen E, Malmivaara A, Ilmarinen J, Ylostalo P, Mutanen P,
Ahonen G, Aro T (2002) EVectiveness of a worksite exercise pro-
gram with respect to perceived work ability and sick leaves among
women with physical work. Scand J Work Environ Health 28:85–93
Nygard CH, Arola H, Siukola A, Savinainen M, Luukkaala T,
Taskinen H, Virtanen P (2005) Perceived work ability and certi-
Wed sickness absence among workers in a food industry. Int Congr
Ser 1280:296–300. doi:10.1016/j.ics.2005.02.062
Reiso H, Nygard JF, Brage S, Gulbrandsen P, Tellnes G (2001) Work
ability and duration of certiWed sickness absence. Scand J Public
Health 29:218–225. doi:10.1177/14034948010290031201
Salonen P, Arola H, Nygard CH, Huhtala H, Koivisto AM (2003)
Factors associated with premature departure from working life
among ageing food industry employees. Occup Med (Lond)
53:65–68. doi:10.1093/occmed/kqg012
Schochet P, Rangarajan A (2004) Characteristics of low-wage workers
and their labor market experiences: evidence from the mid- to late
1990s. Mathematica Policy Research, Princeton (1), 66
Sell L, Faber A, Søgaard K (2008) Impacts from occupational risk fac-
tors on self reported reduced work ability among Danish wages
earners. In: Masaharu K (eds) Promotion of work ability towards
productive aging: selected papers of the 3rd international sympo-
sium on work ability, Hanoi, 22–24 October. Taylor and Francis,
London
Tuomi K, Ilmarinen JE, Jahkola A, Katajarinne L, Tulkki A (1998)
Work ability index, 2nd edn. Finnish Institute of Occupational
Health, Helsinki
UN Department of Economic and Social AVairs PD2 (2007) World
population prospects (revised)
van der Berg TIJ, Alavinia SM, Bredt FJ, Lindeboom D, Elders LAM,
Burhof A (2008) The inXuence of psychosocial factors at work
and life style on health and work ability among professional
workers. Int Arch Occup Environ Health 81:1029–1036.
doi:10.1007/s00420-007-0296-7
Vingard E, Lindberg P, Josephson M, Voss M, Heijbel B, Alfredsson
L, Stark S, Nygren A (2005) Long-term sick-listing among wom-
en in the public sector and its associations with age, social situa-
tion, lifestyle, and work factors: a three-year follow-up study.
Scand J Public Health 33:370–375. doi:10.1080/140349405
10005860
Table 4 continued
Hazard
ratio
95% CI P value
Educational level in years 0.94 0.87–1.02 0.1440
Cigarettes a day 1.92 0.84–4.37 0.1194
Drinks a day 0.87 0.72–1.05 0.1518
Physical activity level
in leisure
0.59 0.43–0.80 0.0007
ERP without work ability
Gender (male = 1,
female = 2)
0.64 0.37–1.09 0.0974
Age in years 1.11 1.08–1.14 <0.0001
Monthly wages after tax
in 1,000 kr.
0.76 0.70–0.82 <0.0001
BMI in 2000 1.03 0.98–1.09 0.2459
Children 1.07 0.82–1.39 0.6189
Cohabiting 2.25 1.35–3.73 0.0017
Educational level in years 1.00 0.91–1.09 0.9807
Cigarettes a day 1.92 0.83–4.43 0.1281
Drinks a day 0.85 0.70–1.03 0.1046
Physical activity level
in leisure
0.55 0.40–0.75 0.0001