Psychosocial work environment factors and weight change: a prospective study among Danish health care workers.

0.21 MB·English

by  Gram Quist, Helle

GramQuistetal.BMCPublicHealth2013,13:43
http://www.biomedcentral.com/1471-2458/13/43
RESEARCH ARTICLE Open Access
Psychosocial work environment factors and
weight change: a prospective study among
Danish health care workers
Helle Gram Quist1*, Ulla Christensen2, Karl Bang Christensen3, Birgit Aust1, Vilhelm Borg1 and Jakob B Bjorner1,2,4
Abstract
Background: Lifestyle variables may serve as importantintermediate factors between psychosocial work
environment and health outcomes. Previous studies, focussing on work stressmodels have shown mixedand weak
results in relation to weight change. This study aims to investigate psychosocial factors outside the classical work
stress models as potential predictorsof change in body mass index (BMI) in a population of health care workers.
Methods: A cohort study, withthreeyears follow-up, was conducted among Danish health care workers (3982
women and 152men). Logistic regression analysesexamined change inBMI (more than+/− 2kg/m2) as predicted
bybaseline psychosocialwork factors (work pace,workload, quality ofleadership,influence atwork, meaning of
work, predictability, commitment, role clarity, and role conflicts) and five covariates (age, cohabitation, physical work
demands, typeof work position and seniority).
Results: Among women, high role conflicts predicted weight gain, whilehigh role clarity predicted both weight
gain and weight loss. Living alone also predicted weight gain among women, whileolder age decreased the odds
ofweight gain. High leadership quality predicted weight loss among men. Associations were generally weak, with
theexceptionof quality of leadership, age, and cohabitation.
Conclusion: This study ofa single occupational group suggesteda few new risk factors for weight change outside
thetraditionalwork stress models.
Keywords: Body mass index, Weightgain, Weight loss, Work stress, Psychosocial workfactors
Background matter of genetic and socioeconomic factors [5,6].
Obesity is a well-known risk factor for mortality and Evidence shows that psychosocial work factors affect our
morbidity, including cardiovascular disease, hypertension, health [7-13] and health behaviour, such as physical
stroke,typeIIdiabetesandsometypesofcancer[1-3].The activity, drinking, smoking and dietary habits [14-19].
prevalence of obesity has grown worldwide, reaching These health behaviours may be intermediate factors in
epidemic proportions. Previous research has found that the relationship between psychosocial work environment
weightchange,bothlossandgain,carriesariskofmortality andhealthrelatedoutcomes.Longitudinalresearchstud-
independentlyoftheinitialweightlevel;andthattheriskof ies on working conditions and weight change are scarce
mortalityfromweightchangeishigherthantheriskfroma and have mainly looked at traditional work stress
stable weight [4]. Maintaining a stable weight seems models, particularly the job-strain model [12] and the
favourablewhenaddressingtheriskofmortality. effort-reward imbalance model [20]. In both models,
Body weight is not just a simple matter of energy it is assumed that work stress can alter our health
balance (calorie intake and physical activity), but also a behaviours, causing - for instance - weight change.
Overall, results from previous studies are inconsistent.
Where some studies have found full support for a
*Correspondence:[email protected]
positive association between poor psychosocial working
1NationalResearchCentrefortheWorkingEnvironment(NRCWE),Lersoe
Parkalle105,2100,Copenhagen,Denmark conditions and weight gain [8,21-23], others have found
Fulllistofauthorinformationisavailableattheendofthearticle
©2013GramQuistetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsofthe
CreativeCommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,
distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.
GramQuistetal.BMCPublicHealth2013,13:43 Page2of9
http://www.biomedcentral.com/1471-2458/13/43
noconsistentsupport[24-26],abi-directionalrelationship risk of both weight gain and weight loss as individual
[27,28], or that poor psychosocial working conditions are responsestostressmayvary.
associated with weight loss [29-31]. Also, research
suggests that stress can cause some people to eat more,
while others react byeatingless[32,33].Epidemiological Methods
researchers have also tied body weight/weight change to Studydesignandstudysample
other work and sociocultural factors, such as shift We used data from a prospective cohort study of
work, working overtime, education, income and marital employees in the eldercare sector in Denmark. The study
status [34-39]. Again results are mixed and vary aims at examining the health and work environment of
between genders. health care workers employed in Danish municipalities
As the job-strain model [12] and the effort-reward (36 out of 65 invited municipalities participated).
imbalance model [20] have shown inconsistent results, it Baseline questionnaire data was collected from fall
is worth broadening the area of psychosocial work envi- 2004 to spring 2005 using mailed questionnaires.
ronment, as it is possible that other psychosocial work These were distributed to 12,746 employees at base-
factors could have an effect on weight change. This was line – of which 9,949 participated (78%). A follow-up
done in a recent study by Berset and colleagues, who was conducted in 2006, preceding a second follow-up
examined the role of social stressors on the relationship in 2008. Participation in the follow-up rounds was
betweenweightgainandworkstress[40].Theyfoundthat 64% and 63%, respectively. Maximizing the follow-up
social stressors, such as tension between colleagues and period to three years, this study only included
the use of reprimands, were predictive of BMI at respondents who participated at baseline and in the
follow-up, while only finding limited support for the second follow-up round (3982 women and 153 men).
role of work stress. This adds to the idea that other Table 1 presents descriptive characteristics of the
aspects of the psychosocial work environment might participants at baseline. An analysis of difference between
be of bigger importance than previously assumed. the study population (those responding to both baseline
Consequently, this explorative study uses a broader and follow-up) and those only examined at baseline,
mapping of the psychosocial work environment, based on showednosignificantdifferenceswithregardstoage,body
the Copenhagen Psychosocial Questionnaire (COPSOQ) massindexandleisure-timephysicalactivity.However,the
[41]. The COPSOQ is theoretically founded but not study population had significant longer tenure (mean 8.9
restricted to a single theory [42] and is developed on the vs. 8.0 years, p<.0001), lower physical demands at work
basis of existing questionnaires, such as the Setterlind (mean18.6 vs. 19.6, p<.0001)and had alower proportion
Stress Profile [43], the Whitehall II questionnaire [44], the of smokers (34% vs 36%, p=0.0151). With regards to the
Job-ContentQuestionnaire[45]andShortForm-36[46]. psychosocial work environment, we found that the
In addition to the dimensions from the job-strain and study population reported higher predictability
effort-reward imbalance models, the COPSOQ includes (mean 58 vs 56, p<.0001), higher influence (mean
other dimensions from the psychosocial area, such as 47 vs 45, p<.0001), higher leadership quality (mean
emotional demands, cognitive demands, meaning of 59 vs. 56, p<.0001), higher involvement (mean 60
work, quality of leadership, predictability, role conflicts, vs. 58, p<.0001), higher meaning at work (mean 79
social support and offensive behaviours. In a recent vs. 77, p=0.0006), but lower role conflicts (mean 41
study, high emotional demands and low meaning of vs. 42, p=0.0133). Thus, it seems that the study
work, was shown to predict ill mental health beyond the population experience a better psychosocial and
job strain and effort-reward imbalance models [47]. physical work environment that the sample only
Further,roleconflicts,highemotionaldemands,lowquality examined at baseline. However, the differences in
of leadership and the demands for hiding emotions, have work environment appear to be small.
been found to predict sickness absence in Danish The study was approved by the Danish Data Protection
populations [48,49]. Using dimensions from the COPSOQ, Agency and followed the regulations for data storage
the aim of the present study is to examine the relationship and protection. Questionnaire research in Denmark
between a broad range of psychosocial work factors and does not require approval by ethic committees and
change in BMI. We see weight change as an indicator of thus approval was not obtained. Before completing
lifestyleandhealthbehaviour.WeexaminedchangeinBMI the questionnaire, participants were informed about
bi-directionally; both as an increase and decrease. We the study and it was made clear that participation
hypothesize that an unfavourable psychosocial work envir- was voluntary. Participants returned the completed
onment (e.g. low meaning of work, low predictability etc.) questionnaires directly to the research group and
is associated with change in BMI. Our analyses take into confidentiality was maintained by using numbers to
account that work environment stressors can increase the identify participants.
GramQuistetal.BMCPublicHealth2013,13:43 Page3of9
http://www.biomedcentral.com/1471-2458/13/43
Table1Baselinedescriptivestatisticsofthestudypopulation
Men Women
Variable N % Mean S.D N % Mean S.D
Baseline2004/2005
Age 153 46 8.25 3982 45 8.45
Smokestatus
Smoker 32 20.92 . . 1389 35.18 . .
Non-smokera 121 79.08 . . 2559 64.82 . .
BMI(kg/m2) 151 25.97 3.30 3804 24.89 4.34
Cohabitation
Livingwithapartnerorspouse 122 80.79 . . 3245 82.38 . .
Livingalone/other 29 19.21 . . 694 17.62 . .
WorkHours
Dayschedule 122 79.74 . . 2658 66.87 . .
Fixedeveningshift 17 11.11 . . 581 14.62 . .
Fixednightshift 5 3.27 . . 181 4.55 . .
Alternatingday/eveningshift 8 5.23 . . 441 11.09 . .
Alternatingevening/nightshift 0 0.00 . . 27 0.68 . .
Alternatingday/evening/nightshift 1 0.65 . . 87 2.19 . .
Typeofworkposition
Managers 45 29.41 . . 267 6.71 . .
Healthcareworkers 73 47.71 . . 3378 84.83 . .
Other(eg.janitors,secretaries) 35 22.88 . . 337 8.46 . .
Follow-up2008
BMI(kg/m2) 150 26.21 3.24 3776 25.37 4.62
ChangeinBMI
Unchangedweight 125 83.89 . . 2853 77.72 . .
BMIgainof2kg/m2 16 10.74 . . 583 15.88 . .
BMIlossof2kg/m2 8 5.37 . . 235 6.40 . .
aincludingprevioussmokers.
Measurements influence on tasks, amount of work and choice of co-
Psychosocialworkcharacteristics workers. Meaning of work was measured by a three-item
The psychosocial work environment was examined with scale about the meaningfulness and importance of work,
items and scales derived from the 1st version of the work motivation and involvement. Commitment to the
Copenhagen Psychosocial Questionnaire (COPSOQ) workplace was measured by a five-item scale concerning
[41,50]. We measured the following psychosocial work personal dedication to the workplace. Predictability was
environment factors: quantitative demands, quality of measured by a two-item scale about information related
leadership, influence at work, meaning of work, commit- to organizational changes and information required to
ment, predictability, role clarity and role conflicts. These carry out ones job well. Role clarity was measured by a
are recognized in the literature as important psychosocial three-item scale regarding expectations, objective and
work exposures and have been used in other studies responsibilities at work. Finally, role conflicts were
[7,10,47,49,51,52]. Quantitative demands were measured measured by a four-item scale about the experience and
byasingle-itemregardingworkpaceandatwo-itemscale acceptance of tasks, contradictory demands and unneces-
about workload. Quality of leadership was measured by a sarytasks.Allpsychosocialscaleswerere-codedfrom0to
four-item scale about the managements’ ability to plan 100 points, with low scores indicating low levels of the
work, solve conflicts and ensure well-being and develop- measured dimension. Depending on the scale, a low
ment opportunities for their employees. Influence at work score can be positive (e.g.lowroleconflicts)ornegative
was measured by four questions related to personal (e.g.lowleadershipquality).
GramQuistetal.BMCPublicHealth2013,13:43 Page4of9
http://www.biomedcentral.com/1471-2458/13/43
Workinghours factors and weight change. Specifically, change in BMI
As working hours are related to psychosocial work was utilized as the dependent variable and psychosocial
environment, e.g. demands and influence [53], we also work factors as predictors. The bi-directional approach
analyzed working hours. Working hours were divided allowed us to determine whether psychosocial work
into two categories: day schedule (fixed dayshift) and factors can lead to increased BMI, decreased BMI or
non-day schedule (fixed evening shift, fixed night both. We conducted both unadjusted and adjusted
shift, alternating day/evening shifts, alternating even- regression analyses. In the adjusted model, a backward
ing/night shift or alternating day/evening/night shift). elimination approach was utilized, starting with all
We collapsed all the non-day schedule groups into variables included to mutually adjust for each predictor
one group, as some of them only had very few variable. The variables were tested for statistical signifi-
participants (see Table 1). cance each at the time, deleting those that were not
significant. Variables with a significance level greater than
Covariates the criterion level of 0.05 were removed from the model
The covariates in this study were age, cohabitation, type of (the least significant first). All psychosocial variables, in
work position, seniority and physical work demands. addition to seniority and physical work demands, were
Cohabitation was dichotomized as living alone or living included as continuous variables in the statistical analyses.
with a spouse or partner. Type of work position included The psychosocial work environment factors were all
managers, health care workers, or others (e.g. janitors, assessed as continuous variables (originally scores from 0
cleaners, secretaries). Seniority specified the number of to 100). However, we changed scoring to 10-points scales
years respondents had worked at their current workplace. (0 to 10), so that the OR represents a 10-point increase in
Finally, physical work demands were calculated as a single the psychosocial variable. Also, age and physical workload
mean score based on 15 questions regarding lifting, werechangedtorepresenta10-yearand10-pointincrease,
position of the arms, and work postures (for example, respectively. All analyses were conducted separately for
“How often does your work require that you kneel down; men and women. The statistical analyses were performed
ononeorbothknees?”or“Howoftendoyouworkwitha withSASProcLogistics,version9.2(SASInstitute).
rotated upper body?”) [54]. Respondents were asked how Since we collapsed all non-day schedule workers into
often they were exposed to these positions and lifts one group, we tested whether any loss of information
(response categories ranged from “never” to “very often”). was caused by this day/non-day dichotomization. Fur-
The higher the mean score, the more demanding physical thermore, in order to evaluate possible loss of informa-
workrequirements. tion due to the categorization of BMI change, we also
performed additional analyses using BMI as a continu-
Outcome ous variable (using SAS Proc Mixed with logarithmic
Body mass index was calculated based on the transformation of BMI to achieve model fit). In addition,
respondents’ self-reported information on height and wealsoconductedsensitivityanalyses;wetestedwhether
weight (weight in kilograms divided by height in meters the association between change in BMI and the psycho-
squared). We measured change in BMI (gain or loss) of social work factors were similar regardless of the work
more than 2 kg/m2. Change in BMI was separated into factors being treated as continuous variable or categor-
the following three groups: (1) unchanged weight, (2) BMI ical variables (Proc Logistic, SAS, version 9.2). Finally,
lossofmorethan2kg/m2,and(3)BMIgainofmorethan we checked for multicollinearity among the predictor
2 kg/m2. Unchanged weight was the reference group. To variables(Proc Reg,SAS,version 9.2).
ourknowledge,thereisnocommonlyacceptedstandardas
to what constitutes a meaningful weight change over time. Results
Research indicates that it is desirable to maintain a stable Table 1 presents sample characteristics for women
weight (in terms of mortality risk) as almost any level of (n=3982) and men (n=153). On average, the women
weightchangeisassociatedwithincreasedriskofmortality wereyoungerandhadslightlylowermeanBMIatbaseline
compared to maintaining a stable weight [4]. To achieve a than men. In both sexes, BMI increased over the 3-year
sufficient group size for a robust analysis, we chose a cut- follow-up period (0.25 kg/m2 for men and 0.48 kg/m2 for
offpoint at+/−2kg/m2asalmost28%oftherespondents women). More women than men were current smokers
hadgained/lostmorethan2kg/m2,whileonly10%gained/ (35% and 21%, respectively) and more than 8 out of 10
lostmorethan5kg/m2. lived with a spouse or a partner (81% and 82%, for men
andwomenrespectively).
Statisticalanalysis A total of 3647 women and 136 men were included in
Using logistic regression for nominal categorical data, the regression analyses as 335 women and 17 men were
we assessed the associations between psychosocial work excluded due to missing values on the response or
GramQuistetal.BMCPublicHealth2013,13:43 Page5of9
http://www.biomedcentral.com/1471-2458/13/43
explanatory variables. In Table 2 the unadjusted and increased the odds of weight loss. For women, high role
adjusted estimates are reported for the association conflicts increased the odds of weight gain (OR=1.13;
between each predictor and change in BMI (odds ratio 95% CI: 1.06-1.21). High role clarity increased the odds
and95%confidenceintervalsarepresented).Amongmen, for both weight gain (OR=1.09; 95% CI: 1.02-1.17) and
high leadership quality (OR=1.55; 95% CI: 1.02-2.36) weightloss(OR=1.14;95%CI:1.03-1.27)amongwomen.
Table2LogisticsregressionwithBMIchangeasdependentvariable
Women Men
Unadjusted Mutuallyadjusteda Unadjusted Mutuallyadjusteda
Variable Response OR 95%CI OR 95%CI OR 95%CI OR 95%CI
Qualityofleadership BMIgain 0.98 0.94-1.02 . . 1.27 0.96-1.69 1.29 0.96-1.73
BMIloss 1.00 0.94-1.07 . . 1.48 1.00-2.19 1.55 1.02-2.36
Influenceatwork BMIgain 1.01 0.96-1.05 . . 1.17 0.90-1.52 . .
BMIloss 0.99 0.93-1.06 . . 1.14 0.77-1.67 . .
Meaningofwork BMIgain 1.02 0.95-1.09 . . 1.10 0.75-1.62 . .
BMIloss 1.07 0.97-1.19 . . 1.95 1.04-3.66 . .
Workload BMIgain 0.99 0.95-1.04 . . 1.03 0.79-1.34 . .
BMIloss 0.97 0.91-1.03 . . 0.75 0.51-1.12 . .
Workpace BMIgain 1.01 0.97-1.06 . . 1.19 0.89-1.59 . .
BMIloss 0.94 0.87-1.01 . . 0.80 0.53-1.20 . .
Commitment BMIgain 0.97 0.92-1.02 . . 1.07 0.78-1.45 . .
BMIloss 0.95 0.88-1.03 . . 1.63 0.98-2.71 . .
Roleclarity BMIgain 1.03 0.97-1.10 1.09 1.02-1.17 0.99 0.71-1.38 . .
BMIloss 1.11 1.00-1.22 1.14 1.03-1.27 1.97 1.03-3.79 . .
Roleconflicts BMIgain 1.13 1.06-1.19 1.13 1.06-1.21 1.00 0.74-1.35 . .
BMIloss 1.02 0.93-1.11 1.04 0.94-1.14 0.84 0.54-1.31 . .
Predictability BMIgain 0.97 0.93-1.02 . . 1.02 0.76-1.38 . .
BMIloss 1.04 0.97-1.12 . . 1.37 0.86-2.17 . .
Workinghours BMIgain 1.04 0.86-1.26 . . 0.54 0.12-2.55 . .
BMIloss 1.14 0.94-1.38 . . 1.27 0.24-6.66 . .
Covariates
Age BMIgain 0.71 0.64-0.78 0.71 0.64-0.79 0.92 0.49-1.71 . .
BMIloss 0.85 0.73-0.99 0.85 0.72-1.01 0.70 0.31-1.59 . .
Cohabitation BMIgain 1.29 1.04-1.62 1.33 1.06-1.68 1.53 0.45-5.19 . .
BMIloss 0.86 0.59-1.25 0.79 0.53-1.17 0.66 0.08-5.60 . .
Workposition:
Managers BMIgain 0.76 0.52-1.11 . . 3.33 1.03-10.75 . .
BMIloss 0.78 0.44-1.36 . . 0.74 0.14-4.02 . .
Othersb BMIgain 0.92 0.67-1.28 . . 0.79 0.14-4.29 . .
BMIloss 0.62 0.35-1.10 . . 0.39 0.04-3.52 . .
Seniority BMIgain 0.98 0.97-1.01 . . 1.01 0.95-1.08 . .
BMIloss 1.00 0.99-1.02 . . 0.92 0.79-1.07 . .
Physicaldemands BMIgain 1.02 0.93-1.12 . . 0.97 0.62-1.52 . .
BMIloss 0.92 0.80-1.06 . . 0.68 0.34-1.36 . .
aSignificantpredictorsafterbackwardselimination.
bE.g.janitors,cleaners,secretaries.
[OR=oddsratio;95%CI=confidenceintervals].
Significantresultsarepresentedinboldface.
GramQuistetal.BMCPublicHealth2013,13:43 Page6of9
http://www.biomedcentral.com/1471-2458/13/43
In addition to the psychosocial work factors, we also Of the included covariates, we found age and cohabit-
found that living alone (OR=1.33; 95% CI: 1.06-1.68) ation (among women) were associated with weight
increasedtheoddsforweightgain,whileolderagesignifi- change. Specifically, living alone increased the odds for
cantlydecreasedtheoddsforweightgain(OR=0.71;95% weight gain, while older age decreased the odds for
CI: 0.64-0.79). Except for age, cohabitation, and quality of weight gain. Smoking and leisure time physical activity
leadershipallassociationswereweak. were not included as covariates, as these can be thought
Resultsfromtheadditionalanalyses,whereweaddressed of as intermediate variables in the relationship between
BMI as a continuous variable, showed weaker results. For psychosocial work environment and health outcomes.
women, weight change was associated with age and For instance, work hours may influence ones possibility
cohabitation, but not with any psychosocial work factors. for participating in leisure time physical activity, which
Among men, weight change was not associated with any inturn,mayimpact weightchange.
psychosocial work factor either (results not shown). The mixed findings of this study contribute to the
However, we would argue that using BMI as a continuous existing body of literature demonstrating inconsistent
variable could limit the possibility of finding bi-directional results in relation to weight change and psychosocial
effects of work environment. Unsuitable weight change work factors. Reviews by Overgaard and colleagues [25]
canoccurinseparatedirectionsdependingontheindivid- andSiegristandRödel[55],foundonlymodestsupportfor
ual; i.e. thesame workfactor can increase the riskof both a consistent relationship between work stress and
weightgainandweightloss. bodyweight/body mass index, and Wardle and colleagues
Asitispossiblethatthedichotomizationofworkhours [56]concludedinameta-analysisthatworkstressonlyhas
could hide important differences between the groups, we very small effect as risk factor for weight gain. Similarly, a
also ran the logistic regression analyses with the original largecohortstudyfoundthattherelationshipbetweenBMI
non-collapsed categories. However, regardless of this, and work stress remains unclear (Nyberg et al., in press,
work hours were still not predictive of weight change 2011). The inconsistent findings can be a result of
(resultsnotshown). differences in the exposure measurement, design, or
Results from the sensitivity analyses, showed that methods,butitmayalsobecausedbyafocusonanoverall
whentreatedascategoricalvariable(three levels),quality association.Generally,weightgainhasbeeninvestigatedas
of leadership was not a significant predictor for weight a function of work related stress. Two prospective studies
change among men. However, role clarity and role have investigated a bi-directional relationship, and both
conflicts remained significant predictors, as did age and found support for the bi-directional effects of work
cohabitation, among women. Finally, we checked for stress on weight [27,28]. We found some support for
multicollinearitybyusingtheVIFoption(VarianceInflation bi-directional effects of psychosocial work factors on
Factor) in the Proc Reg analysis (SAS, version 9.2). weight change, in particularly with role clarity, which
Multicollinearity, i.e. that some of the predictor variables was associated with both BMI gain and BMI loss.
(two or more) are highly correlated with each other, can Our results indicate that factors outside the classical
lead to imprecise parameter estimates. The test indicated work stress have only limited influence on weight
thatmulticollinearitywasnotofmajorconcern;thehighest change, some even in an unexpected direction.
VIF value among women was 1.87 (predictability) and Living alone was predictive of weight gain for
amongmenitwas2.08(commitment). women, but not for men. Marital status has received
some attention as a factor influencing personal health,
Discussion and generally, married individuals tend to weigh more
The purpose of this paper was to investigate psychosocial than non-married [57,58]. However, some research
factors outside the classical work stress models in an has neglected to distinguish between cohabitating
attempttodeterminewhetherthesepredictweightchange. individuals and true singlehood. Making this distinc-
We hypothesized that psychosocial work factors would be tion, Averett and colleagues found that cohabitation
associated with change in BMI when these represented an (and marriage) were associated with increased BMI
unfavourable work environment. Our hypothesis received [59]. Our finding did not support this; the odds of
only limited support. Among men, high leadership quality weight gain were highest among single women. Also,
wasassociatedwithweightloss–againstourexpectations. older age decreased the odds for weight gain.
Among women, we also found an unexpected association Thisstudyhadanumberofstrengths:thebi-directional
between weight change and high role clarity. In support of approach to the analysis of weight change and the use of
our hypotheses, we found that weight change (gain) was psychosocial work factors outside the traditional work
predicted by high role conflicts. Consequently, we cannot stress models. First, the bi-directional strategy allows for
conclude that an unfavourable psychosocial work the investigation of weight change in both directions; an
environment is associated with weight change per se. approach only applied in few previous studies. Second,
GramQuistetal.BMCPublicHealth2013,13:43 Page7of9
http://www.biomedcentral.com/1471-2458/13/43
this study adds to prior research by using a broader psychosocial work factors seem to have some effect
mapping of the psychosocial work environment. on weight change, the associations were generally
Workload, work pace and influence at work, can be weak. Furthermore, some of the associations were in
considered aspects of the job-strain model. Similarly, an unexpected direction. The lack of strong findings may
commitment can be considered an aspect of the implythattheassociationbetweenweightchangeandwork
effort-reward imbalance model. However, we addressed environmentriskfactorssimplyistoovague.Thissuggests
eachoftheseasindependentfactors. that the risk factor model may not be the right theoretical
The assessment within a single occupational group framework for assessing the influence of the workplace on
(health care workers) provides both advantages and weight. Each workplace also represents a social setting
disadvantages, since it reduces the variation in the where the employee interacts with co-workers (and poten-
traditionalpsychosocialfactors,astheworkersexperience tiallycustomers,clientsetc.)thatcansetexamples,provide
similarconditions.Thetraditionaldomainsofquantitative encouragement or lack of encouragement for pursuing a
demands and influence are work factors related more to healthy lifestyle. Thus, the effect of workplace on personal
job type than to place of work. On the other hand, lifestyle might transcend the risk factors evaluated in the
domains like leadership quality and predictability are currentstudy.Consequently,itisrelevanttoinvestigatethe
muchmorerelatedtotheworkplacethantojobtype.The effect of workplace and to examine potential group effects
design of the current study is optimal for detecting withinworkplacesinfuturestudies.
the latter type of effects.
Generally,ourresultsshouldbeinterpretedinthelightof
Abbreviations
the following limitations. First, our data were female- BMI:Bodymassindex;COPSOQ:CopenhagenPsychosocialQuestionnaire;
dominated;weonlyhad153maleparticipants,which limit OR:Oddsratio;CI:Confidenceinterval.
the strength of the analyses and the generalizability.
Caution must be taken when interpreting the results for
Competinginterests
men; the number of male respondents who experienced Theauthorsdeclarethattheyhavenocompetinginterests.
weight change was only 24. Although the attrition rate in
our study was somewhat high, and potentially could cause Authors’contributions
lossofpower,wewouldarguethatbiasfromattritionisnot HGQhasdevelopedtheideaforthisarticleandwasresponsibleforthe
a major concern in this study as we found no differences developmentofthedesign,dataprocessingandthestatisticsanalyses.
Furthermore,HGQhasinterpretedtheresults,writtenthemanuscriptand
on BMI when comparing the study population with those
prepareditforpublication.UChascontributedsubstantiallytothe
who had only responded at baseline. Another potential developmentofthedesign,interpretationoftheresultsandprovidedcritical
limitation of this study is the fact that we conducted revisionstothemanuscript.KBChascontributedtothedevelopmentofthe
designandthestatisticalanalyses.BAhasprovidedcriticalrevisionstothe
multiple testing. Multiple testing can be a problem as it
manuscriptandtothedesignofthestudy.VBhasprovidedcriticalrevisions
increases therisk ofmass significance. Thus, some caution totheinterpretationoftheresultsandthedesignofthestudy.JBBhas
must be taken when interpreting the significant results. supervisedtheentireprocessandsignificantlycontributedtothedesign,
statisticalanalyses,interpretationoftheresultsandmadecriticalrevisionsto
Furthermore, we used a relatively short follow-up period
themanuscript.Allauthorshavereadandapprovedthemanuscriptinits
(three years). A recent study among shift workers [60] finalform.
found that only longer term shift work (10+ years) signifi-
cantly increased the risk of obesity. This suggests that a
Acknowledgements
three year follow-up period, might be insufficient to draw ThecohortstudywasfinancedbytheDanishGovernmentthroughathree-
firm conclusions. It might be that it takes longer for the yeargranttotheFOR-SOSUprogramattheNationalResearchCentreforthe
WorkingEnvironment.Thewritingofthismanuscriptwasfundedbyagrant
psychosocial work environment to have an impact on
fromTheDanishWorkingEnvironmentResearchFund.Noneofthefunding
weightchange,orthatotherfactors(insideandoutsidethe sourcestookpartinthestudydesign,datacollection,interpretationofthe
work environment) play a larger role in this relationship. results,writingofthemanuscript,ordecisionsregardingpublicationofthe
manuscript.TheauthorswishtothankdatamanagerIsabellaGomes
Finally, we relied on self-reported data on weight and
Carneiro,PhD,fromtheDanishNationalResearchCentrefortheWorking
height,whichcancausebiasasheightandweightareoften Environment,whohashelpedwithdataprocessing.
over- and underestimated [61,62]. However, since we
Authordetails
studied weight change, the bias is probably less than
1NationalResearchCentrefortheWorkingEnvironment(NRCWE),Lersoe
in a cross-sectional analysis. Parkalle105,2100,Copenhagen,Denmark.2DepartmentofPublicHealth,
SectionofSocialMedicine,UniversityofCopenhagen,OesterFarigmagsgade
5,1014,Copenhagen,Denmark.3DepartmentofPublicHealth,Sectionof
Conclusion
Biostatistics,UniversityofCopenhagen,OesterFarigmagsgade5,1014,
In conclusion, including alternative psychosocial work Copenhagen,Denmark.4QualityMetric/OptumInsight,24AlbionRoad,
environment factors has added some new information Lincoln,RI02865,USA.
about the relationship between work environment
Received:12April2012Accepted:15January2013
and weight change. Although some of these additional Published:17January2013
GramQuistetal.BMCPublicHealth2013,13:43 Page8of9
http://www.biomedcentral.com/1471-2458/13/43
References 23. NetterstromB,KristensenTS,DamsgaardMT,OlsenO,SjolA:Jobstrainand
1. GelberRP,KurthT,MansonJE,BuringJE,GazianoJM:Bodymassindexand cardiovascularrisk-factors-across-sectionalstudyofemployeddanish
mortalityinmen:evaluatingtheshapeoftheassociation.IntJObes menandwomen.BrJIndMed1991,48(10):684–689.
2007,31(8):1240–1247. 24. BrissonC,LarocqueB,MoisanJ,VezinaM,DagenaisGR:Psychosocial
2. PerryIJ,WannametheeSG,WalkerMK,ThomsonAG,WhincupPH,Shaper factorsatwork,smoking,sedentarybehavior,andbodymassindex:a
AG:Prospective-Studyof Risk-FactorsforDevelopmentofNon- prevalencestudyamong6995whitecollarworkers.JOccupEnvironMed
Insulin-DependentDiabetesinMiddle-AgedBritishMen.BMJ1995, 2000,42(1):40–46.
310(6979):560–564. 25. OvergaardD,GyntelbergF,HeitmannBL:Psychologicalworkloadand
3. MurphyNF,MacIntyreK,StewartS,HartCL,HoleD,McMurrayJJV:Long- bodyweight:isthereanassociation?Areviewoftheliterature.
termcardiovascularconsequencesofobesity:20-yearfollow-upofmore OccupationalMedicine-Oxford2004,54(1):35–41.
than15000middle-agedmenandwomen(theRenfrew-Paisleystudy). 26. NybergST,HeikkiläK,FranssonEI,AlfredssonL,DeBacquerD,BjornerJB,
EurHeartJ2006,27(1):96–106. BonenfantS,BorritzM,BurrH,CasiniA,ClaysE,DraganoN,ErbelR,
4. MikkelsenKL,HeitmannBL,KeidingN,SorensenTIA:Independenteffects GeuskensGA,GoldbergM,HooftmanWE,HoutmanI,JöckelKH,KittelF,
ofstableandchangingbodyweightontotalmortality.Epidemiology KnutssonA,KoskenvuoM,LeineweberC,LunauT,MadsenIEH,
1999,10(6):671–678. MagnussonHansonLL,MarmotMG,NielsenML,NordinM,OksanenT,
5. LahelmaE,RahkonenO,BergMA,HelakorpiS,PrattalaR,PuskaP,UutelaA: PenttiJ,etal:Jobstraininrelationtobodymassindex:pooledanalysis
ChangesinhealthstatusandhealthbehavioramongFinnishadults of160000adultsfrom13cohortstudies.JInternMed2012,
1978–1993.ScandJWorkEnvironHealth1997,23:85–90. 272(1):65–73.
6. LaaksonenM,PrattalaR,HelasojaV,UutelaA,LahelmaE:Incomeand 27. HannerzH,AlbertsenK,NielsenML,TuchsenF,BurrH:Occupational
healthbehaviours.EvidencefrommonitoringsurveysamongFinnish factorsand5-yearweightchangeamongmeninaDanishnational
adults.JEpidemiolCommunityHealth2003,57(9):711–717. cohort.HealthPsychol2004,23(3):283–288.
7. KivimakiM,VirtanenM,VartiaM,ElovainioM,VahteraJ,Keltikangas-Jarvinen 28. KivimakiM,HeadJ,FerrieJE,ShipleyMJ,BrunnerE,VahteraJ,MarmotMG:
L:Workplacebullyingandtheriskofcardiovasculardiseaseand Workstress,weightgainandweightloss:evidenceforbidirectional
depression.OccupEnvironMed2003,60(10):779–783. effectsofjobstrainonbodymassindexintheWhitehallIIstudy.
8. KivimakiM,Leino-ArjasP,LuukkonenR,RiihimakiH,VahteraJ,KirjonenJ: IntJObes(Lond)2006,30(6):982–987.
Workstressandriskofcardiovascularmortality:prospectivecohort 29. NiedhammerI,GoldbergM,LeclercA,DavidS,BugelI,LandreMF:
studyofindustrialemployees.BMJ2002,325(7369):857–860. Psychosocialworkenvironmentandcardiovascularriskfactorsinan
9. NetterstromB,ConradN,BechP,FinkP,OlsenO,RuguliesR,StansfeldS: occupationalcohortinFrance.JEpidemiolCommunityHealth1998,
Therelationbetweenwork-relatedpsychosocialfactorsandthe 52(2):93–100.
developmentofdepression.EpidemiolRev2008,30(1):118–132. 30. OstryAS,RadiS,LouieAM,LaMontagneAD:Psychosocialandother
10. RuguliesR,BultmannU,AustB,BurrH:Psychosocialworkenvironment workingconditionsinrelationtobodymassindexinarepresentative
andincidenceofseveredepressivesymptoms:Prospectivefindings sampleofAustralianworkers.BMCPublicHealth2006,6:53.
froma5-yearfollow-upofthedanishworkenvironmentcohortstudy. 31. AmickBC,KawachiI,CoakleyEH,LernerD,LevineS,ColditzGA:
AmJEpidemiol2006,163(10):877–887. Relationshipofjobstrainandiso-straintohealthstatusinacohortof
11. KuperH,Singh-ManouxA,SiegristJ,MarmotM:Whenreciprocityfails: womenintheUnitedStates.ScandJWorkEnvironHealth1998,
effort-rewardimbalanceinrelationtocoronaryheartdiseaseandhealth 24(1):54–61.
functioningwithintheWhitehallIIstudy.OccupEnvironMed2002, 32. EpelE,JimenezS,BrownellK,StroudL,StoneyC,NiauraR:Arestress
59(11):777–784. eatersatriskforthemetabolicsyndrome?BiobehavioralStressResponse:
12. KarasekRA:Jobdemands,jobdecisionlatitude,andmentalstrain- ProtectiveandDamagingEffects2004,1032:208–210.
implicationsforjobredesign.AdmSciQ1979,24(2):285–308. 33. StoneAA,BrownellKD:Thestress-eatingparadox-multipledaily
13. StansfeldS,CandyB:Psychosocialworkenvironmentandmentalhealth- measurementsinadultmalesandfemales.PsycholHealth1994,
ameta-analyticreview.ScandJWorkEnvironHealth2006, 9(6):425–436.
32(6):443–462. 34. GrothMV,FagtS,StockmarrA,MatthiessenJ,Biltoft-JensenA:Dimensions
14. KouvonenA,KivimakiM,ElovainioM,VirtanenM,LinnaA,VahteraJ:Job ofsocioeconomicpositionrelatedtobodymassindexandobesity
strainandleisure-timephysicalactivityinfemaleandmalepublicsector amongDanishwomenandmen.ScandJPublicHealth2009,
employees.PrevMed2005,41(2):532–539. 37(4):418–426.
15. KouvonenA,KivimakiM,CoxSJ,PoikolainenK,CoxT,VahteraJ:Jobstrain, 35. LallukkaT,LaaksonenM,MartikainenP,Sarlio-LahteenkorvaS,LahelmaE:
effort-rewardimbalance,andheavydrinking:astudyin40,851employees. Psychosocialworkingconditionsandweightgainamongemployees.Int
JOccupEnvironMed2005,47(5):503–513. JObes2005,29(8):909–915.
16. AlbertsenK,HannerzH,BorgV,BurrH:Theeffectofworkenvironment 36. LallukkaT,LahelmaE,RahkonenO,RoosE,LaaksonenE,MartikainenP,
andheavysmokingonthesocialinequalitiesinsmokingcessation. HeadJ,BrunnerE,MosdolA,MarmotM,SekineM,NasermoaddeliA,
PublicHealth2003,117(6):383–388. KagamimoriS:Associationsofjobstrainandworkingovertimewith
17. AlbertsenK,HannerzH,BorgV,BurrH:Workenvironmentandsmoking adversehealthbehaviorsandobesity:EvidencefromtheWhitehallII
cessationoverafive-yearperiod.ScandJPublicHealth2004, Study,HelsinkiHealthStudy,andtheJapaneseCivilServantsStudy.
32(3):164–171. SocSciMed2008,66(8):1681–1698.
18. KawakamiN,TsutsumiA,HarataniT,KobayashiF,IshizakiM,HayashiT, 37. SobalJ,StunkardAJ:Socioeconomic-statusandobesity-areviewofthe
FujitaO,AizawaY,MiyazakiS,HiroH,MasumotoT,HashimotoS,ArakiS: literature.PsycholBull1989,105(2):260–275.
Jobstrain,Worksitesupport,andnutrientintakeamongemployed 38. SobalJ,HansonKL,FrongilloEA:Gender,ethnicity,maritalstatus,and
Japanesemenandwomen.JEpidemiol2006,16(2):79–89. bodyweightintheunitedstates.Obesity2009,17(12):2223–2231.
19. SorensenG,StoddardAM,DubowitzT,BarbeauEM,BigbyJ,EmmonsKM, 39. vanDrongelenA,BootCRL,MerkusSL,SmidT,vanderBeekAJ:Theeffects
BerkmanLF,PetersonKE:Theinfluenceofsocialcontextonchangesin ofshiftworkonbodyweightchange-asystematicreviewof
fruitandvegetableconsumption:resultsofthehealthydirections longitudinalstudies.ScandJWorkEnvironHealth2011,37(4):263–275.
studies.AmJPublicHealth2007,97(7):1216–1227. 40. BersetM,SemmerNK,ElferingA,JacobshagenN,MeierLL:Doesstressat
20. SiegristJ:Adversehealtheffectsofhigh-effort/low-rewardconditions. workmakeyougainweight?Atwo-yearlongitudinalstudy.ScandJ
JOccupHealthPsychol1996,1(1):27–41. WorkEnvironHealth2011,37(1):45–53.
21. BrunnerEJ,ChandolaT,MarmotMG:Prospectiveeffectofjobstrainon 41. KristensenTS,HannerzH,HoghA,BorgV:TheCopenhagenPsychosocial
generalandcentralobesityintheWhitehallIIstudy.AmJEpidemiol2007, Questionnaire-atoolfortheassessmentandimprovementofthe
165(7):828–837. psychosocialworkenvironment.ScandJWorkEnvironHealth2005,
22. KouvonenA,KivimakiM,CoxSJ,CoxT,VahteraJ:Relationshipbetween 31(6):438–449.
workstressandbodymassindexamong45,810femaleandmale 42. KristensenTS:Aquestionnaireismorethanaquestionnaire.ScandJ
employees.PsychosomMed2005,67(4):577–583. PublicHealth2010,38:149–155.
GramQuistetal.BMCPublicHealth2013,13:43 Page9of9
http://www.biomedcentral.com/1471-2458/13/43
43. SetterlindS,LarsonG:Thestressprofile-Apsychosocialapproachto
measuringstress.StressMedicine1995,11(2):85–92.
44. MarmotMG,SmithGD,StansfeldS,PatelC,NorthF,HeadJ,WhiteI,
BrunnerE,FeeneyA:Healthinequalitiesamongbritishcivil-servants-the
Whitehall-IiStudy.Lancet1991,337(8754):1387–1393.
45. KarasekR,BrissonC,KawakamiN,HoutmanI,BongersP,AmickB:TheJob
ContentQuestionnaire(JCQ):aninstrumentforinternationally
comparativeassessmentofpsychosocialjobcharacteristics.JOccup
HealthPsychol1998,3(4):322–355.
46. WareJE,SherbourneCD:TheMos36-ItemShort-FormHealthSurvey
(Sf-36).1.Conceptual-frameworkanditemselection.MedicalCare1992,
30(6):473–483.
47. BurrH,AlbertsenK,RuguliesR,HannerzH:Dodimensionsfromthe
CopenhagenPsychosocialQuestionnairepredictvitalityandmental
healthoverandabovethejobstrainandeffort-rewardimbalance
models?ScandJPublicHealth2010,38:59–68.
48. LundT,LabriolaM,ChristensenKB,BultmannU,VilladsenE,BurrH:
Psychosocialworkenvironmentexposuresasriskfactorsforlong-term
sicknessabsenceamongDanishemployees:ResultsfromDWECS/
DREAM.JOccupEnvironMed2005,47(11):1141–1147.
49. RuguliesR,AustB,PejtersenJH:Dopsychosocialworkenvironment
factorsmeasuredwithscalesfromtheCopenhagenPsychosocial
Questionnairepredictregister-basedsicknessabsenceof3weeksor
moreinDenmark?ScandJPublicHealth2010,38:42–50.
50. PejtersenJH,KristensenTS,BorgV,BjornerJB:Thesecondversionofthe
CopenhagenPsychosocialQuestionnaire.ScandJPublicHealth2010,
38:8–24.
51. BultmannU,KantI,VandenBrandtPA,KaslSV:Psychosocialwork
characteristicsasriskfactorsfortheonsetoffatigueandpsychological
distress:prospectiveresultsfromtheMaastrichtCohortStudy.Psychol
Med2002,32(2):333–345.
52. NiedhammerI,GoldbergM,LeclercA,BugelI,DavidS:Psychosocial
factorsatworkandsubsequentdepressivesymptomsintheGazel
cohort.ScandJWorkEnvironHealth1998,24(3):197–205.
53. HarmaM:Workhoursinrelationtoworkstress,recoveryandhealth.
ScandJWorkEnvironHealth2006,32(6):502–514.
54. HollmannS,KlimmerF,SchmidtKH,KylianH:Validationofaquestionnaire
forassessingphysicalworkload.ScandJWorkEnvironHealth1999,
25(2):105–114.
55. SiegristJ,RodelA:Workstressandhealthriskbehavior.ScandJWork
EnvironHealth2006,32(6):473–481.
56. WardleJ,ChidaY,GibsonEL,WhitakerKL,SteptoeA:Stressandadiposity:
ameta-analysisoflongitudinalstudies.Obesity2011,19(4):771–778.
57. SobalJ,RauschenbachBS,FrongilloEA:Marital-status,fatnessandobesity.
SocSciMed1992,35(7):915–923.
58. LipowiczA,GronkiewiczS,MalinaRM:Bodymassindex,overweightand
obesityinmarriedandnevermarriedmenandwomeninPoland.AmJ
HumBiol2002,14(4):468–475.
59. AverettSL,SikoraA,ArgysLM:Forbetterorworse:Relationshipstatus
andbodymassindex.EconHumBiol2008,6(3):330–349.
60. KuboT,OyamaI,NakamuraT,ShiraneK,OtsukaH,KunimotoM,Kadowaki
K,MaruyamaT,OtomoH,FujinoY,MatsumotoT,MatsudaS:Retrospective
cohortstudyoftheriskofobesityamongshiftworkers:findingsfrom
theIndustry-basedShiftWorkers'Healthstudy,Japan.OccupEnvironMed
2011,68(5):327–331.
61. GorberSC,TremblayM,MoherD,GorberB:Acomparisonofdirectvs.
self-reportmeasuresforassessingheight,weightandbodymassindex:
asystematicreview.ObesRev2007,8(4):307–326.
Submit your next manuscript to BioMed Central
62. KrulAJ,DaanenHAM,ChoiH:Self-reportedandmeasuredweight,height
andbodymassindex(BMI)inItaly,theNetherlandsandNorthAmerica. and take full advantage of:
EurJPublicHealth2011,21(4):414–419.
• Convenient online submission
doi:10.1186/1471-2458-13-43
• Thorough peer review
Citethisarticleas:GramQuistetal.:Psychosocialworkenvironment
factorsandweightchange:aprospectivestudyamongDanishhealth • No space constraints or color figure charges
careworkers.BMCPublicHealth201313:43. • Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit