Objectives The aim of this study was to investigate the association between multiple lifestyle-related risk factors (unhealthy diet, low leisure-time physical activity, overweight/obesity and smoking) and self-rated work ability in a general working population.
Setting Population-based cross-sectional study, in Telemark County, Norway, 2013.
Participants A random sample of 50 000 subjects was invited to answer a self-administered questionnaire and 16 099 responded. Complete data on lifestyle and work ability were obtained for 10 355 participants aged 18–50 years all engaged in paid work during the preceding 12 months.
Outcome measure Work ability was assessed using the Work Ability Score (WAS)—the first question in the Work Ability Index. To study the association between multiple lifestyle risk factors and work ability, a lifestyle risk index was constructed and relationships examined using multiple logistic regression analysis.
Results Low work ability was more likely among subjects with an unhealthy diet (ORadj 1.3, 95% CI 1.02 to 1.5), inactive persons (ORadj 1.4, 95% CI 1.2 to 1.6), obese respondents (ORadj 1.5, 95% CI 1.3 to 1.7) and former and current smokers (ORadj 1.2, 95% CI 1.1 to 1.4 and 1.3, 95% CI 1.2 to 1.5, respectively). An additive relationship was observed between the lifestyle risk index and the likelihood of decreased work ability (moderate-risk score: ORadj 1.3; 95% CI 1.1 to 1.6; high-risk score: ORadj 1.9; 95% CI 1.6 to 2.2; very high risk score: ORadj 2.4; 95% CI 1.9 to 3.0). The overall population attributable fraction (PAF) of low work ability based on the overall risk index was 38%, while the PAFs of physical activity, smoking, body mass index and diet were 16%, 11%, 11% and 6%, respectively.
Conclusions Lifestyle risk factors were associated with low work ability. An additive relationship was observed. The findings are considered relevant to occupational intervention programmes aimed at prevention and improvement of decreased work ability.
- lifestyle risk factors
- work ability
- general working population
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
The study included 10 355 subjects from the general working population in Telemark, Norway.
Inclusion of several lifestyle-related factors allowed for examination of both independent and additive associations between lifestyle and work ability.
The study is strengthened by inclusion of several adjustment variables/possible confounders (age, gender, educational level and occupation) in the regression analyses.
Potential study limitations are selection bias due to non-response, the cross-sectional design, lifestyle and weight self-reports and non-attendance of older than 50.
As in many other European countries,1 Norway’s population and workforce are ageing. The challenges this presents have given rise to government policies with a stronger emphasis on work ability promotion and extension of working life. Work ability is a multifactorial concept encompassing the worker’s health status, physical capacity and psychological resources2 and may be defined as the balance between the self-perceived physical and mental capacity and work demands.2 3
Promoting and maintaining good work ability in all phases of working life is vital, as poor work ability has been linked with increased risk of reduced work quality,4 sickness absence,5–7 long-term disability,7 8 early retirement5 9 and long-term unemployment.7 Good mid-life work ability may also protect against old-age mobility limitation, regardless of type of retirement.10 A person’s work ability may be influenced by various work-related and individual factors.11 12 At the individual level, lifestyle-related factors (such as diet, physical activity [PA], body mass index [BMI] and smoking) are known to have a significant impact on health.13 14 However, the contribution of lifestyle to variation in work ability is not fully understood. The most commonly used method for assessing self-rated work ability is the Work Ability Index (WAI), developed by researchers of the Finish Institute of Occupational Health.15 A corresponding instrument is the first single-item question in the WAI, the Work Ability Score (WAS).16
Previous cross-sectional and longitudinal studies have investigated the relationship between different lifestyle factors and work ability (measured by WAI or WAS).11 12 17–26 A systematic review covering 14 cross-sectional and 6 longitudinal studies of lifestyle and work ability published from 1985 to 2006 has identified low leisure-time PA and obesity as important determinants of decreased work ability in different occupational groups.11 Recent studies support these findings.12 17–23 25–27 A limited number of studies have indicated a positive association between healthy diet indicators (high intake of fibre/fruits and vegetables) and good work ability.11 24 27 Non-smoking has also been associated with good work ability in some studies,6 11 19 although the results on smoking and work ability remain inconclusive.11 17 Previous studies have commonly focused on distinct occupational groups, groups with certain job demands and selected age groups,11 12 17–20 23 26 27 rather than on general working populations.21 22 24 Additional studies assessing large general working populations are warranted to investigate whether lifestyle changes could enhance work ability across occupations and ages.
Lifestyle-related risk factors are often observed together.28 Previous lifestyle and health studies have shown associations between multiple lifestyle risk indicators on non-communicable, chronic diseases and all-cause mortality,29 self-rated health,30 31 long-term work disability (early retirement)8 and sickness absence due to several diseases.32 However, few studies have focused on associations between multifactorial lifestyle risk and work ability. It appears that only one small (n=187) Polish study conducted among professionally active subjects has investigated the additive relationship between multiple, simultaneously applicable lifestyle indicators and modification of work ability. In that study, the authors identified an additive association between a healthy lifestyle index (incorporating recommended PA, normal BMI, non-smoking and fibre intake) and increasing WAI.24 Given the lack of larger studies exploring multifactorial associations between lifestyle and work ability, supplementary studies are needed. Available Norwegian studies have mainly investigated the effect of psychosocial, social and mechanical work exposure on work ability,33 rather than the potential contribution of lifestyle factors.
The aim of the present study was to investigate the association between lifestyle-related risk factors (unhealthy diet, low leisure-time PA, obesity and smoking) and self-rated work ability in a large general population of employed adults in Norway.
Study population and design
The cross-sectional ‘Telemark Study’ was carried out from February to August 2013 in Telemark County, which is located in the Southeastern part of Norway and has a population of about 170 000. A sample of 50 000 males and females aged 16–50 years, from the approximately 80 000 residents in Telemark, was drawn randomly using the services of the Norwegian national population registry. Of the 50 000 who received the questionnaire, 1793 had moved, 4 were deceased, 13 were unable to answer due to disease or disability, 23 could not answer due to language problems and 25 were ineligible for other reasons. Of the 48 142 eligible participants, a total of 16 099 answered the questionnaire, resulting in a response rate of 33%. Participation was highest among the older age groups, women and participants from urban areas. The data collection and recruitment methods and characteristics of the non-responders have been described in detail elsewhere.34
Participants were asked questions on diet, PA, height and weight, and background variables at baseline. Employees were defined as subjects engaged in paid work during the preceding 12 months. Participants aged 16–17 years were excluded from the study due to low work engagement in this group. Complete data for the present analyses (diet, PA, smoking habits, height and weight and work ability) were available for 10 355 participants.
Self-rated work ability was assessed using the first single-item question in the WAI,15 the WAS16: ‘Current work ability compared with the lifetime best’, where a score of 0 represents complete work disability and a score of 10 represents work ability at its best. Previous studies have demonstrated a strong association between WAS and the complete WAI.9 21 WAS has been recommended and used as a simple, reliable indicator of work ability in several population studies.5 9 17 21 35 In this study, work ability was divided into two categories: low work ability (score 0–7) and good work ability (score 8−10).20 21 25 35
Diet was determined using food frequency questions previously used in the Norwegian population-based Nord-Trøndelag Health Study (HUNT3) (2006–2008).36 37 The questions were selected from a larger validated food frequency questionnaire used in the Oslo Health Study of 200138 and covered habitual intake of fruits/berries, vegetables, boiled potatoes, pasta/rice, fat fish, sausages/hamburgers and chocolate/candies, with the response options ‘0–3 times/month’, ‘1–3 times/week’, ‘4–6 times/week’, ‘1 time/day’ and ‘≥2 times/day’. To reflect general dietary advice for improved health,39 the following indicators and cut-off points were used: intake of fruits/berries and vegetables (≥2 times/day), fat fish (1–3 times/week) and sausages/hamburgers and chocolate/candies (≤1–3 times/week). The responses were coded 0 (not meeting general dietary recommendations) or 1 (meeting general dietary recommendations). A diet sum score for each participant (scale 0–4) was calculated by summarising their scores for the four indicators, reflecting the number of recommendations met.40 The diet score was trichotomised into the categories ‘unhealthy’ (total score 0–1), ‘average’ (total score 2) and ‘healthy’ (total score 3-4) diet, to indicate different levels of health risk.
Moderate to vigorous leisure-time physical activity (MVPA) was assessed using questions covering frequency, intensity and duration of exercise used in the HUNT1 (1984–1986) and HUNT3 (2006–2008) studies.41 The questionnaire has previously been validated against objective measurement methods and the International Physical Activity Questionnaire, with good internal consistency.41 The participants reported average weekly frequency of exercise by answering the question, ‘How frequently do you exercise?’, which had the following answer options: ‘never’, ‘less than once a week’, ‘once a week’, ‘2–3 times a week’ and ‘almost every day (4–7 times a week)’. Average intensity was reported by answering the question, ‘If you exercise once or more a week, how hard do you exercise?’, which had the following answer options: ‘I do not become sweaty or breathless’, ‘I become sweaty or breathless’ and ‘I become almost exhausted’. Average duration was reported by answering the question, ‘For how long are you normally physically active?’, which had the following answer options: ‘less than 15 min’, ‘15–29 min’, ‘30 min–1 hour’ and ‘more than 1 hour’. To reflect recommendations on adult MVPA (≥150 minutes/week),39 the responses to the three questions were combined to give a total MVPA score.41 This was labelled ‘PA’ and dichotomised into ‘active’ and ‘inactive’. The weighted scores used to calculate the total score and the cut-off point reflecting recommended MVPA were set according to the values used in the HUNT1 and HUNT3 studies.41 42
BMI categories (underweight, normal weight, overweight and obesity) were calculated based on self-reported height and weight data. Cut-off points were chosen according to WHO reference values for adults: underweight (<18.5), normal weight (18.5–24.9), overweight (25–29.9) and obesity (≥30).43
Smoking was measured by asking three questions. The first was, ‘Do you smoke every day?’ Two follow-up questions were then asked: ‘Do you smoke occasionally?’ and ‘If not, have you smoked in the past?’ Smoking habits were divided into three categories labelled ‘current smoker’ (every day and occasional smoking combined), ‘former smoker’ and ‘never smoked’.
Lifestyle risk index
Based on current knowledge of associations between lifestyle, health and non-communicable diseases, an overall lifestyle risk index was constructed to study the possible association between multiple lifestyle risk factors and low work ability. To indicate overall lifestyle risk, the individual lifestyle factors were given weighted risk scores: 0 (low health risk), 0.5 (intermediate health risk) and 1 (high health risk), and then summed into an overall index ranging from 0 to 4. To study different levels of lifestyle risk, the lifestyle risk index was divided into four categories: ‘low-risk score’ (total score 0–0.5), ‘moderate-risk score’ (total score 1–1.5), ‘high-risk score’ (total score 2–2.5) and ‘very high risk score’ (total score 3–4). The index was labelled ‘lifestyle risk index’.
The participants were all between 18 and 50 years of age, and were grouped into three categories: ‘18–30 years’, ‘31–40 years’ and ‘41–50 years’.
The participants’ educational level was categorised as follows: ‘primary and lower secondary education’ (10 years or less), ‘upper secondary education’ (an additional 3–4 years) and ‘university or university college’.
The participants were classified by a trained research assistant based on self-reported current occupation (2013), using the International Standard Classification of Occupations-88 coding system.44 The 10 occupational groups were further combined into five subgroups for use in the analyses.
Spearman’s r was used to assess the correlation between the individual lifestyle risk factors. Multiple logistic regression analysis was used to assess associations between the four individual lifestyle factors and the multifactorial lifestyle risk index (independent variables), as well as the likelihood of low work ability (dependent variable). The individual lifestyle variables were mutually adjusted in the respective models. ORs with 95% CIs were calculated for the likelihood of low work ability. Forward conditional selection was applied to include available adjustment variables (gender, age, educational level and occupational group) associated with the respective independent variables in the models. The population attributable fraction (PAF) of low work ability was calculated for each lifestyle risk factor and the index.45 PAF is defined as the fraction of all cases of a particular disease or other adverse condition in a population that is attributable to the specific exposure.
Only participants with complete data for all main variables (lifestyle variables and WAS) were included in the analyses. Respondents with missing values for adjustment variables were included with ‘missing’ as a separate adjustment variable category. For all tests, p<0.05 was considered significant. The questionnaires were scanned by Eyes and Hands (Read-soft Forms, Helsingborg, Sweden), while the statistical analyses were carried out using IBM SPSS Statistics for Windows, V.23.
Patient and public involvement
To release the full potential of the study, we have involved user-representatives in the study planning, design and transfer of knowledge. Resourceful user-representatives are engaged in the dissemination of results to the public, policy-makers and to healthcare workers through regional, national and international media on all platforms (newspapers, internet, radio and television). An user-representative is the member of the steering committee and has given valuable contributions in development of questionnaires. In addition user-representatives are involved in piloting the questionnaire.
A total of 16 099 of the 48 142 eligible subjects answered the questionnaire. Of these, 12 932 had been employed during the preceding 12 months and were aged 18 or older. Complete data on lifestyle variables and work ability were obtained for 10 355 respondents. Further background characteristics of the study population are shown in table 1. The distributions of the main variables are specified in table 2. The associations between multiple and independent associations between individual lifestyle factors and the likelihood of low work ability are presented in table 3.
Spearman’s r correlations between individual lifestyle-related risk factors were ranging from 0.027 between BMI and diet to 0.117 between PA and diet. Multiple logistic regression showed independent associations between individual lifestyle factors and the likelihood of low work ability (table 3, model 1). Participants in the category ‘unhealthy diet’ were more likely to have low work ability than participants with a ‘healthy diet’ (ORadj2 1.3; 95% CI 1.02 to 1.5). Inactive subjects were more likely to have low work ability than active individuals (ORadj2 1.4; 95% CI 1.2 to 1.6). Obese participants had lower work ability than normal-weight subjects (ORadj2 1.5; 95% CI 1.3 to 1.7). Former and current smokers were more likely to have low work ability than those who had never smoked (ORadj2 1.2; 95% CI 1.1 to 1.4 and ORadj2 1.3; 95% CI 1.2 to 1.5, respectively). All associations were observed independently of other lifestyle factors and available background variables (gender, age, educational level and occupational group).
An association was observed between the lifestyle risk index and the likelihood of low work ability (table 3, model 2). The figures were as follows: moderate-risk score: ORadj2 1.3; 95% CI 1.1 to 1.6; high-risk score: ORadj2 1.9; 95% CI 1.6 to 2.2; very high risk score: ORadj2 2.4; 95% CI 1.9 to 3.0. The analyses were adjusted for available background variables. The overall PAF of low work ability based on the overall risk scores was 38%, while the PAFs of PA, smoking, BMI and diet were 16%, 11%, 11% and 6%, respectively.
In the present study, consistent associations were found between several lifestyle risk factors and self-rated low work ability in a general working population in Norway. Obesity was the factor which was most strongly associated with low work ability, followed by low PA, current smoking and unhealthy diet/former smoking. Further, an additive relationship was observed between multiple risk factors and work ability. Increasing scores on a multiple lifestyle risk index were associated with increasing likelihood of low work ability. An overall PAF of 38% indicated a substantial contribution of lifestyle to work ability. Of the individual lifestyle factors, low PA had the highest observed PAF (16%). All associations were observed independently of gender, age, educational level and occupation.
A direct comparison with other studies is difficult, due to heterogeneity of study design, definition and measurement of lifestyle indicators, varying population sizes and varying use of complete WAI or WAS. However, some similarities and differences can be noted.
The results agree with previous studies in which unhealthy diet indicators were linked with low work ability.11 24 27 Unhealthy diet, characterised by low consumption of healthy foods or nutrients, has previously been associated with low mental and physical health in a number of population studies.46–50 Work ability has previously been strongly associated with mental and physical health.17 One possible explanation for the findings is that an unhealthy diet may influence self-perceived work ability through decreased physical and mental capacity related to job demands.2 Currently, little information is available on how measures to promote healthy eating at the workplace can have positive impact in this context. However, the results indicate that a diet close to the recommended composition could improve work ability.
There is convincing evidence that regular PA helps to prevent various chronic diseases and improve health-related quality of life.51–53 It is, therefore, likely that physically active individuals are better equipped to meet physical and psychological demands at work and to achieve better work ability. In accordance with previous occupation-specific studies,11 17–20 24 26 27 low leisure-time PA was associated with low work ability in the present sample from the general working population. Earlier studies indicate that the benefits of and need for PA differ between job types. A recently published Danish study focusing on workers performing physically demanding tasks concluded that PA must be of high intensity and long duration to increase work ability.23 In contrast, it has also been suggested that mentally demanding jobs do not necessarily require good physical condition to meet work demands, at least not among younger workers.17 A Swedish prospective study of healthcare workers found that leisure-time PA at the recommended level or higher improved work ability both immediately and in the longer term.18 Correspondingly, the results in the present study show that achieving the recommended level of weekly leisure-time MVPA reduces the likelihood of low work ability, indicating a beneficial effect across occupations and ages. Further, recent research indicates that PA at the workplace may have an additional favourable impact on work ability due to positive effects on social relationships and psychological well-being.54
In line with previous studies,11 17 19 20 24 27 a significant association was observed between obesity and low work ability. Obese respondents had a 50% higher likelihood of low work ability than respondents with a normal weight. In a systematic review published in 2009, five out of seven studies (mainly concentrating on Finnish municipal workers and caregivers) reported an association between obesity and low work ability in different occupational groups.11 A recent Danish study of a general working population of 10 000 adults has shown that increasing BMI above normal range is associated with lower work ability.22 A similar trend was observed in the present study, with the likelihood of decreased work ability increasing gradually as BMI rises. However, the results for the overweight respondents did not reach significance in the adjusted models. There are several possible explanations for the observed association, ranging from individual health problems due to obesity to psychosocial problems and physical limitations at the workplace.55
Smokers (both current and former) showed a higher likelihood of low work ability than non-smokers. However, there is no unanimous agreement on this association. While some studies have failed to demonstrate a significant difference,17 21 27 other studies support our findings.19 20 24 A Dutch study of workers with common diseases found significance only for participants with respiratory diseases,20 while another study found significance for women only.24 In contrast, the effect of occasional smoking on work ability has been found to be more evident for men than for women.56 Contradictory findings may be explained by the fact that earlier studies have examined different occupational groups, not the general working population. A possible explanation for the observed association is impaired health status or chronic conditions due to current or former smoking, which in turn may have impaired work ability.56 The results indicate that former smokers may also be at risk of low work ability, emphasising the importance of assessing this group as well.
Although the individual lifestyle risk factors appeared to be slightly correlated, independent associations with low work ability were observed for each factor. The individual factors were added up to compose a lifestyle risk index. Lifestyle risk indexes can be used as indicators of overall or cumulative risk of non-communicable diseases.29 As suggested by others,24 an additive association was observed between lifestyle risk factors and work ability. Participants with a high or very high risk score on the lifestyle risk index were more than twice as likely to have low work ability, than those with a low-risk score. The effect seems to be additive rather than synergetic as the strength of the associations of more than one risk factor was not stronger than the sum of the risks of the underlying factors.20 Moreover, additional analyses of the most prevalent risk factor combinations did not show any significant synergetic effects either (data not shown). As the relative importance of the lifestyle risk factors to good health, non-communicable diseases and low work ability has not been fully determined, we decided to weight each factor equally in the lifestyle risk index. The decision to weight the single risk factors equally was further supported by the comparable effects of the individual factors on observed WAS (table 3).
A PAF of 38% indicates a substantial contribution of multiple lifestyle risk to low work ability. According to the lifestyle risk index, a considerable proportion (36%) of the participants had a high or very high risk score. Knowing that an unhealthy lifestyle increases the risk of various non-communicable diseases, it can be assumed that lifestyle changes in line with current health recommendations would improve the prognoses of these diseases and indirectly improve work ability. Although PA had the highest PAF, all four risk factors contributed significantly to low work ability, underlining the importance of targeting multiple lifestyle changes.
Although no causality can be claimed based on the present results, the associations indicate that occupational health promotion strategies should target multiple lifestyle changes to reduce the likelihood of decreased work ability. Lifestyle is theoretically modifiable, but often considered a personal matter with no formal responsibility resting with the employer. However, facilitating lifestyle changes through workplace measures may be beneficial for both employers and employees in terms of improved work ability.
The present study has strengths, but also limitations that should be recognised. An important strength is the large study sample, which covers all types of occupational groups and a broad age range. Simultaneous assessment of several lifestyle-related factors has allowed mutual adjustment and examination of both independent and additive relationships. Further, the study has employed validated questions for diet,57 leisure-time MVPA41 and self-assessed work ability.9 21 The dichotomisation of the total MVPA score into ‘active’ and ‘inactive’ gives good information on MVPA by reference to current recommendations on PA.39 41 42 The dietary score appears to be a comprehensive indicator of healthy dietary behaviour, compared with previous studies in which the ‘diet’ variable was either not fully elucidated27 or consisted only of single nutrients or single food items.11 24 The first single-item question of the WAI, the WAS16—‘Current work ability compared with lifetime best’—was used to assess work ability. This item has become established as a practical, simple and valid indicator of work ability,9 21 often replacing complete WAI in clinical practice and research58 59 and increasingly used in population studies.5 9 17 21 35 In accordance with these studies, work ability was considered to be good when the score was between 8 and 10.
Several individual and environmental factors have previously been associated with decreased work ability and/or lifestyle.11 17 60–62 To investigate independent relationships between lifestyle and work ability, several adjustment variables (age, gender, educational level and occupation) were included in the regression analyses. However, the adjustment did not alter the estimates substantially, indicating independent associations and limited risk of overadjustment. Nevertheless, the possibility cannot be excluded that other individual and environmental characteristics such as poor musculoskeletal capacity, chronic disease, psychosocial factors at work and high physical or mental work demands may have attenuated the associations.11 17 20 33 61
The present study did not include workers older than 50 years of age. Therefore, it cannot be concluded that the findings are generalisable to older age groups. Previous research has indicated that lifestyle may be even more important to older workers than younger in terms of good work ability.11 17 Moreover, promoting good work ability through a healthy lifestyle early on may reduce the risk of non-communicable chronic diseases and consequent impaired work ability later in life.14 60
Participants’ self-reported diet and PA data may have caused bias due to under-reporting of unhealthy habits and/or over-reporting of healthy habits, or bias due to deficient recollection. However, the applied questions on food items and PA have demonstrated good reliability and validity when compared with objective measures and other validated questionnaires.41 57 Self-reported weight and height is known to be prone to bias, and misreporting may have influenced the observed associations. Nevertheless, the proportion of participants in the overweight and obese categories was in line with national BMI data for adults.63 As regard to self-reported smoking history, previous studies have indicated high reliability of self-reporting.64 In addition, occasional smokers were included in the current smoker category to capture all at-risk respondents, as infrequent and occasional smokers may still have a nicotine dependency and may under-report.65
Another limitation of the study is the low response rate (33%), which may have caused bias due to non-response.34 There was a predominance of participants from older age groups, women and participants from urban areas. Further, only participants with complete data on lifestyle indicators and work ability were analysed. However, non-response to the postal questionnaire has been assessed,34 showing that responders and non-responders had similar frequencies of respiratory symptoms and asthma, but that young males and past smokers were somewhat under-represented and that weighting according to inverse probability of non-response did not alter the results substantially (data not shown).
Data collection was limited to one Norwegian county, and the results are therefore not necessarily representative of the national population. Finally, the study’s cross-sectional design makes it impossible to identify causal relationships between lifestyle indicators and work ability.
In the present study, significant associations have been identified between several lifestyle risk factors and low work ability in a general working population. Moreover, an additive relationship between multiple lifestyle risk factors and low work ability has been observed. The results indicate that employees in general may benefit from interventions targeting multiple lifestyle changes. Further, the results appear relevant to occupational intervention programmes aimed at preventing and improving low work ability. A follow-up study is planned to investigate the observed associations over time, with a particular focus on ageing and workers with chronic diseases.
The authors thank Regine Abrahamsen, Geir Klepaker and Gølin Finkenhagen Gundersen for data collection.
The authors are grateful to Johny Kongerud for contribution to the design of the main study.
Contributors AKMF and MVS were involved in the conception, design and data collection. IMO was involved in selection of research questions and planning of the statistical analyses. MVS was responsible for the statistical analyses. All authors were involved in the interpretation of the results. IMO drafted the manuscript with the assistance from MMDB. All authors revised the manuscript critically and approved the final manuscript.
Funding The research was supported by funding from University of South-Eastern Norway and Telemark Hospital.
Competing interests None declared.
Ethics approval The study was conducted in accordance with the guidelines laid down in the Declaration of Helsinki and was approved by the Regional Committee for Ethics in Medical Research and the Norwegian Data Protection Authority (REC 2012/1665).
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The datasets generated and/or analysed during the current study are not publicly available due to individual privacy regulations, but are available from the corresponding author on reasonable request.
Patient consent for publication Not required.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.