Participants were originally recruited for a population-based study initiated in 1990 to examine issues related to mental health within two geographically diverse areas in Norway. Further details regarding the study are presented elsewhere 19. A total of 2727 adults from the two study regions (Oslo, urban and Lofoten, rural) were randomly selected by Statistics Norway. Of these, 2014 (74%) individuals participated, who were representative of age and gender for the general Norwegian population. Ten years later, 65% (N = 1300) were available for a follow-up study, plus 1000 new randomly selected participants were added to increase the number of younger individuals and immigrants within the sample. Of the 2300 invited, 1691 persons (803 men and 888 women) participated in the follow-up. For the present study, all women who reported having paid employment were selected for the analyses (597 of 888). All data were collected using structured face-to-face interviews by trained interviewers.

Work ability was assessed by the question "How do you estimate your work ability today?" This question was selected from the Graded Reduced Work Ability Scale, which was constructed for the Norwegian Ministry of Health and Social Affairs 20. Responses are scored on a scale from 1 (extremely reduced) to 6 (not reduced at all). Work ability was recoded into three categories: severe reduced (1–2) work ability, moderately reduced (3–5) and not reduced (6) work ability.

Individual factors were age, ethnicity, marital status, and residence. Ethnicity was dichotomized according to whether the woman was born in Norway or not. Marital status was dichotomized according to whether the woman was married/cohabitating or not. Place of residence variable was categorized according to urban (Oslo), rural (Lofoten), or other (moved elsewhere). Working time was measured by working days per week. Four days and more per week were coded as full time work, and one day and less – part time work.

Physical health was assessed by the question "How satisfied are you with your physical health and well-being?" The score ranged from 1 (extremely satisfied) to 7 (extremely dissatisfied). The item was recoded into three categories: satisfied (from 1 to 3), partially satisfied (4), and dissatisfied (from 5 to 7). Weight and height were registered and body mass index (BMI) was calculated using the formula weight (kg) divided by height (m) squared. Consistent with established criteria, obesity was defined as a BMI of ≥ 30 kg/m² and participants were classified as obese or non-obese.

Mental health status was assessed by Hopkins Symptoms Checklist (HSCL-25) 21. The HSCL-25 is a 25-item self-report questionnaire about the presence and intensity of anxiety and depression symptoms during the previous week. Items are scored on a scale from 1 (no distress) to 4 (extremely distressed). The HSCL-25 score was calculated as the sum score of items divided by number of items answered. To be counted as a valid HSCL-25 score, at least 13 items had to be answered. A score from 1 to 1.54 was defined as "no mental distress," from 1.55 to 1.74 as a "mild mental distress," and score equal to or larger than 1.75 was defined as "severe mental distress" 1922.

The work environment was assessed based on seven questions originally used in work/life household surveys in Norway 23. Physical and mental strain at work were assessed by the questions, "Is your job so physically strenuous that you are often physically tired after work?" and "Does your work demand so much concentration and attention that you often feel exhausted after work?" Job stress was assessed by the question, "Is there a lot of stress at your workplace?" Scores ranged from 1 (never) to 4 (almost always) and were recoded into two categories: never (1) and sometimes/all the time (from 2 to 4). Level of decision-making at work was assessed by the question, "To what extent do you personally make decisions about your work?" Scoring ranged from 1 (I decide myself) to 4 (immediate superior decides what and how), and the score was then recoded into two categories: decide myself (1) and partial control over decisions/immediate supervisor decides what and how (from 2 to 4). Opportunity for effectively utilizing skills at work was assessed by the question, "Do you feel that you utilize your skills and abilities at work?" Scores ranged from 1 (yes) to 4 (no). The answers were recoded to: yes (1) and partially/no (from 2 to 4). Job insecurity was assessed by the question, "Is there any risk that you might lose your current job in the near future?" The score ranged from 1 (yes, absolutely) to 5 (no, very unlikely) and responses were dichotomized into yes (from 1 to 3) and no (from 4 to 5). Job satisfaction was assessed by the question "How satisfied are you with your job?" The score ranged from 1 (very satisfied) to 7 (not satisfied at all) and was recoded to: satisfied (from 1 to 3), partially satisfied/dissatisfied (4–7). These seven work environment variables were included both separately and as a sum score. The work environment sum score (0 to 13) was recoded into three categories: good (0 – 3), average (4 – 6), and poor (7 – 13).

Occupation was registered as free-text, and then recoded according to the Standard Classification of Occupations and dichotomized into either 'skilled' or 'unskilled' work 24.

A chi-square test was used to analyse differences in proportions and Pearson's correlation was calculated for all the variables. Poisson univariate regression was calculated with 95% confidence intervals (CI) to estimate risk ratios (RR) between the dependent variable, reduced work ability, and the selected independent variables (demographic, physical health and well-being, mental health, and work environment). Finally, significant risk factors were entered into a Poisson regression model for both categories of dependent variable. The significance level was set at p < .05. All analyses were performed using STATA, version 8.2.

Table 1 illustrates the distribution of self-reported work ability according to age (years). Of the 597 female employees, 8.9% reported an extremely or very reduced ability to work.

Table 2 shows the level of self-reported work ability categorized into not reduced, moderately reduced and severely reduced according to demographic variables, health status, and psychosocial and organizational aspects of the work environment. The Pearson's chi-square to test proportional differences and the univariate rate ratios (RR; 95% CI) for level of work ability are also presented. Increasing age demonstrated a significant effect on the proportion of women reporting a reduction in work ability. The proportion of women reporting no reduction in work ability fell from 51.0% among the youngest to 19.4% among those 50 years and older.

Work in unskilled occupations (RR = 1.9) and feeling that women do not utilize her abilities at work (RR = 2.2) was associated with severe reduced work ability.

Both partially satisfaction and dissatisfaction with physical health and well-being were associated with moderately reduced work ability (RR = 1.6 and RR = 1.7) and severely reduced work ability (RR = 6.9 and RR = 14.4).

Mild and severe mental distress was also associated with both moderately (RR = 1.4 and RR = 1.9) and severely (RR = 2.7 and RR = 7.5) reduced work ability in the univariate analyses.

Reporting job dissatisfaction was associated with a moderately and severely reduction in work ability compared to satisfied employees (RR = 1.5 and RR = 5.3). Analysis with sum score for work environment showed that women who reported an "average" or "poor" working environment had a higher risk of reporting both moderately reduced work ability (RR = 1.3 and RR = 1.6) and severely reduced work ability (RR = 2.6 and RR = 5.6) than employees who were satisfied with the overall work environment.

Neither ethnicity, marital status, residence, working time, BMI, nor physically and mentally strenuous work, workplace stress, level of decision-making, and job insecurity were associated with work ability.

When examining the pattern of correlations among variables, several significant (p < 0.01) correlations existed: job satisfaction and mental health (r = 0.39), job satisfaction and physical health/well-being (r = 0.36), mental and physical health/well-being (r = 0.37), and workplace stress and mental strain (r = 0.37; data not shown in tables). There was a significant difference in the proportion of women reporting "job dissatisfaction" among those who were satisfied with their physical health/well-being and those who were not (2.3% versus 25.9%, respectively; p < 0.001). Similar results were found for mental distress (1.8% versus 22.4%, respectively; p < 0.001).

In the multivariate regression model, we included the variables, which had significant elevated incidence ratios in univariate analyses (see Table 3). In the multivariate analysis, the following variables were significant for moderately reduced work ability: age groups 30–39, 40–49 years and 50+ years (RR = 1.3, 1.5 and 1.8) and severe mental distress (RR = 1.5). Unskilled occupation (RR = 1.9) and partially satisfaction and dissatisfaction with physical health remained significant for severe reduced work ability (RR = 5.1 and IR = 9.5).