METHODS

Sample selection

The present study was conducted in a Dutch nationwide study group of 658 patients aged 18–59 years diagnosed by a rheumatologist. Eligible patients were identified using the national Standardised Diagnosis Register of Rheumatic diseases (SDR), a representative database with information on diagnoses of the patient population of 80% of all rheumatologists in the Netherlands.⁹ From this database a random sample of patients with AS aged 16–59 years was selected from a geographically representative sample of 34 rheumatologists (34% of all rheumatologists participating in the SDR, 27% of all rheumatologists in the Netherlands), working in 15 practices (42% of all SDR practices) throughout the Netherlands. All patients were contacted by mail by their own rheumatologist, and 658 (60.3% of the initial cohort, 70.0% of the patients traceable) agreed to participate in the study. All patients filled out a self administered questionnaire adjusted to their current employment status (having a paid job, having no paid job but had a paid job in the past, and never had a paid job) and containing several standardised instruments. Data were obtained on sociodemographics, disease characteristics, functional abilities, health related quality of life, present and past working conditions, and coping with the disease. We limit our description of variables to those which were assessed to address our present study purpose. The design of the present study has been described in detail elsewhere.^4,7

Sociodemograhic variables

Year of birth, sex, and highest attained level of education were identified as sociodemographic variables. Attained educational level was separated into three categories based on the Dutch educational system (primary education (0–8 years), secondary education (9–16 years), and higher vocational education/university (17 years and more)).

Disease characteristics

Disease activity was assessed using a Dutch version of the Bath ankylosing spondylitis disease activity index (BAS-DAI)¹⁰ containing six questions focusing on fatigue (one item), axial pain (one item), peripheral pain (one item), enthesopathy (one item), and morning stiffness (two items). Each question was scored on a visual analogue scale (VAS). The BASDAI score (0–10) was calculated as the mean value of the mean of the first four items and the mean of the last two items referring to morning stiffness. A higher score indicates more disease activity.

Patients were asked about the year in which AS was diagnosed. Furthermore, patients reported their current use of medication, whether they had total hip replacement, and whether they had AS related comorbidity and other chronic disorders.

Activities in daily life

To assess functional abilities in daily life, a Dutch version of the Bath ankylosing spondylitis functional index (BASFI) was used. This contains eight questions on function in AS and two questions reflecting the patients' ability to cope with everyday life.¹¹ Each question was scored on a VAS, the mean of which gave the BASFI score (0–10). A higher score indicates worse outcome.

Employment situation

All patients reported on their employment history using a Dutch generic structured instrument, the vocational handicap questionnaire (VHQ), which had been validated in previous studies among Dutch chronically ill people.^12–14 This instrument assesses present occupational situation, vocational handicaps, employment history, and income compensation by work disability pensions. Data were also collected on the year of the first real paid job and job changes. To describe the present or last job, data were collected on field of work, job demands, size of company, accessibility of the workplace and transportation mobility, job adjustments, social support from colleagues, and management in the workplace. Patients who had withdrawn from the labour force were asked to indicate in which year they stopped working and whether they had left for work related reasons, disease related reasons, or other personal reasons. For patients who left the labour force after being diagnosed, the duration of employment after diagnosis was calculated.

Coping

Behavioural coping styles were assessed using the disease specific coping with rheumatic stressors (CORS) instrument.^15,16 The CORS measures eight coping styles directed at the most important chronic stressors of inflammatory rheumatic diseases—that is, pain, limitations, and dependence. Three scales refer to pain: comforting cognition (nine items), decreasing activities (eight items), and diverting attention (eight items). Three scales measure styles of coping with limitations: optimism (five items), pacing—that is, adapting one's level of activity (10 items)—and creative solution seeking (eight items). Two scales refer to dependence: making efforts to accept one's level of dependency (six items), and showing consideration (seven items). For each item, patients reported how often they made use of that particular coping strategy (seldom or never, sometimes, often, very often).

Global wellbeing

The instrument used to assess global wellbeing was the disease specific Dutch version of the Bath ankylosing spondylitis global health index (BASG), containing two questions on the impact of AS on global wellbeing in the past week and in the past six months.¹⁷ Each question was scored on a VAS, the mean of which gave the BASG score (0–10). A higher score indicates more impact of the disease on the global wellbeing of the patient.

Data analysis

The influence of disease duration on the participation in the labour force of patients with AS was studied by comparing age and sex specific levels of employment in our AS population with those in the general population using an indirect standardisation method. We therefore calculated, for our study group, ratios of the observed (O) and expected (E) number of working people for evenly spaced five year categories of disease duration. Our study design meant that the first category ranged from one to five years of disease duration. Expected levels of participation in the labour force were computed using reference data for the general Dutch population relating to age and sex specific levels of employment.¹⁸ The O/E ratio can be interpreted as the relative chance of patients with AS having a paid job compared with the general population, adjusted for age and sex. A rate ratio of 1.0 indicates an equal chance, a rate ratio of less than 1.0 a decreased chance, and a rate ratio of more than 1.0 an increased chance. To examine statistically significant differences between the AS group and the general population, we obtained the variances of the O/E ratios, and estimated their 95% confidence intervals (O/E±1.96×√var(O/E)). If a rate ratio of 1.0 is included in the interval, then the estimate is not statistically significant at the α level of 0.05.¹⁹

In evaluating sociodemographics, disease related variables, coping styles, and various work related factors in relation to withdrawal from the labour force, we compared characteristics of patients who had withdrawn from the labour force any time after being diagnosed with AS and patients still employed at the time of the study. For employment factors, we examined the characteristics of the last job position of the patients who had withdrawn and the characteristics of the present job of the employed patients. Firstly, all variables were subjected to univariate analysis using χ² tests to determine proportional variance between groups and t tests were conducted to analyse differences between group means. In addition, we assessed the relative importance of sociodemographic and disease related factors and work related factors in the relationship with withdrawal from the labour force in a multivariate context. A multiple logistic regression analysis with a forward stepwise procedure was therefore performed to assess the simultaneous associations of those sociodemographic, disease related, and work related factors that were found to be significant at the p≤0.05 level, when tested by univariate analysis. A variable was entered into the equation when the α level for rejecting the null hypothesis was 0.20. This level was chosen to ensure adequate power for the selection of variables to have acceptable performance for statistical testing.¹⁹ For implementation of the results in practice, each continuous variable was dichotomised at the median score of the total group of patients who performed a job at any time after diagnosis.

On the basis of the final logistic regression model of the stepwise procedure, attributable fractions and preventable fractions were subsequently calculated to estimate the overall proportion of withdrawals from the labour force that could be either attributed to, or prevented by, sociodemographic, disease related, or work related factors. The adjusted attributable fraction is the proportion of withdrawals from the labour force, which is attributed to exposure by that factor. It was calculated as the exposed proportion of patients withdrawn from the labour force multiplied by (aOR-1)/aOR where aOR is the odds ratio for that risk factor, adjusted for all other factors in the final model.

The adjusted preventable fraction was the proportion of withdrawals in a situation of non-exposure that could be prevented by exposure to that factor. It was calculated as the unexposed proportion of patients withdrawn multiplied by (1-aOR).

To examine possible recall bias, we performed the same stepwise forward multiple regression analysis with a subsample of patients who had withdrawn from the labour force during the five years before the study and a subsample of patients with five or fewer years of disease. The cut off point of five years was chosen to ensure adequate statistical power for the logistic regression.¹⁹ All statistical analyses were performed using SPSS for Windows, version 10.0.7.

RESULTS

Overall participation in the labour force in this cohort of patients with AS with a mean disease duration of 12.4 years, standardised for age and sex, was 54.2% (63.4% in males and 46.9% in females), which was a significant reduction of 11.0% (15.4% in males and 5.2% in females) compared with the Dutch general population of working age. Data about overall participation in the labour force has been presented extensively in a previous paper.⁴ In further exploring the labour market position of these patients, we studied whether disease duration had an effect on the chance of having a paid job among patients with AS compared with the chance of having a paid job given a patient's age and sex according to reference data from the general population in the working age range. Figure 1 shows rate ratios and their 95% confidence intervals for the relative chance of labour force participation in patients with AS for five year categories of duration of disease. It was found that all of the rate ratios were significantly reduced, which implies that the probability of participating in the labour market was reduced in patients with a relatively short disease duration as well as in patients with long disease duration. Moreover, rate ratios of patients with longer disease duration did not differ significantly from patients with short disease duration—that is, having the disease influences the probability of holding a paid job, but the duration of the disease has no additional influence.

Of the total study group of 658 patients, 81 (12.3%) had either stopped working before diagnosis (n=41 patients (6.2%)), or had no work history at all (n=27 patients (4.1%)) or had an unknown work history (n=13 patients (2.0%)), whereas 577 patients (87.7%) held a paid job, at least temporarily, at some time after being diagnosed with AS. Work history of the total study group has previously been described in more detail.⁷

For studying determinants of withdrawal from the labour force, we restricted our study group to those 577 patients with a paid job sometime after diagnosis. Of these patients, 156 (27.0%) had withdrawn from the labour force after a mean disease duration of 6.3 years; 94.9% of these patients indicated that they stopped working for medical reasons. Most cited AS related reasons for withdrawal were physical limitations (86.1% of the patients withdrawn due to AS) and being fatigued at work (77.4% of the patients withdrawn due to AS). Only 3.2% of the withdrawals had stopped working because of retirement. Table 1 presents patient characteristics, disease status, and outcome of the group of patients that stopped working at any time after being diagnosed with AS compared with the group that was working at the time of assessment.

On the basis of univariate comparisons, it was found that, on average, patients with AS who were younger, male, or more highly educated were more likely to be employed at the time of assessment, a trend similar to that in the general population in the same working age range. Patients who had withdrawn from the labour force had significantly longer disease duration, showed higher disease activity, experienced more disabilities in daily life and used more medication, and more often had total hip replacement. In general, the group of patients who had withdrawn had different coping strategies and experienced less favourable global wellbeing compared with patients who were still in employment.

More than three quarters of the patients who had withdrawn from the labour force had been officially recognised as work disabled to some extent, as they received income compensation in the form of a work disablement benefit. In the Dutch social security system, employees who are on full or partial sick leave for more than a year are entitled to a work disablement benefit after work disability has been officially recognised on the basis of a medical and ergonomic examination. This pension provides income compensation up to a maximum of 70% of the last income earned, for a maximum period of five years. At the end of that period, the person's ability to work has to be reassessed. The level of compensation is dependent on age, number of years worked, and levels of work disablement. People whose ability to work is impaired by 80%–100% are entitled to full work disablement benefit. People whose ability to work is impaired by 15%–80% are entitled to partial work disablement benefit. In the Netherlands, it is therefore possible to receive work disablement benefit and to remain in employment, usually on a part time basis. In our study population, 18.3% of the patients with a paid job had been officially recognised as having a work disablement.

The associations between various employment factors and withdrawal from the labour force are presented in table 2.

Patients with AS who had withdrawn from the labour force were more often active in the industrial sector and less active as professionals (for example, physicians, teachers, and policy makers) than currently employed patients. Fewer patients who had withdrawn were specially trained for their last job position, fewer of them had followed any job training at all, and more of them had performed a physically demanding job at some time during their working career.

Patients who were still in employment worked more often in larger companies and more often had technical or ergonomic work adjustments.

Patients who had withdrawn from the labour force reported difficult access to the workplace and reduced transport mobility at the workplace more often than currently employed patients. With respect to the social environmental factors at the workplace, patients who had withdrawn were much more dependent on their colleagues. Compared with patients still in employment, they also thought more often that attitudes of colleagues and management were negative and that there was insufficient support from both colleagues and management.

Table 3 lists the results of the stepwise forward logistic regression for the combined effect of sociodemographic and disease related factors and work related factors. Because functional ability, use of medication, and global wellbeing were highly correlated with disease activity and field of activity correlated highly with educational level (r>0.6), we chose to enter only the variables for disease activity and educational level into the models to prevent colinearity.

Variables listed in table 3 were entered into the final regression model in sequence as presented. The following variables, which were significant by univariate analysis, were excluded during the forward stepwise procedure: sex, educational level, diverting attention to cope with pain, special training for the last job position, and performed a physically highly demanding job at some time during working career. Overall, the model showed that coping styles and disease related, sociodemographic, and work related factors contributed substantially to withdrawal from the labour force.

When the proportion of patients who had withdrawn from the labour force was calculated in terms of each factor, it emerged that pacing to cope with limitations accounted for 73% of the withdrawals from the labour force. On the other hand, it was found that 73% of the withdrawals could still have been employed if technical or ergonomic adjustments of the workplace had been arranged. Other important factors increasing the risk were often/very often seeking creative solutions to cope with limitations (36%), a high disease activity (32%), older age (32%), and insufficient support from colleagues or management (29%). Other important factors, which reduced the risk, were working in a large company (34%) and frequent/very frequent acceptance to cope with dependency (32%).

In subsample analyses of patients who had withdrawn from the labour force within five years before the study and of patients with five or fewer years of disease, the same factors were identified. The following variables were stronger in the subsample of patients with five or fewer years of disease compared with patients in the full sample or in the subsample of patients withdrawn within five years before the study: pacing to cope with limitations, disease activity, attitude of the management, and access to the workplace.

DISCUSSION

Despite the fact that the overall participation in the labour force of the patients with AS under study was lower compared with the general population,⁴ the present study indicates that the probability of participating in the Dutch labour force is already reduced in patients with AS of relatively short disease duration, and participation in the labour force is not additionally influenced by duration of disease. By contrast with our study, most other reports on participation in the labour force and disease duration among patients with AS deal with groups with a work history, and found employment rates which still ranged from 79% to 85% after 20 years of AS, and from 53% to 81% after 30 years of AS.^20–22 Our study also considered patients without a work history and controlled for the common risk factors of participation in the labour force: age and sex. Therefore, our results give a deeper understanding of the position in the labour market of patients with AS among the general working population in relation to their duration of disease.

We report that sociodemographic factors, coping styles, disease related factors, and various work related factors substantially contribute to withdrawal from the labour force in patients with AS. In a previous study we already identified coping as an important factor related to work among patients with AS.⁷ In this study, we further explored behavioural coping simultaneously with other factors. Pacing to cope with limitations—that is, adapting the level of activity indicated by “I take more time for my activities”, “I stop doing my activities in time”, or “I take full account of my limitations”—was the major positively correlated factor. This implies that a passive way of coping with the limitations is associated with withdrawing from the labour force. Moreover, we found that five of the eight different coping styles assessed in this study significantly contributed to withdrawal from the labour force. With the exception of our previous study, we know of no earlier study that has covered the effect of behavioural coping in relation to employment in patients with AS. A recent study by Barlow et al investigated cognitive coping in relation to work in patients with AS.⁸ In fact, for patients with rheumatic diseases in general, only a few studies have examined the independent effect of behavioural coping on outcome of illness, and only two studies in rheumatoid arthritis found that patients use behavioural coping strategies when dealing with problems at work.^23–26 Earlier studies indicated that patients with AS used adaptive coping mechanisms to cope with stressful situations and that these patients were comparable with patients with other rheumatic diseases in the use of active or passive coping mechanisms.^27,28 Therefore, our findings in patients with AS stresses the importance of behavioural coping in relation to employment in patients with rheumatic diseases in general and in patients with AS in particular.

The major employment factor negatively associated with withdrawal from the labour force was technical or ergonomic adjustment of the workplace. One earlier study by Fellman et al found that 62% of patients with AS had pain when working which was induced by a wrong or monotonous posture, long working times, and unfavourable climatic factors at work.²⁹ Guillemin et al found that exposure to cold conditions and prolonged standing increased the risk of long term work disability.⁶ Yelin et al, on the other hand, found no relation between getting equipment and the improvement of employment status of persons with musculoskeletal conditions. Compared with our study, Yelin et al found a relatively low frequency of people getting special equipment (5.3%) in a relatively old population with various musculoskeletal conditions.³⁰ On the basis of those earlier studies and our study, we conclude that the ergonomic and climatic conditions at the workplaces of patients with AS specifically should be analysed at an early stage and adjusted to the needs of these patients to prevent problems at work.

Other major work related factors with negative correlations were working in a large company and additional job training after diagnosis, whereas insufficient support from colleagues or management, reduced transportation mobility at the workplace, difficult access to the workplace, and negative attitude of the management were positively correlated. Only a few of these factors have been previously investigated in a multivariate context among patients with AS.⁵ In line with our finding, vocational counselling and job training had already been identified as a protective factor against long term disability by Guillemin et al.⁶

It should be pointed out that our study has some methodological limitations. Data were collected by a self administered questionnaire and we did not validate these data using objective measures. However, the questionnaire included several well validated instruments for assessing disease activity and outcome independently of employment status.^10–17 So it is not likely that our results were substantially and systematically affected in any direction by differential misclassification; in other words there was no systematic difference in rating of disease activity and outcome between patients who had withdrawn and patients who still had a paid job.

Although data about employment were collected retrospectively, we think that considerable recall bias is less likely, because we found that work related factors were equally important in a subsample of patients who had recently withdrawn from the labour force.

With respect to coping, the cross sectional design of our study does not allow us to state that the explanatory variables have a causal relationship with withdrawal from the labour force. Our study does not rule out the competing explanation that withdrawal from the labour force invokes passive coping styles such as “pacing” as a way of coping with limitations or that behavioural coping and withdrawal from the labour force are reciprocally interrelated. On the other hand, research on coping in AS found that the coping mechanisms used did not vary with the duration of AS,¹ meaning that patients with AS may develop a coping style early in the illness and use their favourite strategies to deal with stressful situations. A more longitudinal approach is needed to assess whether the association between coping and withdrawal from the labour force in AS is one of effect rather than cause.

To provide information on the extent to which withdrawal from the labour force is due to a specific factor, and to estimate potential maximum effect of intervention programmes in the light of these factors, we calculated attributable and preventable fractions. Application of this technique is relatively new in this field of research. It is only found in two other studies in patients with RA.^24,31 Considering our results with respect to, for instance, pacing to cope with limitations (accounting for 73% of the withdrawals) and disease activity (accounting for 32% of the withdrawals), it can be calculated that the sum of the fractions exceeds 100%. Despite being aware that any withdrawal from the labour force could be caused or prevented by various factors, there might be a tendency to think that the sum of these fractions should be 100%. In fact, separate components of the underlying mechanism are not mutually exclusive. Therefore, the upper limit of the total of all fractions of all the components is not 100% but infinity. Only a fraction of withdrawal from the labour force attributable to or preventable from a single component cannot exceed 100%.¹⁹ This is also the case in our study.

In summary, our findings show that it is important to consider sociodemographic factors, disease related factors, behavioural coping styles, and work related factors in both research and vocational management of patients with AS. Research on the subject should give more attention to the study of behavioural coping and attitudes towards work depending on the nature and conditions of the work and on the social work environment and their effects on employment. Therefore, we emphasise that developers of intervention programmes dealing with work problems in patients with AS should consider and evaluate the involvement of a multidisciplinary team of professionals with medical, paramedical, psychosocial, and occupational health expertise. Furthermore, we think that cooperation from employers and from colleagues is important to achieve maximum reduction in early withdrawal from the labour force among these patients. In practice, patients should be encouraged to remain active despite the stressors of their disease. Moreover, patients should be encouraged and guided to recognise and develop their talents, and if possible, they should extend their education. In general, companies should be encouraged to promote positive attitudes towards people with disabilities.