Article Text

Extended report
Impact of ankylosing spondylitis on sick leave, presenteeism and unpaid productivity, and estimation of the societal cost
  1. Annelies Boonen1,2,
  2. Tjinta Brinkhuizen3,
  3. Robert Landewé1,2,
  4. Désirée van der Heijde4,
  5. Johan L Severens2,5
  1. 1Department of Internal Medicine, Division of Rheumatology, Maastricht University Medical Center, Maastricht, The Netherlands
  2. 2CAPHRI School of Public Health and Primary Care, Maastricht University, Maastricht, The Netherlands
  3. 3Faculty of Medicine, University of Maastricht, Maastricht, The Netherlands
  4. 4Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
  5. 5Department of Health, Organisation Policy and Economics, University Maastricht, Maastricht, The Netherlands
  1. Correspondence to Dr Annelies Boonen, Department of Internal Medicine, Division of Rheumatology, Maastricht University Medical Center, P Debyelaan 25, Maastricht 6229HX, The Netherlands; a.boonen{at}


Aim To describe the influence of ankylosing apondylitis (AS) on sick leave, presenteeism and unpaid work restrictions and to estimate related productivity costs.

Methods 142 consecutive and unselected patients with AS under the care of rheumatologists participated in a longitudinal observational study and completed the Health and Labour Questionnaire (HLQ) assessing disease-related sick leave, presenteeism and restrictions in unpaid work over the previous 2 weeks. Logistic regressions explored which explanatory variables were associated with work outcome. Productivity loss was valued in monetary terms.

Results Among 72 patients in paid employment, 12% had sick leave over a period of 2 weeks and 53% experienced an adverse influence of AS on work productivity while at work. Over this period they reported on average of 5.8 h sick leave and 2.4 inefficient working hours, for which they estimated an extra 1.9 h were needed to complete unfinished work. Among all patients (n=137), 71% had experienced restrictions in unpaid work during the previous 2 weeks with 42% needing help for these tasks for an average of 8 h. The annual production costs for the total group were €1451 (95% CI 425 to 2742) per patient for sick leave, €967 (95% CI 503 to 1496) to compensate for hours worked inefficiently while at work and €1930 (95% CI 1404 to 2471) to substitute loss of unpaid work production.

Conclusion Patients with AS not only have substantial sick leave but also experience restrictions while being at work and when performing unpaid tasks. Limitations in physical functioning are strongly associated with work restrictions. Societal costs of formal and informal care are comparable with the costs of sick leave and presenteeism combined.

Statistics from

Request Permissions

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.


One of the consequences of ankylosing spondylitis (AS) is restrictions in paid and unpaid work. The chronic nature of the disease and the start of first symptoms at an early age contribute to the relevance of worker participation. The restrictions experienced are important for patients but, because of the costs of loss in productivity, also for society.1

Loss of productivity in paid work concerns absenteeism, which is the temporary (such as sick leave) or permanent (such as work disability) time missed from work, but also presenteeism, which is the impaired performance or productivity while at work. In studies that compared work participation in patients with AS with the general population, employment was slightly decreased, especially in men, but (partial) work disability was remarkably increased.2,,4 Interestingly, employment and work disability varied across countries.4 5 Evidence on sick leave in comparison with the general population is scarce. One study showed that about 50% of patients annually have an episode of sick leave and that the number of days of sick leave per year was higher than in the general population.5 Although two studies explored presenteeism in patients with musculoskeletal diseases, data specific for AS are not available.6 7

The impact of the disease on the ability to perform unpaid tasks was studied in interviews with a convenience sample of 111 patients with AS8; 86% experienced some restrictions in housework, 49% in shopping and 31% in preparing meals. A review of four ‘cost of illness’ studies showed that between 14% and 20% of patients needed help from family or friends (a proxy for unpaid productivity loss), which accumulated to 29–44 h/year. In addition, 14–15% had paid household help for 12–17 h/year.9

Accepting that restrictions in paid and unpaid work are important for the patient as well as the societal perspective, we need valid instruments and a common theoretical framework to assess these effects and estimate societal costs. However, there is no consensus on such an approach.10 11 Most instruments are not fully validated and only a few allow quantification of productivity loss and calculation of the associated productivity costs.12 The Health and Labour Questionnaire (HLQ) was developed to measure loss of paid and unpaid productivity and to estimate productivity costs.13 In this study we applied the HLQ in a group of unselected patients under the care of a rheumatologist to better understand the impact of AS on participation in paid and unpaid work.



One hundred and forty-two patients who fulfilled the modified New York criteria constituted the study sample.14 The patients were part of an unselected prevalence cohort with longitudinal follow-up of patients with AS under the care of a rheumatologist (OASIS cohort). A description of patient recruitment has been published elsewhere.15 For the present analyses, data for the third year of the survey (1999–2000) were used since in this year the HLQ was added.


Patients completed questionnaires addressing sociodemographic factors and disease-related variables. Sociodemographic factors included sex, age, marital status, family situation, education, profession and work status. In the Dutch Social Security system a person is considered work-disabled if, as a consequence of disease, he/she has a reduced capacity to earn the same income as a person with a similar education and work experience but without disease.16 Work disability between 80% and 100% is considered full work disability; all other levels are partial work disability, in which case subjects can continue in a (part-time) paid job. AS-specific function and disease activity were measured by the Bath AS Functional Index (BASFI)17 and the Bath AS Disease Activity Index (BASDAI),18 respectively. To reflect structural damage, the modified Stokes AS Spinal Score (mSASSS) for the second year of the cohort evaluation was included.19

Patients also completed the HLQ, a validated questionnaire with four modules assessing qualitative and quantitative aspects of paid and unpaid work in the last 2 weeks.13 Patients were asked to report on restrictions as a consequence of AS (including AS-related disease), except for the days of sick leave where patients were asked to report separately sick leave because of AS and for health reasons other than AS. Module I assesses the number of (half) working days absent at paid work. In cases where the episode of sick leave exceeded 2 weeks, the patient had to state the total duration of the sick leave. Module II describes productivity losses while being at work, both quantitatively and qualitatively. Module III assesses the number of hours spent on unpaid work per week for different areas of unpaid work tasks. Module IV asks, for each of the unpaid tasks, whether subjects were hindered (yes/no) or not able (yes/no) to perform the tasks because of disease. In addition, the weekly number of hours that other persons (family, friends, private paid help or professional help) had to take over any task was recorded.

Statistical analyses

The sociodemographics, clinical characteristics and results of each module of the HLQ are presented using descriptive statistics. The time of lost paid productivity (modules I and II of the HLQ) and the time needed for help with unpaid tasks (module IV) were used to estimate the productivity costs.

The results are always presented for the patients who had completed the specific question. The description of sick leave and presenteeism is limited to patients with (part-time) paid work. To reflect the number of hours worked inefficiently while at work, self-rated efficiency (visual analogue scale (VAS) from 0 (completely inefficient) to 10 (as efficient as normal)) is related to the number of hours worked with AS-related discomfort ((10 − VAS efficiency/10) × hours worked with discomfort). To compare sick leave and time spent at unpaid tasks with the general Dutch population, the electronic databank on social and economic data of Statistics Netherlands was consulted.

Costs over 2 weeks are extrapolated to yearly costs and averaged over all patients in the sample. Since the costs were skewed, variability is represented by the bootstrapped 95% CI after 5000 iterations.20 The costs of sick leave (absenteeism) are calculated based on the friction method, which considers that productivity loss of absenteeism occurs only in the friction period which is the time needed to replace a person (3 months at the time of this study). To convert (half) days of absence into hours of absence, it was assumed that a full work day represents 8 h. The costs of presenteeism were based on the extra hours of work patients estimated are needed to compensate for the inefficient hours. The costing method for unpaid production followed the replacement theory, assuming the hours other persons provide help are a proxy for tasks not performed by the patient.10 Consistent with the friction cost method, the cost of 1 h of paid production is valued by the ‘added value of productivity’ per hour corrected by an elasticity factor of 0.8 and ranges from €21.13 to €50.36/h according to age categories and sex (see online Appendix I). The cost of 1 h of unpaid production is valued by the hourly market cost to substitute tasks which is consistent with the replacement method (see online Appendix II).21 All costs were adjusted for differential timing to values for 2006 using consumer price indices.

To determine which variables were associated with restrictions in productivity, logistic regression analyses were performed with sick leave (yes/no), presenteeism (work adversely influenced by AS; yes/no), restrictions in unpaid work (yes/no) and need for help for unpaid tasks (yes/no) as dependent variables. When sick leave and presenteeism were the outcomes, the sample was limited to those with paid work and the independent variables were age, disease duration (diagnosis), sex, educational level (≤12 years or more), profession (manual vs non-manual), having a partial work disability, BASFI, BASDAI and mSASSS. When unpaid work was the dependent variable, the total group was considered and the independent variables were age, disease duration (diagnosis), sex, educational level (≤12 years or more), having or not having paid work, BASFI, BASDAI and mSASSS. Variables that were univariate associated with the outcome, while controlling for age, gender and disease duration, were included in the final model. Bootstraps were performed in Excel and all other analyses in SPSS 16.0.


Patients and work status

A total of 142 patients with AS participated in the third year of the survey. The sociodemographic characteristics of the sample are shown in table 1. At that time, none of the patients was being treated with biological disease-modifying antirheumatic drugs (DMARDs). Seven of 142 patients did not complete the questions on work status and paid work restrictions. They more frequently (but not significantly) had a higher educational level (33% vs 19%) and less frequently did non-manual jobs (40% vs 60%), had a lower BASDAI (2.7 vs 3.8) and better BASFI (2.3 vs 3.9). Five patients did not complete the questions on unpaid work restrictions. They were all male, more frequently had a paid job (80% vs 51%) and tended to have a (non-significant) lower BASDAI (2.6 vs 3.9), better BASFI (2.6 vs 3.8) and less radiographic damage (7.4 vs 14.7). Of the 135 who provided information on their working status, 72 (53%) had paid work (70% of male patients and 35% of female patients). The mean (SD) hours worked per week were 32.8 (10.0) and 26.5 (9.9) for men and women, respectively. Only 6% of all patients reported housekeeping as their main work status.

Table 1

Demographic characteristics of the total group of AS patients in the study

Module I: Sick leave

Descriptives are shown in table 2; 73% of all episodes of sick leave and 80% of all days of sick leave were attributable to AS. Overall, 9.5% of official working hours were lost due to overall sick leave and 7.2% to AS-related sick leave.

Table 2

Absenteeism and presenteeism in patients with ankylosing spondylitis (AS) with paid work

The only variable independently associated with episodes of sick leave was BASFI (table 3). At the time of the study, 6.6% of working persons in the Netherlands had at least one episode of sick leave over a period of 1 year compared with 15.9% over a period of 2 weeks in our sample. The friction costs in the working patients were €3689 (95% CI 1546 to 6266) per patient per year for all causes and €2923 (95% CI 922 to 5406) per patient per year for AS-related sick leave. Averaged over the total cohort, this was €1832 (95% CI 728 to 3198) and €1451 (95% CI 425 to 2742) per patient per year, respectively (table 5).

Table 3

Variables associated in multivariate logistic regressions with sick leave, presenteeism, impact on unpaid work and need for formal or informal care as a consequence of ankylosing spondylitis (AS)

Table 5

Yearly productivity costs because of ankylosing spondylitis (AS)-related absenteeism, presenteeism and need for help in unpaid work for the total group

Module II: Presenteeism

Seventy-one per cent of patients experienced AS-related discomfort while at work during at least 1 day and, on average, patients worked with discomfort during 7.23 h/2 weeks. About 53% of working patients reported that discomfort adversely influenced their work (table 2). The type of impediments at work that were most frequently encountered can be seen in table 2. Patients estimated their work efficiency during hours worked as 7.7 on a scale of 0–10 (where 10 is normal). When adjusting the hours worked with discomfort for the work efficiency score, this would result in 2.4 inefficient hours/2 weeks. Twenty-two per cent of patients considered that extra work would be needed to compensate for the production loss due to inefficient working hours. An extra 1.91 h/2 weeks was perceived necessary to compensate this loss. Variables associated with presenteeism (work adversely influenced) were BASDAI and BASFI (table 3). All patients with an episode of sick leave experienced presenteeism in the period before or after sick leave.

Accepting that extra working hours to compensate for productivity loss reflect the costs of presenteeism, these would be €1947 (95% CI 1052 to 2976) per working patient per year. Averaged over the total group, costs were €967 (95% CI 503 to 1496) per patient per year (table 5).

Modules III and IV: Unpaid work

Men with AS work 17.6 (14.9) h/week in unpaid tasks and women with AS work 32.4 (22.2) h/week. Compared with the general population, time spent was reduced for all tasks except for odd jobs and child care (table 4). When expressing time spent by patients as a percentage of time spent by the general Dutch population, the reductions were more pronounced in men (40%) than in women (12%). Within patients, between 9.6% and 61.9% of all patients mentioned they were hindered or even had not performed planned tasks (table 4). For all these tasks, more women than men complained of being hindered. In agreement with these findings, more women than men (55.8% vs 39.4%) reported needing help from others. Patients received help mostly from their children or partner and from family and friends. Variables associated with restrictions in unpaid work and need for help were female gender and BASFI (table 3). Limiting regressions to patients with paid work (n=72), having presenteeism or sick leave was associated with more restrictions in unpaid work and more need for help in univariate analyses, but this could be explained entirely by the effect of BASFI and/or BASDAI.

Table 4

Overview of unpaid work for the total group and by gender

Following the replacement method, the unpaid productivity costs are €1930 (95% CI 464 to 2471) and are driven by help from partner or children (table 5).


Although sick leave in AS has been described previously, this study is the first to address the problems experienced by patients with AS while at work and when doing unpaid tasks. While 11.6% of patients with paid work had an episode of AS-related sick leave in the previous 2 weeks (absenteeism), no fewer than 52.8% felt their work was adversely influenced by AS (presenteeism). In the entire sample, 71% experienced restrictions in different types of unpaid tasks. Limitations in physical function (BASFI) were consistently associated with work outcome. It is well known that disease activity (BASDAI) is a major determinant of function (BASFI). However, only when problems while at paid work are considered (presenteeism) does disease activity has a role which is independent of its effect on physical function. Interestingly, the inclusion of an objective measure of radiographic damage (mSASSS) did not change the results. The self-report function (BASFI) is therefore extremely important when understanding work outcome. The extent to which coping explains part of the self-reported function cannot be answered in this study. When comparing those with or without paid employment, fewer employed patients were involved in unpaid tasks (96% vs 100%) and specifically in housekeeping tasks (79% vs 94%). However, they did not report more or fewer restrictions in unpaid tasks or need for help. Interestingly, employed patients with sick leave or presenteeism experienced more restrictions in unpaid tasks than those without sick leave or presenteeism, showing that the different domains of participation restrictions are related. When comparing genders, employed women had no more sick leave or presenteeism than men. However, women experienced more restrictions in unpaid work and needed more help from caregivers. Since women spend more time than men in most of the unpaid tasks, they are more likely to experience restrictions in unpaid tasks. Furthermore, women and men may have different expectations on the quality at which the unpaid tasks should be done and therefore may feel more restricted.

The interpretation of restrictions in paid and unpaid work is more meaningful when comparisons with the general population can be made, but this was not possible for all outcomes. Sick leave occurred more frequently than in the general population. Interestingly, patients spent less time in housekeeping but more time in performing odd jobs than the general population. It may be that it is easier to organise help for housekeeping tasks than asking for help to perform odd jobs, which they would rather perform themselves at the expense of additional time and effort. Not surprisingly, men with AS find it easier to reduce housekeeping time than women with AS.

Although in the total sample the time for which patients with AS needed help for unpaid tasks (8 h/2 weeks) exceeded the time patients were absent or were inefficient at paid work (3.34 h/2 weeks), the cost of loss of unpaid production was comparable to the cost of loss of paid production. This is mainly attributable to the lower monetary value attached to (in)formal care compared with paid production. The HLQ follows the friction cost method to estimate the societal costs of absence from paid work. Although theoretically sound at the societal level, this does not reflect the burden of disease associated with work disability, nor does it capture the gains in productivity once a disabled person engages again in paid work.22 Furthermore, to estimate the costs of presenteeism,12 the HLQ assumes that not all inefficient hours result in production costs since patients reported that they worked 2.4 h inefficiently but estimated that only 1.9 h would be needed to compensate. The HLQ does not ask whether such compensation took place in real life, which could well be the case, further reducing the productivity costs.23 24 One study found that one-third of absenteeism is compensated, but this has not been studied for presenteeism.24 Alternatively, some health economists stipulate that the absence of one worker may affect the productivity of another worker, thus amplifying production losses.25 The lower productivity costs among women with AS can be explained by the lower work participation compared with men (35% vs 70%). The data on differences in gender in paid work participation should be interpreted with care in view of the low number of working women (n=14/72).

The HLQ asked about the impact of AS on paid and unpaid worker participation. Certainly, it cannot simply be assumed that this impact is additional to the impact of the ‘common’ health problems on productivity.5 Large case–control studies recording AS-specific and overall health-related restrictions could solve this question. Since we found that restrictions attributed to AS already exceeded the restrictions reported in the general population (sick leave episodes and time spending), the impact of AS alone is already considerable. Also, the positive effect on paid and unpaid work in randomised controlled trials comparing a tumour necrosis factor blocker with placebo showed that this is at least partially related to the effects of AS on (un)paid work.26

The HLQ has recently been used in patients with rheumatoid arthritis who were eligible for treatment with a biological DMARD. In patients in paid work (n=145), 22.1% reported having had sick leave in the previous 2 weeks (compared with 11.6% in our study), 68% experienced at least mild restrictions while at work (compared with 53% in our study), VAS efficiency was 8.8 (compared with 7.7 in our study) and expected additional hours needed to complete unfinished work were 2.7 h/2 weeks (compared with 1.9 in our study). In the total group (n=389), 18% of patients with rheumatoid arthritis needed help from formal caregivers and 52% from unpaid caregivers, compared with 14% and 34%, respectively, in our study of patients with AS.27 It is important to realise that our cohort is unselected and had differences in age, gender and professions compared with the patients with rheumatoid arthritis.

In summary, patients with AS not only have substantial sick leave but also experience restrictions while being at work and when performing unpaid tasks. This adds to the personal and societal impact of the disease. Limitations in physical functioning are strongly associated with work restrictions.


Supplementary materials

  • Web Only Data ard.2009.116764

    Files in this Data Supplement:


  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the METC, Maastricht, The Netherlands and patients gave informed consent.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Patient consent Obtained.