Article Text
PDF
Abstract
Objectives To examine predictors of work ability gain and loss after anti-tumour necrosis factor (TNF) start, respectively, in working-age patients with rheumatoid arthritis (RA) with a special focus on disease duration.
Methods Patients with RA, aged 19–62 years, starting their first TNF inhibitor 2006–2009 with full work ability (0 sick leave/disability pension days during 3 months before bio-start; n=1048) or no work ability (90 days; n=753) were identified in the Swedish biologics register (Anti-Rheumatic Treatment In Sweden, ARTIS) and sick leave/disability pension days retrieved from the Social Insurance Agency. Outcome was defined as work ability gain ≥50% for patients without work ability at bio-start and work ability loss ≥50% for patients with full work ability, and survival analyses conducted. Baseline predictors including disease duration, age, sex, education level, employment, Health Assessment Questionnaire, Disease Activity Score 28 and relevant comorbidities were estimated using Cox regression.
Results During 3 years after anti-TNF start, the probability of regaining work ability for totally work-disabled patients was 35% for those with disease duration <5 years and 14% for disease duration ≥5 years (adjusted HR 2.1 (95% CI 1.4 to 3.2)). For patients with full work ability at bio-start, disease duration did not predict work ability loss. Baseline disability pension was also a strong predictor of work ability gain after treatment start.
Conclusions A substantial proportion of work-disabled patients with RA who start anti-TNF therapy regain work ability. Those initiating treatment within 5 years of symptom onset have a more than doubled 3-year probability of regaining work ability compared with later treatment starts. This effect seems largely due to the impact of disease duration on disability pension status.
- Anti-TNF
- Rheumatoid Arthritis
- DMARDs (biologic)
- Social work
- Outcomes research
Statistics from Altmetric.com
Introduction
Rheumatoid arthritis (RA) is the most common rheumatic inflammatory disease with an estimated prevalence of 0.6–0.8%.1 ,2 Without effective treatment, the disease usually leads to deterioration of physical function and work disability.3
The introduction of tumour necrosis factor (TNF) antagonists has played a major role with better control of symptoms, halting of radiological progression and improvement of functional capacity.4 ,5 The impact of these effects on work ability has not been as well studied, although there is emerging evidence that this treatment could affect work ability in a favourable way compared with treatment with a non-biological disease-modifying anti-rheumatic drug (DMARD), and in patients who have responded poorly to an adequate course of non-biological DMARDs.6–8
As work ability is a domain with high priority for patients with RA, both financially and socially,9 data on which patients would benefit the most from anti-TNF therapy from a work perspective could have important implications for treatment decisions, but also for resource allocations. In addition, it could indicate groups of patients with a poorer work prognosis who might benefit from early non-pharmacological interventions. Until now, such reports have been scarce with a few findings suggesting that short disease duration at anti-TNF start could have a favourable effect on future ability to work.7 ,10 Higher Health Assessment Questionnaire (HAQ) and older age at bio-start have been associated with a less favourable work prognosis,11–13 but larger predictor studies are needed.
The objective of this study was to examine predictors of work ability gain and loss, respectively, in working-age patients with RA during the first 3 years after starting anti-TNF therapy, with a special focus on the impact of disease duration at treatment start.
Methods
Setting
In 2009, the Swedish population was 9.3 million, 60% being of working age (19–64 years). The Swedish healthcare system offers universal access, and patients with RA are generally diagnosed and treated by rheumatologists for their disease and by general practitioners and other specialists for their comorbidities. The official retirement age is 65 years, but workers have the right to enter old-age pension at 61 or stay at work until 67 years. Prescription drugs are provided free of charge above a threshold of SEK 1800 annually (≈€190/US$260).
The Swedish social insurance system
All residents in Sweden aged 16–64 years can be granted economic compensation from the Social Insurance Agency in case of sickness, disability or injury. The compensation can take the form of sick leave or disability pension and it is possible to have both compensation forms simultaneously, though never exceeding 100% together. Only periods ≥14 days are contained in the register (for detailed information on the Swedish social insurance system, see online supplementary text).
supplementary text
Patient cohort
Patients with RA, 19–62 years, starting treatment with their first TNF inhibitor 2006–2009 were identified in the Swedish Rheumatology Quality Register. This is a national register comprising both the Swedish biologics register (Anti-Rheumatic Treatment In Sweden (ARTIS)) set up in 1999, covering 87–95% of all biologics-treated patients with RA in Sweden, and the register for follow-up of incident RA (according to 1987 American College of Rheumatology criteria), started in the mid-1990s.14 ,15 From this cohort, we selected patients with no work ability at bio-start (90 days with sick leave or disability pension during the three months prior to bio-start) and patients with full work ability (0 days with sick leave or disability pension during the three months before bio-start). The patient inclusion is described in a flow chart (see online supplementary figure S1). Sixty-two years was chosen as upper age limit to avoid that patients enter old-age pension during follow-up.
Data were linked to the Social Insurance Agency Register including day-level data on sick leave and disability pension (2005–2010), and the Longitudinal Integration Database for Health Insurance and Labor Market Studies including data on education level and unemployment benefits by calendar year (2005–2009). Furthermore, data on comorbidities were retrieved from inpatient and outpatient care visits registered in the National Patient Register (2001–2009), and from drug treatment registered in the nationwide Prescribed Drug Register (2005–2009; online supplementary table S1).
Outcome
For patients with full work ability at bio-start, the outcome measure was time to loss of either ≥50% or 100% work ability, respectively, during a month (≥15 net days with sick leave or disability pension out of 30; 30 days out of 30). For patients with no work ability at bio-start, the outcome measure was time to regain of either ≥50% or 100% work ability, respectively, during a month (≤15 net days with sick leave or disability pension out of 30; 0 days out of 30). A month was chosen as window to be long enough to reflect sustained changes but short enough to do so with good resolution and in close connection to when changes appear.
Net days of sick leave and disability pension were used as primary outcome because both compensation forms may be complete (100%) or partial, in specific predefined steps (eg, 25%, 50% or 75%), and were calculated by multiplying the degree of compensation with the number of days. Since patients often move from sick leave to disability pension and the two compensation forms can also coexist, they were analysed jointly, as has been recommended.16 The maximum number of days was set as 30 per month.
Follow-up
Patients were followed from day of anti-TNF start until time of event, death, emigration or end of follow-up (maximum 36 months), whichever came first.
Primary exposure
The primary variable investigated was disease duration at treatment start (defined as time from onset of RA symptoms to start of anti-TNF therapy). We also investigated disability pension status at bio-start in patients without initial work ability since it has been reported an important predictor in related conditions.17
Covariates
Covariates included known to drive work disability, comprised demographic variables (age, sex and education level as well as employment status),18 ,19 clinical variables (HAQ and Disease Activity Score 28 (DAS28)),18 ,19 treatments (non-biological DMARD(s), glucocorticoids) and comorbidities (depression and anxiety disorders, cardiovascular disease, malignancies, hypertension, diabetes and hip/knee replacement).20
Statistics
Statistical analyses were conducted using SPSS (V.20.0). Differences in baseline characteristics between work ability categories at bio-start as well as differences between groups of disease duration within each work ability category were assessed by one-way analysis of variance for continuous variables and χ2 tests for ordinal ones.
Survival analyses with Kaplan-Meier curves were conducted using Cox-regression models to assess baseline predictors. Independent variables included in the prediction models were chosen based on subject matter knowledge.18–20 The variables were first tested in univariate analysis and only those generating p values ≤0.1 in the analysis for work ability gain were included in the multivariate models, except for disease duration, sex, age at bio-start, HAQ and DAS28, which were always included. Furthermore, to assess the association between disease duration and disability pension status, logistic regression was used, including the same covariates as in the survival analyses.
Patient lacking data for any of the independent variables at bio-start were excluded from the multivariate Cox-regression and logistic regression models. Reported p values are two-sided and p<0.05 was considered statistically significant.
Results
We identified 753 patients with no work ability and 1048 with full work ability at bio-start. In the group with full work ability at bio-start, 512 patients (49%) had <5 years disease duration compared with 264 patients (35%) in the group with no work ability (p<0.001; table 1).
Baseline characteristics for patients with rheumatoid arthritis (RA) 18–62 years starting first anti-tumour necrosis factor (TNF) therapy 2006–2009 in Sweden
Among patients with no work ability at bio-start, those with <5 years disease duration were slightly younger than patients with longer disease duration (52 vs 54 years; p<0.001) and had substantially less full-time disability pension (49% vs 82%; p<0.001). They were more often taking methotrexate (75% vs 63%; p=0.001) and steroids (61% vs 49%; p=0.001) at bio-start, and had less often a history of hip or knee replacement (7% vs 20%; p<0.001).
In patients with full work ability at bio-start, those with <5 years disease duration were again younger than patients with ≥5 years disease duration (44 vs 47 years; p<0.001), they were more often on steroids (48% vs 36%; p<0.001) and had a lower prevalence of drug-treated hypertension (11% vs 16%; p=0.03) and previous hip or knee replacement (2% vs 7%; p<0.001).
Work gain in patients with no work ability at bio-start
Univariate analysis
In RA-patients with no work ability at bio-start and with disease duration <5 years, the probability of regaining ≥50% work ability was 35% during the first 3 years after treatment start, while the probability in patients with no work ability and ≥5 years disease duration was 14% (unadjusted HR 3.0 (95% CI 2.1 to 4.2)) (figure 1A and table 2). In sensitivity analyses, a less pronounced difference was seen already with a cut-off point of 2 years for disease duration (see online supplementary table S2). The corresponding probabilities of regaining 100% work ability were 24% for patients with disease duration <5 years and 8% for those with ≥5 years disease duration (unadjusted HR 3.7 (95% CI 2.3 to 5.7)) (see figure 1C and online supplementary table S3).
Predictors of work ability gain ≥50% for patients with rheumatoid arthritis (RA) with no work ability at bio-start* assessed by Cox regression analysis during 3 years after bio-start for patients aged 18–62 years starting treatment 2006–2009 in Sweden
Cumulative probability of work ability gain ≥50% and 100% for Swedish patients with rheumatoid arthritis (RA) (19–62 years) with no baseline work ability from start of first anti-tumour necrosis factor (TNF) therapy (2006–2009) and 36 months forth, stratified on disease duration at bio-start (A and C) and disability pension status at bio-start (B and D).
Patients who were not on full-time disability pension (but instead had full-time sick leave or part-time sick leave in combination with part-time disability pension) had a cumulative probability of regaining ≥50% work ability of 56% during the first 3 years after treatment start, while it was only 4% for patients who were on full-time disability pension (unadjusted HR 18.4 (95% CI 11.2 to 30.2)) (see figure 1B and table 2). The corresponding probabilities of regaining 100% work ability were 38% and 2% (unadjusted HR 20.2 (95% CI 10.5 to 39.1)) (see figure 1D and online supplementary table S3).
Multivariate analysis
Disease duration <5 years, male sex, younger age, history of hip/knee replacement and unemployment at bio-start were predictors of work ability gain ≥50% during the first 3 years after starting first anti-TNF therapy in multivariate analysis. When adding baseline disability pension status to the model, only disability pension status and younger age remained significant predictors of work ability gain (adjusted HR 14.3 (95% CI 7.7 to 26.5) for full/partial sick leave vs full disability pension and 1.5 (95% CI 1.2 to 1.8) per 10 years decrease in age) (table 2).
Disease duration and disability pension status
Exploring the relationship between disease duration and disability pension status in RA-patients with no work ability at bio-start, a close association was found: the longer the disease duration at anti-TNF start, the higher the share of patients with full-time disability pension. In patients starting their first anti-TNF treatment within 1 year of symptom onset, 38% had full-time disability pension, while 87% of those starting after 15 years or more were on full-time disability pension (figure 2). In multivariate logistic regression, the adjusted OR of having full-time disability pension between patients with ≥5 years and <5 years disease duration at bio-start was 4.9 (95% CI 3.2 to 7.5; online supplementary table S4).
Disability pension/sick leave status at start of first anti-tumour necrosis factor (TNF) therapy (2006–2009) in relation to disease duration for Swedish patients with rheumatoid arthritis (19–62 years) without work ability at bio-start.
Work loss in patients with full work ability at bio-start
In patients with RA with full work ability at bio-start, there were no differences in work ability loss over 3 years after treatment start between patients with <5 years versus ≥5 years disease duration. The cumulative probabilities of ≥50% work ability loss were 28% and 25%, respectively (adjusted HR 0.9 (95% CI 0.7 to 1.2)) (figure 3 and table 3), while cumulative probabilities of 100% work ability loss were 10% and 9%, respectively (adjusted HR 1.3 (95% CI 0.8 to 2.0)) (see figure 3 and online supplementary table S5).
Predictors of work ability loss ≥50% for patients with rheumatoid arthritis (RA) with full work ability at bio-start*, assessed by Cox regression analysis during 3 years after bio-start for patients aged 18–62 years starting treatment 2006–2009 in Sweden
Cumulative probability of work ability loss ≥50% (A) and 100% (B) for Swedish patients with rheumatoid arthritis (RA) (19–62 years) with full baseline work ability from start of first anti-tumour necrosis factor (TNF) therapy (2006–2009) and 36 months forth, stratified on disease duration at bio-start.
Concerning other predictors, a history of depression/anxiety disorder was associated with higher risk of losing work ability (adjusted HR 1.7 (95% CI 1.1 to 2.6)), which was also the case for hip/knee replacement (adjusted HR 2.2 (95% CI 1.3 to 3.8)) and unemployment (adjusted HR 1.7 (95% CI 1.1 to 2.6)). It was furthermore shown that lower HAQ scores and lower DAS28 values were associated with less risk of losing work ability (adjusted HR 0.6 (95% CI 0.4 to 0.9) for HAQ <1.0 vs HAQ ≥1.5 and adjusted HR 0.6 (95% CI 0.3 to 0.9) for DAS28≤3.2 vs >5.1).
Discussion
In this study, we found that the adjusted likelihood of returning to work for patients with RA lacking work ability at bio-start was more than doubled if starting treatment within 5 years of disease onset compared with later treatment starts. However, when making additional adjustment for disability pension status at bio-start, which in itself was a strong predictor, the effect of disease duration was no longer significant. Additional analyses though showed an almost fivefold increase in the risk of being on full-time disability pension for patients with disease duration ≥5 years at bio-start compared with patients starting treatment earlier. Once on full-time disability pension, very few patients regain any work ability.
This is an observational study and we cannot infer causality. However, these results suggest that disease duration at bio-start is important for the chance of regaining work ability in patients lacking this: the longer the disease duration, the higher the share of patients on disability pension and the lower the probability that anti-TNF therapy could bring patients back to work.
Among patients with full work ability at anti-TNF start, there was no difference in the risk of losing it between patients starting treatment with <5 years disease duration and those starting treatment later. This might reflect the findings in two recent Swedish studies showing a peak in work disability during the first years after RA diagnosis,18 ,21 and that after 5 years patients who are still capable of work represent a selected group less prone to lose work ability later on.
To the best of our knowledge, predictors of work ability gain and loss after bio-start have not been estimated separately in a structured way before, and interestingly we could show that predictors for the two outcomes were different.
Previous research
Our findings confirm results in an earlier Swedish study by some of us, which suggested that patients with RA starting anti-TNF therapy with <5 years disease duration have more reversibility in their work disability than patients starting treatment later in the disease course. However, that study had a strictly descriptive approach and was not designed to specifically analyse predictors of work ability.7 Concerning disease duration as a predictor of work ability loss our results are consistent with findings by Verstappen et al 11 and Wolfe et al,12 showing no such association. Allaire et al 10 though reported less employment loss in patients starting anti-TNF therapy within 11 years of RA onset, but their analysis used work status registered by self-report, subjected to non-response and recall bias.22
Our results concerning HAQ agree with previous research, demonstrating it to be an important predictor of different work outcomes after bio-start,11–13 which is also true for age,10–13 while results for sex as a predictor have been more mixed. Comparing predictors of work ability after bio-start in the current study with earlier studies on predictors of work ability after RA diagnosis, we conclude that higher HAQ scores, older age and lower education level seem to be predictors of a worse outcome in both these settings.19
Strengths and limitations
A strength of the current study was the objectively assessed data on sick leave and disability pension as well as data on relevant comorbidities from an independent source, eliminating non-response and recall bias,22 but also the population-based design. We could furthermore analyse sick leave and disability pension jointly, as recommended.16 Having access to data from the period preceding bio-start allowed for an important stratification on baseline work ability including its components.
The present study also has limitations. First, the information available on work disability is restricted to sick leave and disability pension representing absenteeism measures. We are aware of the aspects of work disability while at work (presenteeism), which according to a systematic review is also an important component of work disability,3 thus limiting our results to one side of the work disability problem.
Second, sick leave episodes ≤14 days were generally not captured in the Social Insurance Agency Register, unless appearing within 4 days of a previous episode. However, a recent Swedish study reported only 2% of sick leave episodes in patients with RA to be 8–14 days,23 and a Finnish study reported only 3% of sick leave episodes to be <10 days in patients with RA, constituting 0.2% of total days,16 indicating that underestimation is likely to be small. Furthermore, we have no data on the frequency of voluntary part-time work (around 20% in the general population).24 Nor did we have information on the causes of sick leave/disability pension, but some of us have previously shown that Swedish patients with RA starting anti-TNF therapy have around four times more absence days than matched population subjects, suggesting that direct and/or indirect effects of RA is the main absence driver. In contrast to previous studies, we also included major absenteeism-related comorbidities.
Third, earlier studies have suggested that there are important additional explanatory variables, such as working hours, physical job demand, work environment, financial situation and coping ability, which we did not have access to, that weakens our results.18 Furthermore, no data were available on obesity and smoking status, two strong predictors of disability pension in the general population,25 nor on radiographic erosions.
Fourth, data on sick leave and disability pension were available until October 2010, why patients starting treatment after October 2007 did not provide data for the entire 3-year follow-up period in the survival analyses. Adjustment was, however, made for year of bio-start in the Cox-regression models. Furthermore, the multivariate models had approximately 20% missing in ≥1 of the covariates, but main results were basically similar after multiple imputation (see online supplementary tables S6 and S7).
Finally, Sweden has a generous welfare system in an international perspective, potentially limiting generalisability, although a systematic review of productivity losses in RA found largely similar rates of work disability in the US and Northern Europe despite different social systems and study methodologies.3
Conclusions
RA-patients without work ability, who started anti-TNF treatment within 5 years of symptom onset, had a more than doubled probability of regaining work ability over the following 3 years compared with patients with longer disease duration at bio-start. This effect seemed to be channelled through the impact of disease duration on disability pension status, suggesting a potential for increasing return to work if anti-TNF therapy is started before patients enter full-time disability pension. Furthermore, patients without a history of hip-knee replacement at anti-TNF start had more work ability gain and less work ability loss, while patients with high HAQ and high DAS28 values had a higher risk of losing work ability.
Acknowledgments
The authors express their gratitude to all rheumatologists contributing data to the Swedish Rheumatology Quality Register.
References
Footnotes
Handling editor Tore K Kvien
Collaborators The ARTIS Study Group: J Askling (Karolinska Institutet), E Baecklund (Uppsala University), S Ernestam (Karolinska Institutet), N Feltelius (Medical Product Agency), H Forsblad-d'Elia (Sahlgrenska Academy), L Jacobsson (Sahlgrenska Academy), A Kastbom (Linköping University), L Klareskog (Karolinska Institutet), E Lindqvist (Lund University), S Rantapaa-Dahlqvist (Umeå University), C Turesson (Lund University), R van Vollenhoven (Karolinska Institutet).
Contributors TO, MN, IP, LTHJ, PG, JA and ME planned the study. JKE and MN provided the data. TO and MN performed the analysis and drafted the paper. All the other authors contributed to editing the draft for content and approved of the final version. All authors had full access to all the data (including statistical reports and tables) in the study and take responsibility for the integrity of the data and accuracy of the data analysis.
Funding This study was funded by unrestricted grants from Lund University and Region Skåne. For the maintenance of the Swedish Biologics Register (ARTIS), the Swedish Society for Rheumatology has received funding, independent of the conduct of these scientific analyses, from Merck Sharp and Dohme, Bristol-Myers Squibb, Pfizer, AbbVie, SOBI, Union Chimique Belge Pharma, AstraZeneca and Roche.
Competing interests JA has research agreements (mainly for commissioned safety monitoring in ARTIS) with UCB, Roche, Merck, Pfizer, Samsung, Lilly and Abbvie, and has received remuneration for an ad board organised by Lilly. JKE has participated in research projects fully or partly funded by Novo Nordisk and COMBINE Sweden and served as an external consultant to AbbVie. MN has participated in advisory boards for Pfizer (rheumatology) and Abbott (non-rheumatology) and in research projects fully or partly funded by Schering-Plough, AstraZeneca, Novo Nordisk, Pfizer and Roche (unrelated to the current work). LTHJ has received fees for speaking and consultancy from Pfizer, Novartis, Abbvie and Celgene. IFP has reported ownership of options in Ana Mar.
Ethics approval The Ethics Committee at Karolinska Institutet.
Provenance and peer review Not commissioned; externally peer reviewed.