BMJ Open Is Committed to Open Peer Review. As Part of This Commitment We Make the Peer Review History of Every Article We Publish Publicly Available

BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay- per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected]

http://bmjopen.bmj.com/ on September 29, 2021 by guest. Protected copyright. BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open

Predicting End of Work Disability After Cancer. A 7 years Follow-up Population-based Cohort Study Using Competing Risks Analysis

For peer review only Journal: BMJ Open

Manuscript ID bmjopen-2016-014094

Article Type: Research

Date Submitted by the Author: 30-Aug-2016

Complete List of Authors: Kiasuwa Mbengi, Régine; Scientific Insitute of Public Health, Belgian Cancer Centre ; Universite Libre de Bruxelles, Centre de Recherche en Santé Environnement et Santé au travail Nicolaie, Alina; Universiteit Gent, Statistical Department Goetghebeur, Els; Universiteit Gent, Applied mathematics, computer science and statistics Otter, Renee; Scientific Institute of Public Health, Belgian Cancer Centre Mortelmans, Katrien; KaMoCo Missinnne, Sarah; Scientific Insitute of Public Health, Belgian Cancer Centre Arbyn, Marc; Scientific Insitute of Public Health, Unit Cancer Epidemiology http://bmjopen.bmj.com/ Bouland, Catherine; Universite Libre de Bruxelles, Centre de Recherche en Santé Environnement et Santé au travail de Brouwer, Christophe; Universite Libre de Bruxelles, Centre de Recherche en Santé Environnement et Santé au travail

Primary Subject Epidemiology Heading:

Secondary Subject Heading: Public health on September 29, 2021 by guest. Protected copyright. Epidemiology < ONCOLOGY, ONCOLOGY, Organisation of health services < Keywords: HEALTH SERVICES ADMINISTRATION & MANAGEMENT

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 1 of 36 BMJ Open

1 2 3 4 1 Predicting End of Work Disability After Cancer. A 7-year Follow-up 5 2 Population-based Cohort Study Using Competing Risks Analysis. 6 7 8 3 R. Kiasuwa Mbengia,c, M.A Nicolae b, Els Goetghebeurb, R. Ottera, K. Mortlemansd, S. Missinne a, 9 4 M. Arbyne, C. Boulandc and C. de Brouwerc 10 11 5 a Belgian Cancer Centre, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 12 6 b Stat-Gent CRESCENDO, University of Ghent 13 7 c Research Centre for Environmental and Occupational Health, Brussels School of Public Health, Université Libre de Bruxelles (ESP-ULB) 8 d Dr. Katrien Mortelmans, PhD Consulting (KaMoCo) 14 9 e Unit Cancer Epidemiology, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 15 10 For peer review only 16 11 Corresponding author: Régine Kiasuwa Mbengi 17 12 [email protected] 18 13 14 rue Juliette Wytsman, 1050 Bruxelles - Belgium 19 14 Office: +32 2 642 57 65 20 15 Mobile: +32 479 3926 58 21 22 23 16 ABSTRACT 24 25 17 Objectives 26 18 The number of workers who are disabled due to cancer is increasing dramatically worldwide. One of the 27 19 main priorities is to preserve their quality of life (QoL) and the sustainability of financing systems for 28 20 sickness absence benefits. We have carried out this study in order to assess factors associated with the 29 21 return to ability to work after cancer-related work disability. This should help with planning of 30 22 rehabilitation needs and tailored programmes. 31 23 Participants 32 24 We conducted this register-based cohort study using individual data from the Belgian Disability http://bmjopen.bmj.com/ 33 25 Insurance. Data on 15, 543 socially insured Belgian people who entered into work disability due to cancer 34 26 in 2007-2011 were used. 35 36 27 Primary and secondary outcome measures 37 28 We estimated the time spent in disability and the cause-specific cumulative incidences of return to ability 38 29 to work stratified by age, gender, occupational class, year of entrance into work disability for 11 cancer 39 30 groups using Kaplan Meier analyses and Cox regression allowing for competing risks.

40 31 Results on September 29, 2021 by guest. Protected copyright. 41 32 The overall median time spent into work disability is 1.59 years (95%CI [1.52-1.66]), ranging from 0.75 42 33 to 4.98 across the 11 cancer groups. By the end of follow-up, more than one-third of work disabled cancer 43 34 survivors were able to RTW (35%). While larger proportions of women were able to RTW at the end of 44 35 follow-up, men who do RTW are able to do so sooner. Women, white-collar, younger and having 45 36 haematological, male genital or breast cancers were the most likely to be able to RTW. 46 37 Conclusion 47 38 Good prognostic factors for the ability to RTW are young age, being female, being a white-collar worker, 48 49 39 and having breast or haematological cancers. 50 40 Looking at our results together with the cancer incidence predictions up to 2025 has a high value for the 51 41 purposes of social security and rehabilitation planning and ascertaining patients’ perspectives. 52 42 53 43 Key words: sickness absence, cancer survivors, competing risks, predictive model, social inequalities 54 44 55 45 56 46 57 47 58 59 Version 30 août 2016 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 2 of 36

1 2 3 48 Strengths and limitations of this study: 4 49 • (Good) Generalizability: we used a population-based dataset without loss of follow-up; the 5 6 50 external validity is therefore barley not limited; 7 51 • Methods: we add competing risks to the traditional survival analysis to respect the complexity of 8 52 the outcomes which is still rarely performed in disability studies; 9 53 • High value of our results (if linked with cancer incidence predictions) for the planning of 10 54 rehabilitation needs of cancer patients, up to 2025; 11 55 • Identification of cancer survivors being at risk to be socio-economically excluded; 12 56 • We miss information on treatments and job demands: these could have helped to (1) precise the 13 57 risk profile and (2) tailor RTW interventions 14 58 15 For peer review only 16 59 17 60 BACKGROUND 18 19 61 The direct and indirect effects of work disability represent an important burden for people who are absent 20 21 62 due to sickness, their families and employers [1]. Long-term work disability may lead to social exclusion, 22 23 63 deprivation or economic insecurity [2], as well as poor health [3]. The negative impact of work disability 24 25 64 on both social and health status is of high importance for public health [4] but studies identifying those 26 27 65 cancer survivors who are at risk of experiencing long-term work disability and identifying the avoidable 28 29 30 66 proportion of work disability are lagging behind. 31 32 67 http://bmjopen.bmj.com/ 33 68 Work disability imposes significant costs on society [5,6] with up to 5% of GDP in Organisation for 34 35 36 69 Economic Co-operation and Development (OECD) countries being spent on disability benefits [5]. In 37 38 70 2010, the OECD published a report describing the barriers to (re)integration in the labour market for 39

40 71 people in work disability (i.e. greater competition, heavier workload and work pressure) [5]. The report on September 29, 2021 by guest. Protected copyright. 41 42 72 also describes the underlying social and economic tragedies. Since the results for Belgium were poor, 43 44 73 with a decrease in the number of people with disabilities employed over the past decade, authorities and 45 46 74 social security administrators have been looking for measures or interventions to reverse the trend. A 47 48 49 75 number of studies have been performed to support the authorities, but these are mainly qualitative and are 50 51 76 based on a small sample of cancer survivors [7-13]. 52 53 77 54 55 56 57 58 59 Version 30 août 2016 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 3 of 36 BMJ Open

1 2 3 78 Insurance medicine researchers and epidemiologists acknowledge differences between diagnoses in terms 4 5 79 6 of the duration of work disability [14,15]. Overall, the leading causes of work disability are 7 8 80 musculoskeletal disorders and mental health problems, which have been widely studied [16]. 9 10 81 In Belgium, cancer is the fifth largest cause of work disability, with 18,462 people in work disability due 11 12 82 to cancer in 2013 (6.2% of all disabled workers in Belgium) [17] (Table 1). Each year, more than 25, 000 13 14 83 Belgian inhabitants of working age (20-64 years) are diagnosed with cancer1. 15 For peer review only 16 84 During the last decade, cancer treatments in middle and high income countries have seen major 17 18 85 19 improvements, leading to increased rates of cancer survival [18,19]. Despite improved survival, the 20 21 86 disclosure of cancer still causes important distress among the people affected and their relatives [20], and 22 23 87 it is associated with work disability or death by their colleagues and supervisors [7,21-24]. 24 25 88 This automatic association of cancer with death is becoming obsolete, however, as was notably shown 26 27 89 through the study by Dal Maso et al. [25] which showed that a quarter of Italian cancer survivors have 28 29 90 reached a death rate similar to that of the general population. 30 31

91 http://bmjopen.bmj.com/ 32 Cancer survivors can experience physiological and/or psychosocial symptoms, due to side or long-term 33 34 92 effects of treatment [26] and are more likely to report fair or poor health overall in all age groups [27]. 35 36 93 For these survivors, work can represent a return to health or “normality” or a need to safeguard their 37 38 94 financial security, self-esteem and social contacts [28-33]. 39

40 95 on September 29, 2021 by guest. Protected copyright. 41 42 96 Little is known about social inequalities in relation to return to work (RTW) among cancer survivors [34]. 43 44 97 45 However, the well-established relationship between socio-economic position (SEP) and long-term 46 47 98 sickness absence predicts that returning to work will be more difficult for cancer survivors working in 48 49 99 manual occupations [35,36]. Previous research has shown that working conditions and psychosocial 50 51 100 conditions in manual occupations act as additional barriers [35,37-39]. Alongside the impact of working 52 53 54 55 1 Numbers are computed online. Official website of the Belgian Cancer Registry: 56 http://www.kankerregister.org/default.aspx?lang=EN 57 58 59 Version 30 août 2016 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 4 of 36

1 2 3 101 conditions, the unequal use of cancer rehabilitation services [40] may also lead to social inequalities in 4 5 102 6 terms of returning to work. It has also been shown that cancer survivors with a low SEP more commonly 7 8 103 become unemployed [41] or take early retirement, which can act as a substitute for sickness absence 9 10 104 benefits or unemployment [41-43]. 11 12 105 13 14 106 This article describes and discusses the results of a population-based cohort study of people with long- 15 For peer review only 16 107 term cancer-related work disability, i.e. receiving sickness absence benefits for more than one year. We 17 18 108 19 will refer to this population below as ‘disabled workers’. They are followed for 3 to 7 years in order to 20 21 109 measure the outflow from work disability to either retirement, ability to RTW or death. 22 23 24 110 The Belgian context 25 26 111 In Belgium, cessation of work due to sickness has to be reported to the employer immediately. The 27 28 112 employer pays the “guaranteed salary” for 14 working days for blue-collar workers (manual workers) 29

30 2 31 113 and 28 for white-collar workers (intellectual workers) . For self-employed workers or unemployed 32 http://bmjopen.bmj.com/ 33 114 people, the SSS covers salary replacements starting after 28 working days. The absence due to sickness 34 35 115 has to be confirmed by a general practitioner or a specialist doctor. 36 37 116 After the period of guaranteed income, the social security system (SSS) takes over the provision of a 38 39 117 replacement income. These benefits due to sickness-related absences vary between 40% and 65% of the

40 on September 29, 2021 by guest. Protected copyright. 41 118 reference salary, depending on the family situation. 42 43 44 119 The SSS distinguishes between short-term and long-term work disability. Short-term work disability has 45 46 120 duration of 1 year, while long-term work disability is for periods exceeding 1 year. The division reflects a 47 48 121 different evaluation method to assess the worker’s eligibility for sickness absence benefits and also the 49 50 122 calculation of the level of sickness absence benefits. 51 52 53 54 2 The legal distinction between both OCs is based on the “nature” of the work performed: mainly manual jobs for blue-collar 55 workers and intellectual jobs for white-collar workers. Official website of the national authority of employment: 56 http://www.belgium.be/fr/emploi/contrats_de_travail/types_de_contrats/nature_du_travail. 57 58 59 Version 30 août 2016 60 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 5 of 36 BMJ Open

1 2 3 123 Entitlement to long-term sickness absence benefits starts from the second year after stopping work (13th 4 5 124 6 month) until the age of retirement, with no limit to its duration. This applies to employees, self-employed 7 8 125 and unemployed socially insured Belgian citizens, while civil servants (almost 20% of the Belgian 9 10 126 workforce) benefit from a specific social security scheme. In Belgium, more than 90% of citizens are 11 12 127 socially insured and covered by compulsory health insurance [44]. 13 14 128 Our work reflects research on work disability due to cancer. Work disability is defined or measured as a 15 For peer review only 16 129 legal status based on administrative definitions, i.e. eligibility for benefit. In 2011, a new measure was 17 18 130 19 implemented, allowing disabled workers to resume work without prior agreement of the health insurer's 20 21 131 medical advisor. 22 23 132 24 25 133 There is an important knowledge gap in Belgium regarding quantitative assessment of the impact of 26 27 134 cancer on work disability. The following aspects need to be better understood: how long it lasts, how it 28 29 135 ends, which workers are more at risk, etc. Our research helps to fill this gap. It is based on a recent model, 30 31

136 http://bmjopen.bmj.com/ 32 developed in 2011 to study RTW after cancer [22] which proposes a comprehensive list of influencing 33 34 137 factors. Among these, we have been able to collect and analyze the following: age, gender, occupational 35 36 138 class, cancer site and work-related outcomes (ability to work, (early) retirement, death and disability). 37 38 139 39

40 140 This study is part of the scientific approach initiated in 2012 at the request of the Federal Ministry of on September 29, 2021 by guest. Protected copyright. 41 42 141 Public Health and Social Security [45,46], to provide evidence and support to the decision-making 43 44 142 45 process to improve and facilitate the professional reintegration of cancer survivors. 46 47 143 48 49 144 50 51 145 52 53 146 54 55 147 56 57 58 148 59 Version 30 août 2016 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 6 of 36

1 2 3 4 149 Methods 5 6 7 150 Design & Statistical analysis 8 9 10 151 We conducted a register-based cohort study, using data from the disability register of the National 11 12 152 Institute for Health and Disability Insurance (NIHDI). Our research has three goals. Our first goal was to 13 14 153 measure the between time spent in disability and age, gender, year of entrance, occupational class and 11 15 For peer review only 16 154 cancer groups covering all cancer sites observed. To achieve this first goal, we calculated the Kaplan- 17 18 155 Meier estimate of time to event and non-parametric and model-based cause-specific cumulative 19 20 156 incidences. 21 22 23 157 Secondly, following the taxonomy set out in theories of work disability [48], our study aimed to build a 24 25 158 prognostic model to estimate the distribution of time spent in disability in the presence of three competing 26 27 159 events: death, (early) retirement, and ability to RTW. For this second objective, we fitted a Cox 28 29 160 proportional hazards model to the cause-specific hazards using a stratified (Model 1, Table 4) and 30 31 161 separated approach (Model 2, Table 5), for each age group and gender, accounting for occupational class, 32 http://bmjopen.bmj.com/ 33 162 cancer groups and year of entrance. 34 35 36 163 A third objective was to investigate social inequalities in work disability among cancer survivors, paying 37 38 164 attention to differences in age, gender and occupational class. For this, we fitted a Cox proportional 39

40 165 hazards model to the cause-specific hazards using a stratified and separated approach for each age group on September 29, 2021 by guest. Protected copyright. 41 42 166 and gender, accounting for occupational class and four categories of cancer: those with low3, medium4 43 44 167 and high5 survival rates, according to the age-standardized 5-year relative survival (ASRS), calculated by 45 46 168 the Belgian Cancer Registry [49]. The –fourth- missing category includes those cancer sites for which the 47 48 49 50 51 52 3 "Esophagus", "Stomach", "Colon & rectum", "Pancreas", "Oth. Mal. Neop. of digestive organs and peritoneum", "Mesothelioma", "Trachea and lung", "Myeloid and others", "CNS", "Oth. malignancies and undefined sites, invasive" 53 4 54 "Lip and oral cavity, nasal cavities, middle ear and accessory sinuses, pharynx, larynx","Non Hodgkin's Disease", "Uterus", "Cervix","Ovary","Oth. mal. neop. of female genitals", "Cervix", "Ovary", "Oth. mal. neop. of female genitals", "Bladder", "Urinary system other 55 than bladder", "Bone and connective tissue" 56 5 "Hodgkin Disease","Acute lymphoid leukemia and lymphoid leukemia, other", "Breast female","Uterus","Kidney","Melanoma of the skin", 57 "Thyroid and other endocrine glands" 58 59 Version 30 août 2016 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 7 of 36 BMJ Open

1 2 3 169 information on the ASRS was not available6. The two rationales behind this approach were as follows: 4 5 170 6 first, it generates a parsimonious model (it avoids the lack of convergence due to the large size of the data 7 8 171 set). Second, this approach makes it possible to account for the “severity” of the disease. 9 10 172 11 12 173 Study population 13 14 174 15 We included all sociallyFor insured peer Belgian people review who were recognized only as work disabled due to cancer on 16 st st 17 175 1 January 2007 and up to 31 December 2011. It must be remembered that civil servants are not included 18 19 176 in the NIHDI database. This cohort comprises 21,701 individuals of whom 6,098 were excluded due to 20 21 177 work disability having started before 1st January 2007 (and non-equivalent follow-up time) or due to 22 23 178 inconsistent records (see Figure 1). The last update of the data was on 31st December 2013, leading to a 24 25 179 maximum of 7 years follow-up. 26 27 28 29 180 Independent prognostic variables 30 31 181 Socio-demographic characteristics included in our study are age at entrance into work disability, 32 http://bmjopen.bmj.com/ 33 182 gender and occupational class (OC). The age variable was based on the date of birth and was further 34 35 183 categorized into four groups: 17-39; 40-49; 50-59 and 60+ years. Occupational classes were based on 36 37 38 184 four categories: blue-collar workers, white-collar workers, self-employed people and assisting spouses 39

40 185 and were recoded into a 3-level variable: blue-collar workers, white-collar workers and self-employed on September 29, 2021 by guest. Protected copyright. 41 42 186 people. 43 44 187 In total, 39 cancer sites have been identified using the ‘pathology codes’ transmitted by the NIHDI and 45 46 188 registered by their ICD-9 codes (Table 2). For the sake of comparability, we translated these into ICD-10 47 48 189 codes7 and gathered them into 11 groups (Table 2). 49 50 51 190 The year of entrance in disability was a continuous variable ranging from 2007 to 2011. We decided to 52 53 191 recode the year of entrance into a two-level variable: 2007-2010 and 2011. This decision is based on an 54 55 6 "Benign tumour", "Mal. neop. of skin other than melanoma", "Oth. malignancies and undefined sites, CIS", "Tumours of uncertain and 56 unspecified behavior" 57 7 Online validated tool to translate ICD-9 codes into ICD-10 codes: http://www.icd10data.com/Convert . 58 59 Version 30 août 2016 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 8 of 36

1 2 3 192 exploratory analysis that showed significant difference in survival pattern between disability acquired 4 5 193 6 before or after 2011. We recoded the year of entrance into a two-level variable: 2007-2010 and 2011 (log- 7 8 194 rank test=502, df=1, p-value < 0.001) (Figure 2). 9 10 11 195 Outcome variables: the three competing events 12 13 196 The outcome variable is the event that causes the end of work disability. We defined three mutually 14 15 197 exclusive events: Fordeath, retirement peer and ability review to return-to-work (RTW).only 16 17 18 198 The status “able to RTW” means that the cancer-disabled worker has been recognized by a health 19 20 199 insurer's occupational doctor as “able to work”. In practice, this might lead to a return to work, 21 22 200 unemployment or a decision to be a housewife/husband. “Retirement” means that the worker is definitely 23 24 201 out of the labour market due to his age and will receive social benefits until death. 25 26 202 Those long-term disabled workers, who have not experienced any event by the end of follow-up, i.e. on 27 28 203 31st December 2013, were administratively censored (31%, Table 3). 29 30 31 32 http://bmjopen.bmj.com/ 33 204 Results 34 35 36 37 205 Description of the study population 38 39 206 No observed workers were lost to follow-up. Table 3 describes the main characteristics of work disabled

40 on September 29, 2021 by guest. Protected copyright. 41 207 cancer survivors included in the study. 42 43 208 The overall mean age at entry is 48.5 years (SD=8.7 years), with a mean age among men of 50.4 years 44 45 209 46 (SD=9 years) and 47.3 years among women (S.D=8.33 years). The majority (77%) of the cancer-disabled 47 48 210 workers are aged 40–59 years. 49 50 211 Women are over-represented (62%), younger at entry (median age is 48 vs. 53 years in men) and mostly 51 52 212 white-collar workers (46% vs. 21% among men) or blue-collar workers (43% vs. 60%). The outcome for 53 54 213 the majority of cancer disabled women is ability to RTW (44.10%), while for most men the outcome is 55 56 214 death (50.86%). 57 58 59 Version 30 août 2016 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 9 of 36 BMJ Open

1 2 3 215 The most frequent cancer site is breast with 35% (n= 5,949) of disabled workers, followed by 15% 4 5 216 6 (n=2,400) of digestive tract and 9% (n=1,417) of respiratory tract cancers. 7 8 217 Regarding OC, half of the disabled workers are blue-collar workers, the majority of whom (38.2% of the 9 10 218 total) die by the end of follow-up. White-collar workers (37%) have the shortest median time spent in 11 12 219 work disability (1.30 years vs. 1.79 for the others) and the majority of them (47.7%) are able to RTW by 13 14 220 the end of follow-up. Self-employed disabled workers represent 13% of the cohort and the majority of 15 For peer review only 16 221 them (38.2%) die by the end of follow-up. 17 18 222 19 By the end of follow-up, 69% of the cohort have experienced one of the three competing events (32.2% 20 21 223 died, 2.2% retired and 34.6% have been able to RTW). The other 31% remained disabled, distributed as 22 23 224 follows: 35% of those who entered in 2009 and 2010, 31% of those aged 40-49 years, 27% of women, 24 25 225 28% of blue-collar workers and 44% of those with benign or in situ tumours. 26 27 226 Figures 3a-e show the non-parametric cause-specific cumulative incidences of time to ability to RTW in 28 29 227 the presence of competing risks. Figures 4a-d show the box plots of time to any event stratified by each 30 31

228 http://bmjopen.bmj.com/ 32 prognostic variable, respectively. 33 34 229 For all prognostic variables, the curves show a steep increase in ability to RTW within the first two years; 35 36 230 later on, the curves virtually level off. 37 38 231 Young workers (17-39 years) have the highest rates of ability to RTW at the end of follow-up (Figure 3a) 39

40 232 and the shortest periods spent in work disability (Figure 4b). Older workers present the shortest time spent on September 29, 2021 by guest. Protected copyright. 41 42 233 in work disability (Figure 4b), mainly due to death and retirement (77%, Table 3). 43 44 234 45 Women have higher rates of ability to RTW compared to men (Figure 3b) but spend longer periods in 46 47 235 work disability (Figure 4a). White-collar workers have higher rates of work disability and spend shorter 48 49 236 periods in it (Figure 4c). Regarding the cancer groups, workers with breast or haematological cancer have 50 51 237 the highest rates of ability to RTW by the end of follow-up (Figure 3c), but the longest periods spent in 52 53 238 work disability (Figure 4e). Those with respiratory tract, head and neck, digestive or CNS cancers have 54 55 239 the lowest rates of ability to RTW (Figure 3c) and a short time spent in work disability (Figure 4e), 56 57 58 240 mainly due to death. 59 Version 30 août 2016 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 10 of 36

1 2 3 241 Those cancer survivors who entered into work disability in 2011 had higher rates of ability to RTW by the 4 5 242 6 end of follow-up (Figure 3e) and shorter periods of time spent in work disability (Figure 4c). 7 8 9 243 The prediction of ending work disability patterns (model 1) 10 11 244 In Table 4 we report the results of a Cox proportional hazards model on the cause-specific hazards of each 12 13 245 competing event, with stratification by age and gender and allowing interaction between gender and the 14 15 246 prognostic variables.For Good prognostic peer factors reviewfor ability to RTW for onlyboth men and women are disability 16 17 18 247 experienced before 2011 and being a white-collar worker. Regarding the 11 cancer groups, men with 19 20 248 haematological or genital organ cancers are the most likely to be able to RTW. Among women, the 21 22 249 cancer groups with the best chance for ability to RTW are haematological and breast. 23 24 250 Concerning deaths among men, disabled workers with respiratory tract, CNS, bone & connective tissue 25 26 251 cancers are most at risk. Among women, those with respiratory tract, female genital organs, digestive 27 28 252 tract and head and neck cancers are the more at risk. 29 30 31 32 253 Social inequalities in work disability of cancer survivors (model 2) http://bmjopen.bmj.com/ 33 34 254 In the second model we stratify by age and gender and allow interactions between both these variables 35 36 255 and occupational class and survival categories (Table 5). The absence of individuals in certain age 37 38 256 categories experiencing retirement (17-49 years, Table 2) leads to a convergence issue when modelling 39

40 on September 29, 2021 by guest. Protected copyright. 41 257 the cause-specific hazard for this type of event and this is therefore not reported. 42 43 258 Table 5 shows that among men, workers aged 40-49 and 50-59 years and blue-collar workers are less 44 45 259 likely (just over half as likely) to be able to RTW compared to white-collar workers. These results 46 47 260 translate larger social inequalities in the 40-49 and 50-59 age groups compared to the youngest cancer 48 49 261 survivors (blue-collar workers are 22% less likely to be able to RTW in age category 17-39 compared to 50 51 262 55% less likely in age category 40-49). There are no differences between blue-collar and white-collar 52 53 54 263 workers for cancer survivors older than 60 years. Similar results are found for females. 55 56 57 58 59 Version 30 août 2016 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 11 of 36 BMJ Open

1 2 3 264 For both genders, self-employed people are less likely to be able to RTW work than white-collar workers 4 5 265 6 from the age of 40 onwards. 7 8 266 9 10 267 DISCUSSION 11 12 13 268 In this study, we aimed to identify the factors that influence the reason for exiting from work disability 14 15 269 and the length of workFor disability peer among cancer review survivors. only 16 17 270 To achieve this, we first measured the association between the time spent in work disability and age, 18 19 271 gender, occupational class, the year of entrance and 11 cancer groups. Secondly, we estimated the 20 21 272 distribution of reasons for ending work disability in the presence of three competing events and thirdly we 22 23 273 investigated social inequalities in work disability among cancer survivors. 24 25 26 274 27 28 275 Since not many of the studies in this field are population-based, making comparisons is not easy. 29 30 276 However, the impact of these – risk - factors on labour market participation has been tested in previous 31 32 277 studies [47]. Authors report that overall, older age at entry and being male are both factors that decrease http://bmjopen.bmj.com/ 33 34 278 the chance of being economically active. Our results show that being older (>60 years) increase the risk 35 36 279 of dying or retiring and that workers aged 40-49 are the least likely to be able to RTW. Being male does 37 38 39 280 indeed reduce the likelihood of being able to RTW but women spend more time in work disability overall.

40 on September 29, 2021 by guest. Protected copyright. 41 281 42 43 282 Regarding the cancer groups, we found a strong association between respiratory tract, head and neck and 44 45 283 digestive tract cancers and death. The first two include smoking-related cancer sites [50], which represent 46 47 284 major sources of work disability and death in the working age population. 48 49 285 Haematological cancers in both genders, breast cancer among women and male genital cancers among 50 51 52 286 men, were found to be positively associated with the ability to RTW. 53 54 287 Other studies have compared different cancer sites to assess their association with employment status 55 56 288 after cancer diagnosis. In line with our results, workers with respiratory and female genital cancers 57 58 59 Version 30 août 2016 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 12 of 36

1 2 3 289 present smaller proportions of employment than workers with breast or haematological cancers, mainly 4 5 290 6 due to poor self-reported health status [26,27, 51]. 7 8 291 9 10 292 Compared to white-collar workers, blue-collar workers and self-employed disabled workers are less likely 11 12 293 to be able to RTW, especially men aged 40-59 years. This is in line with previous research showing that, 13 14 294 overall, blue-collar workers and self-employed people are less likely to be able to work compared to 15 For peer review only 16 295 white-collar workers [26]. In addition to working conditions [52], later stage at diagnosis, differences in 17 18 296 19 treatment [53] and lower participation in rehabilitation services [54] can contribute to these social 20 21 297 inequalities. However, in our competing risk analysis, such inequalities were not found among men, while 22 23 298 the opposite was found among women. OC is also strongly associated with the level of income which 24 25 299 may represent an incentive to RTW when these are significantly different (higher) than sickness absence 26 27 300 benefits [36]. 28 29 301 A different impact of the OC on the risk of work disability according to age and gender has been shown to 30 31

302 http://bmjopen.bmj.com/ 32 exist in a population in a Norwegian county, where young workers with blue-collar jobs are more at risk 33 34 303 than older men [55]. The association between of age and RTW has been reported with contradictory 35 36 304 results in the literature, but the majority found higher age to be associated with later RTW or reduced 37 38 305 chance of employment [34]. Our results show that for Belgian cancer survivors, the opposite is found with 39

40 306 a larger impact of occupational class from the age of 40 years onwards, compared to their younger on September 29, 2021 by guest. Protected copyright. 41 42 307 counterparts. 43 44 308 45 46 47 309 Demographic change and the rising retirement age will increase the number of disabled workers and the 48 49 310 length of work disability, which combined with the effects of the economic crisis (i.e. greater competition 50 51 311 and emphasis on maximum performance), will worsen the situation if we do not implement measures, 52 53 312 interventions and rehabilitation programmes to (re)integrate disabled workers better in the labour market 54 55 313 [5]. 56 57 58 59 Version 30 août 2016 60 12

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 13 of 36 BMJ Open

1 2 3 314 The measures introduced by the Belgian government by the end of 20108 seem to already have had an 4 5 315 6 impact, as those disabled workers who entered in 2011 show better outcomes than the others. Further 7 8 316 studies need to be carried out in future to confirm this trend. However, at the end of follow-up, only 9 10 317 34.6% were able to RTW and 31% were –administratively- censored, remaining disabled. 11 12 318 13 14 319 Strengths, limitations and needs for further research 15 For peer review only 16 17 320 The main strength of this study is the representativeness of the data and the generalizability of our results. 18 19 321 We included in the analysis all Belgian workers disabled due to cancer between 2007- 2011, excluding 20 21 322 civil servants and individuals for whom we detected coding errors. 22 23 323 24 25 324 In most work disability studies, survival analyses are used to estimate the time to an event of interest. The 26 27 325 end of work disability is, however, more complex than this and it may be caused by multiple factors. 28 29 30 326 Therefore, the use of competing risks analysis becomes appropriate to avoid over or under-estimating the 31 32 327 probability of experiencing each event [56]. This model is still rarely used in work disability studies and http://bmjopen.bmj.com/ 33 34 328 its use should be encouraged. 35 36 329 37 38 330 Regarding the objective of predicting disability, the two models showed their ability and effectiveness in 39 40 331 predicting the length and the reasons for ending work disability among Belgian cancer survivors. Our on September 29, 2021 by guest. Protected copyright. 41 42 43 332 second model presents original findings, using the survival rates to identify social inequalities. 44 45 333 Nevertheless, differences still exist among workers of the same age, gender, occupational class and cancer 46 47 334 type; this calls for a more complex model with information on treatment and symptoms [57] and on the 48 49 335 working environment [58]. This is feasible in the future, e.g. by linking data from Cancer Registries to 50 51 336 data on employment and socio-economic status. Results could be used to develop rehabilitation 52 53 337 programmes for cancer survivors similar to those that already exist in other countries [59-62]. 54 55 56 8 Since 2011, disabled workers can resume work (to an adequate extent) without prior authorisation from the health insurer's 57 occupational doctor. 58 59 Version 30 août 2016 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 14 of 36

1 2 3 338 4 5 339 6 While our paper focuses on work disability among cancer survivors in Belgium, it is important to realize 7 8 340 that the methods and principles used are generic and applicable to addressing work disability as whole. 9 10 341 Therefore, this report is also relevant to other conditions and SSS. This paper contributes towards closing 11 12 342 the gap on the transition among cancer survivors from long-term work disability to ability to RTW. 13 14 343 Linking these important results to predictions of cancer incidence9 should make it possible to plan cancer 15 For peer review only 16 344 rehabilitation needs and related sickness absence benefits. 17 18 19 20 21 345 Contributorship statement 22 346 23 24 347 Régine Kiasuwa Mbengi, MPH, PhD Student 25 26 348 Designed the study 27 349 Requested/Collected the data 28 350 Performed the analysis 29 351 Wrote the paper 30 352 31 32 353 Mioara Alina Nicolae, MSc,PhD http://bmjopen.bmj.com/ 33 354 Provided statistical support to build the models 34 355 Provided advices on the writing of the sections on methods and results 35 356 36 357 Prof. Els Goetghebeur, MSc,PhD 37 358 Provided statistical support to build the models 38 39 359 Provided advices on the writing of the sections on methods and results

40 360 on September 29, 2021 by guest. Protected copyright. 41 361 Renee Otter, MD, PhD 42 362 Provided clinical expertise support to build cancer categories 43 363 Substantially contributed to the writing 44 45 364 46 365 Katrien Mortelmans, MD, PhD 47 366 Substantially contributed to the writing 48 367 49 368 Sarah Missine, MSc, PhD 50 369 Provided advice and support to integrate the inequalities perspective in the paper and in the 51 52 370 statistical model 53 371 Supported the writing of the paper 54 55 9 56 Cancer incidence predictions up to 2025 (Belgian Cancer Registry): http://www.kankerregister.org/Tenyears. 57 58 59 Version 30 août 2016 60 14

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 15 of 36 BMJ Open

1 2 3 372 4 5 373 Marc Arbyn, MD 6 374 Substantially contributed to the preparation of the tables and figures and the abstract section 7 375 8 376 Prof. Catherine Bouland, MsC, PhD 9 377 As the co-superviser of the PhD of Miss Kiasuwa, Prof. C. Bouland substantially contributed to 10 378 11 the preparation of the study, the design and the revision of the paper 12 379 13 380 Prof. Christophe de Brouwer, MD, PhD 14 381 As the superviser of the PhD of Miss Kiasuwa, Prof. C. de Brouwer substantially contributed to 15 382 the preparation Forof the study, peer the design and review the revision of the paperonly 16 383 17 384 18 19 385 Funding and Competing interests 20 21 386 22 This study has been funded by the National Institute for Health and Disability Insurance. We 23 24 387 have read and understood BMJ policy on declaration of interests and declare that we have no 25 26 388 competing interests. 27 28 29 389 This article does not contain any studies with human participants or animals performed by any of 30 31 390 the authors. 32 http://bmjopen.bmj.com/ 33 391 34 35 36 392 Data sharing statement 37 38 393 39 The complete and anonym dataset used for this study can be made available from the

40 on September 29, 2021 by guest. Protected copyright. 41 394 corresponding author for researchers interested in comparative studies. This request would be 42 43 395 subject of approval from the Belgian National Institute for Health and Disability Insurance, from 44 45 46 396 the Scientific Institute of Public Health and the Université Libre de Bruxelles. 47 48 397 49 398 50 51 399 52 53 400 54 401 55 56 402 57 58 403 59 Version 30 août 2016 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 16 of 36

1 2 3 404 Reference List 4 405 5 6 406 1. Amir Z, Wilson K, Hennings J, Young A: The meaning of cancer: implications for family finances and 7 407 consequent impact on lifestyle, activities, roles and relationships. Psychooncology 2012, 21: 1167-1174. 8 9 408 2. Daniel C.Lustig: Causal Relationships Between Poverty and Disability. Rehabilitation Counseling Bulletin 10 409 2007, 50: 194-202. 11 12 410 3. Schuring M, Robroek SJ, Otten FW, Arts CH, Burdorf A: The effect of ill health and socioeconomic status 13 411 on labor force exit and re-employment: a prospective study with ten years follow-up in the 14 412 Netherlands. Scand J Work Environ Health 2013, 39: 134-143. 15 For peer review only 16 413 4. Reichard A., Stransky M., Phillips K., Drum C., Mc Clain M.: Does Type of Disability Matter to Public 17 414 Health Policy Practice? Californian Journal of Health Promotion 2015, 13: 25-36. 18 19 415 5. OECD. Sickness, Disability and Work. Breaking the Barriers. 2010. 20 416 Ref Type: Report 21 22 417 6. Luengo-Fernandez R, Leal J, Gray A, Sullivan R: Economic burden of cancer across the European Union: 23 418 a population-based cost analysis. Lancet Oncol 2013, 14: 1165-1174. 24 25 419 7. Betsch N: Kankerpatiënten en werkhervatting: rechten en plichten. Instituut Voor Arbeidsrecht, KU Leuven; 26 420 2013. 27 28 421 8. Desiron HA, Donceel P, de RA, Van HE: A conceptual-practice model for occupational therapy to 29 422 facilitate return to work in breast cancer patients. J Occup Rehabil 2013, 23: 516-526. 30 31 423 9. Desiron HA, Donceel P, Godderis L, Van HE, de RA: What is the value of occupational therapy in return http://bmjopen.bmj.com/ 32 424 to work for breast cancer patients? A qualitative inquiry among experts. Eur J Cancer Care (Engl ) 33 425 2015, 24: 267-280. 34 35 426 10. Desiron HA, Crutzen R, Godderis L, Van HE, de RA: Bridging Health Care and the Workplace: 36 427 Formulation of a Return-to-Work Intervention for Breast Cancer Patients Using an Intervention 37 428 Mapping Approach. J Occup Rehabil 2016. 38 39 429 11. Rommel W. Werken na kanker:welke problemen ervaren (ex-)patiënten die het werk hervatten? 2012. 40 430 Vlaams Liga Tegen Kanker. on September 29, 2021 by guest. Protected copyright. 41 431 Ref Type: Report 42 43 432 12. Tiedtke C, Donceel P, Knops L, Desiron H, Dierckx de CB, de RA: Supporting return-to-work in the face 44 433 of legislation: stakeholders' experiences with return-to-work after breast cancer in Belgium. J Occup 45 434 Rehabil 2012, 22: 241-251. 46 47 435 13. Tiedtke C, Dierckx de CB, Donceel P, de RA: Workplace support after breast cancer treatment: 48 436 recognition of vulnerability. Disabil Rehabil 2014, 1-7. 49 50 437 14. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). Chapter 51 438 12. Future need for research. Scand J Public Health Suppl 2004, 63: 256-258. 52 53 439 15. Alexanderson K, Hensing G: More and better research needed on sickness absence. Scand J Public Health 54 440 2004, 32: 321-323. 55 56 57 58 59 Version 30 août 2016 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 17 of 36 BMJ Open

1 2 3 441 16. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). Chapter 4 442 1. Aim, background, key concepts, regulations, and current statistics. Scand J Public Health Suppl 2004, 5 443 63: 12-30. 6 7 444 17. INAMI. Rapport annuel. Satistiques des indemnités. 2014. 8 445 Ref Type: Report 9 10 446 18. Coleman MP, Gatta G, Verdecchia A, Esteve J, Sant M, Storm H et al.: EUROCARE-3 summary: cancer 11 447 survival in Europe at the end of the 20th century. Ann Oncol 2003, 14 Suppl 5: v128-v149. 12 13 448 19. Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, Rosso S, Coebergh JW, Comber H et al.: Cancer incidence 14 449 and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 2013, 49: 1374-1403. 15 For peer review only 16 450 20. Pryce J, Munir F, Haslam C: Cancer survivorship and work: symptoms, supervisor response, co-worker 17 451 disclosure and work adjustment. J Occup Rehabil 2007, 17: 83-92. 18 19 452 21. Kennedy F HCMFPJ: Returning to work following cancer: a qualitative exploratory study into the 20 453 experience of returning to work following cancer. Eur J Cancer Care 2007, 16: 17-25. 21 22 454 22. Mehnert A: Employment and work-related issues in cancer survivors. Crit Rev Oncol Hematol 2011, 77: 23 455 109-130. 24 25 456 23. Mehnert A, de BA, Feuerstein M: Employment challenges for cancer survivors. Cancer 2013, 119 Suppl 26 457 11: 2151-2159. 27 28 458 24. Verbeek J, Spelten E, Kammeijer M, Sprangers M: Return to work of cancer survivors: a prospective 29 459 cohort study into the quality of rehabilitation by occupational physicians. Occup Environ Med 2003, 60: 30 460 352-357. 31 32 461 25. Dal Maso L, Guzzinati S, Buzzoni C, Capocaccia R, Serraino D, Caldarella A et al.: Long-term survival, http://bmjopen.bmj.com/ 33 462 prevalence, and cure of cancer: a population-based estimation for 818 902 Italian patients and 26 34 463 cancer types. Ann Oncol 2014, 25: 2251-2260. 35 36 464 26. Kiasuwa MR, Otter R, Mortelmans K, Arbyn M, Van OH, Bouland C et al.: Barriers and opportunities for 37 465 return-to-work of cancer survivors: time for action-rapid review and expert consultation. Syst Rev 38 466 2016, 5: 35. 39

40 467 27. Hewitt M, Rowland JH, Yancik R: Cancer survivors in the United States: age, health, and disability. J on September 29, 2021 by guest. Protected copyright. 41 468 Gerontol A Biol Sci Med Sci 2003, 58: 82-91. 42 43 469 28. Amir Z, Brocky J: Cancer survivorship and employment: epidemiology. Occup Med (Lond) 2009, 59: 44 470 373-377. 45 46 471 29. Collins C.G, Ottati A., Feuerstein M: Cancer Survivorship. In Handbook of Work Disability. Prevention and 47 472 Management. Springer edition. Edited by Loisel P, Anema J.R. New York: 2013:289-302. 48 49 473 30. Tiedtke C, de RA, Dierckx de CB, Christiaens MR, Donceel P: Experiences and concerns about 'returning 50 474 to work' for women breast cancer survivors: a literature review. Psychooncology 2010, 19: 677-683. 51 52 475 31. Tiedtke C, Dierckx de CB, de RA, Christiaens MR, Donceel P: Breast cancer treatment and work 53 476 disability: patient perspectives. Breast 2011, 20: 534-538. 54 55 477 32. van Muijen P., Duijts SF, van der Beek AJ, Anema JR: Prognostic factors of work disability in sick-listed 56 478 cancer survivors. J Cancer Surviv 2013, 7: 582-591. 57 58 59 Version 30 août 2016 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 18 of 36

1 2 3 479 33. Grunfeld EA, Cooper AF: A longitudinal qualitative study of the experience of working following 4 480 treatment for gynaecological cancer. Psychooncology 2012, 21: 82-89. 5 6 481 34. van Muijen P., Weevers NL, Snels IA, Duijts SF, Bruinvels DJ, Schellart AJ et al.: Predictors of return to 7 482 work and employment in cancer survivors: a systematic review. Eur J Cancer Care (Engl ) 2013, 22: 8 483 144-160. 9 10 484 35. Sterud T, Johannessen HA: Do work-related mechanical and psychosocial factors contribute to the social 11 485 gradient in long-term sick leave: a prospective study of the general working population in Norway. 12 486 Scand J Public Health 2014, 42: 329-334. 13 14 487 36. Piha K, Laaksonen M, Martikainen P, Rahkonen O, Lahelma E: Interrelationships between education, 15 488 occupationalFor class, income peer and sickness absence.review Eur J Public Health only 2010, 20: 276-280. 16 17 489 37. Melchior M., Krieger N, Kawachi I., et al.: Work Factors and Occupational Class Disaprities in Sickness 18 490 Absence: Findings From the GAZEL Cohort Study. American Journal of Public Health 2005, 95: 1206- 19 491 1212. 20 21 492 38. Melchior M, Goldberg M, Krieger N, Kawachi I, Menvielle G, Zins M et al.: Occupational class, 22 493 occupational mobility and cancer incidence among middle-aged men and women: a prospective study 23 494 of the French GAZEL cohort*. Cancer Causes Control 2005, 16: 515-524. 24 25 495 39. Melchior M, Krieger N, Kawachi I, Berkman LF, Niedhammer I, Goldberg M: Work factors and 26 496 occupational class disparities in sickness absence: findings from the GAZEL cohort study. Am J Public 27 497 Health 2005, 95: 1206-1212. 28 29 498 40. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in cancer 30 499 rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 31 32 500 41. Lindbohm ML, Kuosma E, Taskila T, Hietanen P, Carlsen K, Gudbergsson S et al.: Early retirement and http://bmjopen.bmj.com/ 33 501 non-employment after breast cancer. Psychooncology 2014, 23: 634-641. 34 35 502 42. Carlsen K, Hoybye MT, Dalton SO, Tjonneland A: Social inequality and incidence of and survival from 36 503 breast cancer in a population-based study in Denmark, 1994-2003. Eur J Cancer 2008, 44: 1996-2002. 37 38 504 43. Carlsen K, Dalton SO, Diderichsen F, Johansen C: Risk for unemployment of cancer survivors: A Danish 39 505 cohort study. Eur J Cancer 2008, 44: 1866-1874.

40 on September 29, 2021 by guest. Protected copyright. 41 506 44. Gerkens S, Merkur S: Belgium: Health system review. Health Syst Transit 2010, 12: 1-266, xxv. 42 43 507 45. Kiasuwa Mbengi R. La réinsertion socioprofessionnelle des patients atteints de cancer. Exposé des positions. 44 508 2014. 45 509 Ref Type: Online Source 46 47 510 46. Onkelinx L. Toespraak van de Minister van Volksgezondheid en Sociale Zaken op het Symposium Evaluatie 48 511 van het Kankerplan. 2012. 49 512 Ref Type: Report 50 51 513 47. Rijken M, Spreeuwenberg P, Schippers J, Groenewegen PP: The importance of illness duration, age at 52 514 diagnosis and the year of diagnosis for labour participation chances of people with chronic illness: 53 515 results of a nationwide panel-study in The Netherlands. BMC Public Health 2013, 13: 803. 54 55 516 48. de Rijk A.: Work Disability Theories: A Taxonomy for Researchers. In Handbook of Work Disability. 56 517 Prevention and Managment. Edited by Loisel P., Anema J.R. New York: Springer; 2013:475-499. 57 58 59 Version 30 août 2016 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 19 of 36 BMJ Open

1 2 3 518 49. Belgian Cancer Registry. Cancer Survival in Belgium. 2004-2008. 1-109. 2012. 4 519 Ref Type: Report 5 6 520 50. Regional variation in incidence for smoking and alcohol related cancers in Belgium. Edited by Kris Henau, 7 521 Elizabeth Van Eycken, Geert Silverman, Eero Pukkala. Cancer Epidemiology 39[1], 55-65. 2016. 8 522 Ref Type: Generic 9 10 523 51. Mehnert A, Koch U: Predictors of employment among cancer survivors after medical rehabilitation--a 11 524 prospective study. Scand J Work Environ Health 2013, 39: 76-87. 12 13 525 52. Ervasti J, Kivimaki M, Dray-Spira R, Head J, Goldberg M, Pentti J et al.: Socioeconomic gradient in work 14 526 disability in diabetes: evidence from three occupational cohorts. J Epidemiol Community Health 2016, 15 527 70: 125-131.For peer review only 16 17 528 53. Woods LM, Rachet B, Coleman MP: Origins of socio-economic inequalities in cancer survival: a review. 18 529 Ann Oncol 2006, 17: 5-19. 19 20 530 54. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in cancer 21 531 rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 22 23 532 55. Krokstad S, Johnsen R, Westin S: Social determinants of disability pension: a 10-year follow-up of 62 000 24 533 people in a Norwegian county population. Int J Epidemiol 2002, 31: 1183-1191. 25 26 534 56. Noordzij M, Leffondre K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ: When do we need competing 27 535 risks methods for survival analysis in nephrology? Nephrol Dial Transplant 2013, 28: 2670-2677. 28 29 536 57. Glimelius I, Ekberg S, Linderoth J, Jerkeman M, Chang ET, Neovius M et al.: Sick leave and disability 30 537 pension in Hodgkin lymphoma survivors by stage, treatment, and follow-up time-a population-based 31 538 comparative study. J Cancer Surviv 2015. 32 http://bmjopen.bmj.com/ 33 539 58. Bradley CJ, Bednarek HL: Employment patterns of long-term cancer survivors. Psychooncology 2002, 34 540 11: 188-198. 35 36 541 59. Scott DA Mills M, Black A, Cantwell M, Campbell A, Cardwell CR, Porter S et al.. Multidimensional 37 542 rehabilitation programmes for adult cancer survivors (Review). Cochrane.Database.Syst.Rev. [3]. 2016. 38 543 Ref Type: Generic 39

40 544 60. Mewes JC, Steuten LM, Groeneveld IF, de Boer AG, Frings-Dresen MH, IJzerman MJ et al.: Return-to- on September 29, 2021 by guest. Protected copyright. 41 545 work intervention for cancer survivors: budget impact and allocation of costs and returns in the 42 546 Netherlands and six major EU-countries. BMC Cancer 2015, 15: 899. 43 44 547 61. de Boer AG, Taskila TK, Tamminga SJ, Feuerstein M, Frings-Dresen MH, Verbeek JH: Interventions to 45 548 enhance return-to-work for cancer patients. Cochrane Database Syst Rev 2015, 9: CD007569. 46 47 549 62. Zaman AC, Bruinvels DJ, de Boer AG, Frings-Dresen MH: Supporting cancer patients with work-related 48 550 problems through an oncological occupational physician: a feasibility study. Eur J Cancer Care (Engl ) 49 551 2015. 50 552 51 553 52 53 54 55 56 57 58 59 Version 30 août 2016 60 19

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 20 of 36

1 2 3 Predicting End of Work Disability After Cancer. A 7-years Follow-up Population-based Cohort Study 4 Using Competing Risks Analysis. 5

6 a,c b b a d a 7 R. Kiasuwa Mbengi , M.A Nicolae , Els Goetghebeur , R. Otter , K. Mortlemans , S. Missinne , e c c 8 M. Arbyn , C. Bouland and C. de Brouwer 9 10 11 12 Group of diseases 2007 2008 2009 2010 2011 2012 2013 13 Mental heath 74,054(33%) 78,112(34%) 83,247(34%) 88,535(34%) 92,899(34%) 98,171(35%) 104,291(35%) 14 Muskuloskeletal and connective 58,032(26%) 60,595 (26%) 65,146(27%) 69,583(27%) 74,192(28%) 79,643(28%) 86,071(29%) 15 Circulatory diseases For19,372(9%) peer 19,216(8%) review 19,427(8%) 19,571 only(8%) 19,549(7%) 19,772(7%) 19,963(7%) 16 Traumatic injuries and poisoning 15,302(7%) 15,776(7%) 16,538(7%) 17,080(7%) 17,635(7%) 18,383(6%) 18,955(6%) 17 Tumours* 13,592(6%) 14,266(6%) 15,103(6%) 16,083(6%) 16,742(6%) 17,591(6%) 18,462(6%) 18 Others (13 other conditions) 43,332(19%) 44,188(19%) 45,748(19%) 47,083(18%) 48,482(18%) 49,981(18%) 51,666(17%) 19 TOTAL 223,684 232,153 245,209 257,935 269,499 283,541 299,408 20 (100%) 21 Table 1. Number of cause-specific disabled workers in Belgium. Top 5 Evolution 2007-2013. 22 Annual Report NIHDI, 2014. *including cancers and benign tumours 23 24 25 Belgian socially insured workers in work disability due to cancer on 1st January 2007 + 26 st 27 those who entered up to 31 December 2011 n=21.701 28 29 30 Exclusion of observations with entrance into 31 work disability before 01-Jan-2007, n= 6.098 32 http://bmjopen.bmj.com/ 33 34 35 Belgian socially insured workers in work 36 disability starting between 2007- 2011 37 n=15.603 38 39

40 Exclusion of records with inconsistencies: on September 29, 2021 by guest. Protected copyright.

41 42 In dates of start and end of disability 43 n=5 (end date of disability before start date) 44 45 Gender/cancer site, n=33 46 men with breast cancer, female, n=23 female with breast cancer, male, n=6 47 women with male genital organs cancer, n=2 48 men with ovary cancer, n=1 49 men with female genital organ cancer, n=1 50 51 duplicated records n=22 52

53 54 Belgian socially insured workers with work 55 disability due to cancer starting in 2007 – 56 2011 and no inconsistencies in records 57 n=15.543 58 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 21 of 36 BMJ Open

1 2 3 Figure 1. Flowchart of the number of workers disabled because of cancer between 2007 and 2011 in Belgium. 4 5 GROUPS The 11 cancer groups by anatomical location 5-year 5-year Survival Frequency 6 relative relative category observed in 7 survival rate survival rate the data 8 men*(%) women*(%) 9 1 Other malignancies and undefined sites, CIS NA 1 247 10 Benign tumours 11 2 Head and neck: lip, oral cavity, nasal cavities, middle ear 50.0 57.0 medium 877 and accessory sinuses, pharynx, larynx 12 3 Digestive tract 22.8 22.7 low 2 400 13 14 Esophagus 257 15 Stomach For peer review only 218 16 Colon & rectum 1 479 17 Pancreas 209 18 Other malignant neoplasms of digestive organs and 237 19 peritoneum 4 Respiratory tract 14.6 19.5 low 1 417 20

21 Trachea and lung 1 404 22 mesothelioma 13 23 5 Haematological 1 660 24 25 Hodgkin Disease 86.1 85.0 high 263 26 Non-Hodgkin Disease 67.0 68.9 medium 711 27 Acute lymphoid leukemia and Lymphoid leukemia. Other 81.3 76.7 high 161 28 Myeloid leukemia and others 38.5 40.6 low 307 6 Breast 78.2 88.0 high 5 511 29 30 Brest female 5 494 31 Breast male 17 32 7 Female genital organs 821http://bmjopen.bmj.com/ 33 34 Corpus uterus - cerv. ut. 69.8 medium 273 35 Cervix uteri corp. ut. 79.6 high 147 36 Ovary ovary 54.1 medium 362 Others 37 8 Male genital organs 95.3 - high 486 38 39 Prostate 377

40 Testis 94 on September 29, 2021 by guest. Protected copyright. 41 Others 16 42 9 Urinary tract 388 43 Kidney 71.0 0.7 high 147 44 Bladder 56.6 49.2 medium 178 45 Others 63 10 Central nervous system (CNS) 22.7 25.8 low 709 46 11 Bone and connective tissue (sarcoma’s) 61.9 59.7 medium 47 Melanoma of the skin 86.2 91.0 high 48 Malignant neoplasms of skin other than melanoma NA 49 Thyroid and other endocrine glands 89.3 94.1 high 50 Other malignancies and undefined sites, invasive 51.5 39.1 medium/low 51 Tumours of uncertain and unspecified behavior NA 52 TOTAL 15 543 53 Table 2. The 11 cancer groups used for the analysis. 54 *Reference: Belgian Cancer Registry. Cancer Survival in Belgium, 2004-2008.

56 57 58 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 22 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 Figure 2. Kaplan-Meier estimator for the time in work disability, stratified by the year of entrance into work 24 disability. 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from 4 4 ) )

(censored) (censored) Work disability disability Work 7.46 (17.76- 37.17) 37.17) (17.76- 7.46 27.75 (9.83- 45.68) (9.83- 45.68) 27.75 (7.11- 40.51) 23.81 (0.00- 48.43) 24.05 (0.00- 48.49) 21.33 41.29) (14.28- 27.78 (0.00- 41.44) 20.42 57.68) (31.57- 44.63 (0.00- 48.53) 19.96 (2.65- 32.13 17.39 11.36 (0.00- 40.58) (0.00- 40.58) 11.36 2 33.25) (23.11- 28.18 (0.00- 39.94) 19.49 (5.78- 36.86) 20.50 27.49) (10.29- 18.89 37.47) (16.68- 27.07 31.53) (12.07- 21.80 36.24) (25.95- 31.09 (8.60- 38.44) 23.51 (NA) 0.01 < 33.29) (16.47- 24.88 33.44) (21.40- 27.41 37.31) (26.13- 35.02 41.90) (28.15- 35.03 32.00) (10.23- 21.11

work work Ability to return-to- to Ability 41.09) (33.10- 37.09 24.54) (18.25- 21.40 28.55) (24.68- 26.61 (8.51- 11.76) 10.14 45.17) (40.18- 42.67 59.11) (49.24- 54.17 35.89) (29.19- 32.54 (35.76-45.28) 40.52 28.23) (19.41- 23.82 22.36) (16.18- 19.27 32.63) (25.40- 29.01 31.85) (29.62- 30.74 51.58) (43.80- 47.70 33.32) (28.96- 30.96 26.75) (24.39- 25.57 46.79) (41.41- 44.10 56.27) (51.82- 54.05 41.98) (36.06- 40.52 33.93) (27.55- 30.74 25.63) (20.60- 23.12 36.23) (32.02- 34.12 36.41) (33.20- 34.80 35.73) (32.41- 34.07 37.22) (32.54- 34.88 49.47) (44.04- 46.76 % (IC95%) % (IC95%) - - - Reason of ending work disability disability work ending of Reason Retirement Retirement (1.01 - 4.20) 2.61 (0.86 - 4.24) 2.55 (3.86 - 7.18) 5.52 (0.68 - 6.45) 3.57 (2.07 - 4.26) 3.16 (2.28 - 3.95) 3.12 (0.00 - 1.00) 0.47 (9.72 - 20.21) 14.96 (3.92 - 11.65) 7.78 (0.00 - 0.90) 0.42 (0.20 - 4.97) 2.58 (2.08 - 3.51) 2.80 (1.63 - 3.98) 2.80 (8.04 - 12.31) 10.25 (3.75 - 5.60) 4.68 (2.40 - 4.20) 3.30 (2.23 - 5.07) 3.65 33.63) (28.09- 30.86 (3.87 - 5.60) 4.73 (2.69 - 4.00) 3.35 (1.23 - 2.26) 1.75 (0.42 - 1.02) 0.72 (0.14 - 0.93) 0.54 http://bmjopen.bmj.com/

BMJ Open Death Death - 20.66) (10.68 15.67 - 61.91) (50.27 56.09 - 53.01)) (47.94 50.48 - 77.45) (72.42 74.93 - 29.20) (24.20 26.70 - 16.14) (13.77 14.96 - 47.45) (38.91 43.18 - 24.17) (16.77 20.47 - 53.05) (41.07 47.06 - 56.50) (48.54 52.52 - 52.37) (43.59 47.98 - 39.61) (36.96 38.28 -31.45) (28.57 30.01 - 40.40) (34.33 38.28 5017 (32.28%) 5017 (2.20%) 342 (34.63%) 5383 (30.89%) 4801 52.52) (49.21- 50.86 26.59) (24.46- 25.53 26.13) (22.19- 24.16 29.88) (26.89- 28.38 43.65) (40.53- 42.09 48.97) (43.08- 46.02 37.91) (34.61- 36.26 36.06) (32.80- 34.42 34.11) (30.82- 32.46 31.02) (27.72- 29.37 - 34.17) (29.01 31.59 on September 29, 2021 by guest. Protected copyright.

.30 [1.23-1.37] [1.23-1.37] .30 [3.24- 4.98 [1.41-1.93] 1.69 [1.18-1.41] 1.31 [0.69-0.83] 0.75 [1.66-2.08] 1.83 [1.95-2.30] 2.10 [1.37-1.82] 1.56 [1.52-2.24] 1.73 [1.32-2.16] 1.67 [1.18-1.89] 1.46 [1.03-1.52] 1.24 1.79 [1.63-1.97] [1.63-1.97] 1.79 [1.72-1.89] 1.79 1 (years) ( IC 95%) IC 95%) (years) ( in work disability disability work in Median time spent spent time Median 1.59 [1.52-1.66] [1.52-1.66] 1.59 sp median time (overall ent in disability) [1.12-1.26] 1.18 [1.84-2.05] 1.94 [1.415-1.69] 1.51 [1.637-1.88] 1.76 [1.580-1.79] 1.70 [0.797-1.06] 0.91 [1.780-2.048] 1.89 [1.717-1.960] 1.83 [1.621-1.889] 1.76 [1.410-1.670] 1.54 [0.413-0.487] 0.45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only n (%)n 703 (37%) (37%) 703 15543 15543 (4%) 614 (5.6%) 877 (15.4%) 2400 (9.1%) 1417 (10.7%) 1660 (35.%) 5511 (5.3%) 821 (3.1%) 486 (2.5%) 388 (4.6%) 709 (4.2%) 660 2125 (13%) (13%) 2125 individuals individuals TOTAL TOTAL (50%) 7715 5

(38%) 5874 (62%) 9669 (15.6%) 2421 (32.5%) 5052 (44.7%) 6946 ( 7.21%) 1121 (22.2%) 3454 (24.2%) 3760 (23.4%) 3630 (21.8%) 3388 (8.4 %) 1311

Table 3. DescriptionTable 3. of the characteristics of the individuals included in the cohort, mediantheir spenttime the in disability, rate of those one experiencing of the three competing events at the end of andfollow-up the rate of those remaining work disabled theat end of follow-up.

n=15,543 n=15,543 TOTAL n (%) n TOTAL Bone&Con (sarc.)/Skin/Thyroid Bone&Con Central nervous system (CNS) system nervous Central Urinary tract Urinary Male genital organs genital Male Female genital organs Female Breast Breast Haematological Haematological Respiratory tract Respiratory Digestive tract Digestive Head & Neck & Head CIS/Ben CIS/Ben Cancer site site Cancer Assisting spouse & & spouse Self-employed Assisting White collar White Blue collar collar Blue Occupational Class Occupational Women Women Men Men Gender Gender >=60 >=60 50-59 50-59 40-49 40-49 17-39 17-39 Age at entry entry Age at 2011 2010 2010 2009 2009 2008 2008 2007 2007 Year of entry Year entry of

Page 23 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 24 of 36 5 5

3c 3c

http://bmjopen.bmj.com/ BMJ Open on September 29, 2021 by guest. Protected copyright. 3e 3e 3b

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only

and occupational class group, d) gender, cancer c) b) age, a) per follow-up,end of stratified the by ability RTW of to incidence cumulative 3Non-parametric a-e. Figures 3d 3a e) year of entrance. entrance. of year e) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from 6 6

http://bmjopen.bmj.com/ BMJ Open on September 29, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only Figures 4 a-d. Box4 Figures a-d. plots of time to any of the by gender, three group,events, age year of entrance Youoccupationaland class. could add a box by plot survival category of the cancer.

Page 25 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 26 of 36 7 7

http://bmjopen.bmj.com/ BMJ Open on September 29, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only ox plot of time to events, by cancer group by cancer to time events, of plot ox B Figure 5 e.Figure 5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from

8 8

1.00 0.75 0.80 1.00 2.20 1.00 1.66 1.84 2.36 1.48 1.54 1.87 1.60 1.30 1.92 Women Disability

Censored

1.00 0.78 0.67 1.00 1.51 1.76 2.61 1.33 0.42 1.26 1.45 1.62 1.78 1.00 1.92 Men Disability

1.00 1.00 1.00 0.69 1.28 1.04 1.26 1.06 0.28 1.16 1.79 1.00 2.15 1.61 Retirement

c c c c c a a

Women 1.00 1.04 0.93 0.99 1.00 0.93 0.92 1.00 0.77 0.68 0.45 0.41 RTW 2.36 1.21 1.50 Ability to Ability gender

c b b c c b

1.85 6.16 4.79 3.52 3.06 1.00 1.00 1.00 1.84 0.80 6.07 2.81 0.88 5.13 Death 11.25 stratified stratified

http://bmjopen.bmj.com/ a a a b a BMJ Open

1.00 0.24 1.44 0.90 0.62 0.72 1.00 0.72 1.00 0.38 0.47 0.35 0.32 0.36 Retirement Hazard ratios, Hazard

c c c b a a b b Men

1.00 0.91 1.07 1.00 0.78 1.13 1.00 0.64 0.57 2.17 0.62 1.30 0.15 0.60 1.49 Ability Ability toRTW on September 29, 2021 by guest. Protected copyright.

1.00 0.87 0.69 1.00 1.00 2.70 2.91 5.67 1.51 0.87 0.95 2.41 3.52 3.05 1.78 Death For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only

p< p< 0.05 p< 0.01 able 4. Model 4. able Estimated1: hazard for ratios the status theat endof follow-up, foradjusted age. p≈0.00

SelfEmployedor Assisting spouse Blue collar Blue OCCUP. CLASS OCCUP. White collar Skin/Thyroid /Uncertain/Undefined/ 11 Bone 11 & (Sarcomas) Connective 10 Central 10 nervoussystem 9 Urinary tractUrinary 9 8 Male Male 8 organs genital 7 Female 7 genital organs 6 Breast 6 5 Haematological 5 4 Respiratory 4 tract 3 Digestive Digestive 3 tract 2 Head& 2 Neck CANCERTYPE CIS and1 Benign 2011 20072010

T a b c Year of Year entrance Page 27 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 28 of 36

1 2 3 4 Hazard ratios for death and ability to RTW, stratified by age and 5 gender 6 MEN WOMEN 7 8 Death Ability to RTW Death Ability to RTW 9 17-39 10 low 1.00 1.00 1.00 1.00 c a 11 medium 0.80 1.46 0.75 1.42 c c c b 12 high 0.36 1.71 0.46 1.43 b b 13 missing 0.05 1.18 0.05 1.18 14 15 white collars 1.00For peer1.00 review1.00 1.00 only 16 blue collars 0.81 0.78c 0.85 0.70b 17 self-employed 0.99 0.74 0.98 0.74 18 40-49 19 low 1.00 1.00 1.00 1.00 20 medium 0.68c 1.26 0.62c 1.21 21 high 0.40c 1.92c 0.45c 1.90c 22 missing 0.29c 1.38 0.28c 1.35 23 24 white collars 1.00 1.00 1.00 1.00 25 blue collars 0.86 0.45c 0.85 0.45c 26 self-employed 0.90 0.56c 0.89 0.50b 27 28 50-59 low 1.00 1.00 1.00 1.00 29 c c 30 medium 0.32 0.98 0.60 0.99 c a c 31 high 0.64 1.25 0.41 1.25 c c 32 missing 0.34 0.78 0.34 0.79 http://bmjopen.bmj.com/ 33 34 white collars 1.00 1.00 1.00 1.00 35 blue collars 0.93 0.59c 0.94 0.60c 36 self-employed 0.70c 0.59c 0.70c 0.60c 37 ≥60 38 low 1.00 1.00 1.00 1.00 39 medium 0.64c 1.10 0.61c 1.09

40 high 0.31c 1.24 0.46b 0.61 on September 29, 2021 by guest. Protected copyright. 41 missing 0.49 1.45 0.48 1.44 42 43 white collars 1.00 1.00 1.00 1.00 44 blue collars 1.20 0.89 1.14 0.80 45 b b a c self-employed 0.70 0.49 0.70 0.38 46 Table 5. Cox proportional hazards models stratified per age category and gender accounting for time to 47 48 competing events: hazard ratios associated with employment type and survival categories. 49 50 51 52 53 54 55 56 57 58 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 29 of 36 BMJ Open

1 2 3 4 5 6 List of tables and figures in the supplementary material 7 8 Figure 1. Employment rates among people with disability in the 1990s and 2000s in OECD countries 9 10 Figure 2. Multivariate model of Cancer and Return to Work (Mehnert, 2011) 11 Figure 3. Overall survival stratified by gender, cancer group, age and year of entry into work disability 12 13 Figure 4. Diagram presenting the focus of the study and the included population 14 15 Figure 5. The estimatedFor cause-specific peer cumulative review incidence functions only for the 11 cancer groups, among 16 white collars aged 17-39 and 40-49, men and women. 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39 40 on September 29, 2021 by guest. Protected copyright. 41 Figure 1. Employment rates among people with disability in the 1990s and 2000s in OECD countries. 42 Source: OECD, Sickness, Disability and Work, 2011. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 30 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 Figure 2. Multivariate model of Cancer and Return to Work (Mehnert, 2011) 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 31 of 36 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35

36 Figure 3. Overall survivals straiffied by gender, 11 cancer groups, age groups and year of entry into work 37 disability 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44

45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 32 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 Figure 4. Diagram presenting the focus of the study and the included population. We do include workers 27 28 who are in long-term (>1 year) work disability due to cancer (between 2007 and 2011). 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 Figure 5. The estimated cause-specific cumulative incidence functions for the 11 cancer groups, among 49 50 white collars aged 17-39 and 40-49, men and women. We easily observe that for certain cancer groups, 51 the rates of experiencing ability to RTW and death vary according to gender and age. This is the case for 52 53 head and neck cancers, digestive tract, urinary tract and bone and connective tissues cancer. 54 55 56

57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from

Bone Con. &

CNS CNS

Urinary

Male Male gen.

Fem. Gen.

Breast http://bmjopen.bmj.com/

BMJ Open Haemato

on September 29, 2021 by guest. Protected copyright. Respiratory

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

Digestive tr. For peer review only

49 39

Head&Neck - -

CIS/Benign collars aged 40 aged collars

Wom Wom White White Men Wom Wom White collars aged 17 aged collars White Men

Page 33 of 36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 34 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 35 of 36 BMJ Open

1 2 STROBE Statement—checklist of items that should be included in reports of observational studies 3 4 Item 5 No Recommendation 6 Title and abstract 7 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract 8 page 1 9 (b) Provide in the abstract an informative and balanced summary of what was done 10 and what was found page 1 11 12 Introduction 13 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 14 pages 2-5 15 Objectives For3 peer State specific objectives, review including any prespecified only hypotheses lines 95-98 and 16 17 122-131 18 Methods 19 Study design 4 Present key elements of study design early in the paper lines 95-98 and 140-160 20 21 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, 22 exposure, follow-up, and data collection lines 163-168 23 Participants 6 (a) Cohort study—Give the eligibility criteria, and the sources and methods of 24 selection of participants. Describe methods of follow-up lines163-168 and figure 1 25 26 Case-control study—Give the eligibility criteria, and the sources and methods of 27 case ascertainment and control selection. Give the rationale for the choice of cases 28 and controls 29 Cross-sectional study—Give the eligibility criteria, and the sources and methods of 30 selection of participants 31

http://bmjopen.bmj.com/ 32 (b) Cohort study—For matched studies, give matching criteria and number of 33 exposed and unexposed 34 Case-control study—For matched studies, give matching criteria and the number of 35 controls per case 36 37 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect 38 modifiers. Give diagnostic criteria, if applicable lines 170-192 39 Data sources/ 8* For each variable of interest, give sources of data and details of methods of

40 measurement assessment (measurement). Describe comparability of assessment methods if there on September 29, 2021 by guest. Protected copyright. 41 is more than one group lines 170-192 42 43 Bias 9 Describe any efforts to address potential sources of bias lines 165-167 44 Study size 10 Explain how the study size was arrived at lines 163-168 45 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, 46 describe which groupings were chosen and why lines 152-160 and 170-192 47 48 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding 49 lines 140-160 50 (b) Describe any methods used to examine subgroups and interactions 51 (c) Explain how missing data were addressed 52 53 (d) Cohort study—If applicable, explain how loss to follow-up was addressed 54 Case-control study—If applicable, explain how matching of cases and controls was 55 addressed 56 Cross-sectional study—If applicable, describe analytical methods taking account of 57 sampling strategy 58 59 (e) Describe any sensitivity analyses 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml1 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 36 of 36

1 2 Results 3 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, 4 5 examined for eligibility, confirmed eligible, included in the study, completing follow-up, and 6 analysed lines 163-168 7 (b) Give reasons for non-participation at each stage 8 (c) Consider use of a flow diagram figure 1 9 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and 10 11 information on exposures and potential confounders lines 195-231 12 (b) Indicate number of participants with missing data for each variable of interest no missing 13 data (see Figure 1 for exclusion of individuals with error in the records) 14 (c) Cohort study—Summarise follow-up time (eg, average and total amount) lines 167-168 15 For peer review only 16 Outcome data 15* Cohort study—Report numbers of outcome events or summary measures over time lines 17 185-192 18 Case-control study—Report numbers in each exposure category, or summary measures of 19 exposure 20 21 Cross-sectional study—Report numbers of outcome events or summary measures 22 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their 23 precision (eg, 95% confidence interval). Make clear which confounders were adjusted for 24 and why they were included Tables 2 and 3 25 (b) Report category boundaries when continuous variables were categorized 26 27 (c) If relevant, consider translating estimates of relative risk into absolute risk for a 28 meaningful time period 29 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity 30 analyses lines 243-254 31 32 Discussion http://bmjopen.bmj.com/ 33 Key results 18 Summarise key results with reference to study objectives lines 271-326 34 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. 35 36 Discuss both direction and magnitude of any potential bias lines 309-326 37 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, 38 multiplicity of analyses, results from similar studies, and other relevant evidence lines 328- 39 333

40 on September 29, 2021 by guest. Protected copyright. 41 Generalisability 21 Discuss the generalisability (external validity) of the study results lines 309-311 42 Other information 43 Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, 44 for the original study on which the present article is based lines 335-337 45 46 47 *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and 48 unexposed groups in cohort and cross-sectional studies. 49 50 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and 51 52 published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely 53 available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at 54 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is 55 available at www.strobe-statement.org. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml2 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open

Assessing Factors Associated with Work Disability After Cancer. A Population-Based Cohort Study Using Competing Risks Analysis with a 7-year Follow-Up

For peer review only Journal: BMJ Open

Manuscript ID bmjopen-2016-014094.R1

Article Type: Research

Date Submitted by the Author: 06-Apr-2017

Primary Subject Epidemiology Heading:

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 1 of 39 BMJ Open

1 2 3 4 1 Assessing Factors Associated with Work Disability After Cancer. A 5 2 Population-Based Cohort Study Using Competing Risks Analysis with 6 7 3 a 7-year Follow-Up 8 9 4 R. Kiasuwa Mbengia,c, M.A Nicolae b, Els Goetghebeurb, R. Ottera, K. Mortlemansd, S. Missinne a, 10 e c c 11 5 M. Arbyn , C. Bouland and C. de Brouwer 12 6 a Belgian Cancer Centre, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 13 7 b Stat-Gent CRESCENDO, University of Ghent 14 8 c Research Centre for Environmental and Occupational Health, Brussels School of Public Health, Université Libre de Bruxelles (ESP-ULB) 15 9 d Dr. Katrien Mortelmans,For PhD Consulting peer (KaMoCo) review only 16 10 e Unit Cancer Epidemiology, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 17 11 18 12 Corresponding author: Régine Kiasuwa Mbengi 19 13 [email protected] 14 20 14 rue Juliette Wytsman, 1050 Brusss - Belgium 15 Office: +32 2 642 57 65 21 16 Mobile: +32 479 3926 58 22 23 24 25 17 ABSTRACT 26 18 Objectives 27 19 The number of workers with cancer is increasing dramatically worldwide. One of the main priorities is to 28 20 preserve their quality of life and the sustainability of financing systems for sickness absence. We have 29 21 carried out this study in order to assess factors associated with the ability to work after cancer. Such 30 31 22 insight should help with the planning of rehabilitation needs and tailored programmes. 32 23 Participants http://bmjopen.bmj.com/ 33 24 We conducted this register-based cohort study using individual data from the Belgian Disability 34 25 Insurance. Data on 15,543 socially insured Belgian people who entered into the work disability system 35 26 between 2007 and 2011 due to cancer were used. 36 27 Primary and secondary outcome measures 37 28 We estimated the duration of work disability and the cause-specific cumulative incidence of ability to 38 29 work stratified by age, gender, occupational class and year of entering the work disability system for 11 39 30 cancer sites using Kaplan Meier analyses and Fine and Gray model allowing for competing risks. 40 31 Results on September 29, 2021 by guest. Protected copyright. 41 32 The overall median time of work disability was 1.59 years (95%CI [1.52-1.66]), ranging from 0.75 to 42 33 4.98 years across the 11 sites of cancer. By the end of follow-up, more than one-third of the cancer 43 44 34 survivors in the work disability system were able to work (35%). While a large proportion of the women 45 35 were able to work at the end of follow-up, the men who were able to work could do so sooner. Women, 46 36 white-collar, younger and having haematological, male genital or breast cancers were the most likely to 47 37 be able to RTW. 48 38 Conclusion 49 39 Good prognostic factors for the ability to work were youth, female, white-collar, and having breast, male 50 40 genital or haematological cancers. 51 41 Reviewing our results together with the cancer incidence predictions up to 2025 offers a high value for 52 42 the purposes of social security and rehabilitation planning, and for ascertaining patients’ perspectives. 53 43 54 44 Key words: sickness absence, cancer survivors, competing risks, predictive model, social inequalities 55 45 56 57 46 Strengths and limitations of this study: 58 Version 08 March 2017 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 2 of 39

1 2 3 47 • (Good) Generalisability: we used a population-based dataset without loss of follow-up; the 4 48 external validity is therefore largely not limited; 5 6 49 • Methods: competing risks were added to the traditional survival analysis in order to respect the 7 50 complexity of the outcomes . This is still rarely done in disability studies; 8 51 • Value: our results offer high value (when linked with cancer incidence predictions) for the 9 52 planning of rehabilitation needs for cancer patients, up to the year 2025; 10 53 • The results identify cancer survivors at risk of socio-economicexclusion; 11 54 • Information on treatments and job demands is lacking: this could have helped to (1) identify the 12 55 the risk profile more precisely and (2) tailor return-to-work (RTW) interventions 13 56 14 57 15 For peer review only 16 58 BACKGROUND 17 18 59 The direct and indirect effects of work disability represent a significant burden for people who are absent 19 20 60 due to sickness, and to their families and their employers [1]. Long-term work disability may lead to 21 22 61 social exclusion, deprivation or economic insecurity [2], as well as poor health [3]. The negative impact 23 24 62 of work disability on both social and health status is of high importance for public health [4] but studies 25 26 63 identifying those cancer survivors who are at risk of experiencing long-term work disability and 27 28 64 29 identifying the avoidable proportion of work disability are lagging behind. 30 31 65 32 66 Work disability imposes significant costs on society [5,6] with up to 5% of GDP in Organisation for http://bmjopen.bmj.com/ 33 34 67 Economic Co-operation and Development (OECD) countries being spent on disability benefits [5]. In 35 36 37 68 2010, the OECD published a report describing the barriers to (re)integration in the labour market for 38 39 69 people with disability (i.e. greater competition, more demanding workload and work pressure) [5]. The

40 on September 29, 2021 by guest. Protected copyright. 41 70 report also describes the underlying social and economic tragedies. As the results for Belgium were poor, 42 43 71 with a decrease in the number of people with disabilities employed over the past decade, authorities and 44 45 72 social security administrators have been looking for measures or interventions to reverse the trend. A 46 47 73 number of studies have been performed to support the authorities, but these are mainly qualitative and are 48 49 50 74 based on small samples of cancer survivors [7-13]. 51 52 75 53 54 55 56 57 58 Version 08 March 2017 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 3 of 39 BMJ Open

1 2 3 76 Insurance medicine researchers and epidemiologists acknowledge differences between diagnoses in terms 4 5 77 6 of the duration of work disability [14,15]. Overall, the leading causes of work disability are 7 8 78 musculoskeletal disorders and mental health problems, which have been widely studied [16]. 9 10 79 In Belgium, cancer is the fifth greatest cause of work disability, with 18,462 people on work disability 11 12 80 due to cancer in 2013 (6.2% of all workers on work disability in Belgium) [17] (Table 1). Each year, more 13 14 81 than 25,000 Belgian inhabitants of working age (20-64 years) are diagnosed with cancer. 15 For peer review only 16 82 Over the last decade, cancer treatments in middle and high income countries have greatly improved, 17 18 83 19 leading to increased rates of cancer survival [18,19]. Despite these improved survival rates, a cancer 20 21 84 diagnosis still causes great distress among individuals and their relatives [20], and is associated with work 22 23 85 disability or death by their colleagues and supervisors [7,21-24]. 24 25 86 This automatic association of cancer with death is becoming less and less accurate, however, as was 26 27 87 notably demonstrated in the study by Dal Maso et al. [25] that a quarter of Italian cancer survivors have 28 29 88 reached a death rate similar to that of the general population. 30 31

89 http://bmjopen.bmj.com/ 32 Cancer survivors can experience physiological and/or psychosocial symptoms, due to side effects or long- 33 34 90 term effects of treatment [26] and are more likely to report fair or poor health overall in all age groups 35 36 91 [27]. For these survivors, work can represent a return to health or normality; a safeguard of their financial 37 38 92 security, self-esteem and social contacts [28-33]. 39

40 93 on September 29, 2021 by guest. Protected copyright. 41 42 94 Many studies have highlighted social inequalities in relation to return to work (RTW) among cancer 43 44 95 45 survivors [34] The well-established relationship between socioeconomic position (SEP) and long-term 46 47 96 sickness absence predicts that returning to work will be more difficult for cancer survivors in manual 48 49 97 occupations [35,36]. Previous research has shown that working conditions and psychosocial conditions in 50 51 98 manual occupations act as additional barriers [35,37-39]. Alongside the impact of working conditions, the 52 53 99 unequal use of cancer rehabilitation services [40] may also lead to social inequalities in terms of RTW. It 54 55 100 has also been shown that cancer survivors with a low SEP more commonly become unemployed [41] or 56 57 58 Version 08 March 2017 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 4 of 39

1 2 3 101 take early retirement, which can act as a substitute for sickness absence benefits or unemployment [41- 4 5 102 6 43]. 7 8 103 9 10 11 104 The Belgian context 12 13 105 In Belgium, cessation of work due to sickness must be reported to the employer immediately. The 14 15 106 employer pays theFor guaranteed peer salary for 14 workingreview days for blue-collar only workers (manual workers) and 16 17 18 107 28 working days for white-collar workers (intellectual workers). For self-employed or unemployed 19 20 108 individuals, the social security system (SSS) covers salary replacement after 28 working days. The 21 22 109 absence due to sickness must be confirmed by a general practitioner or a specialist doctor. 23 24 110 After the period of guaranteed income from the employer, the SSS takes over the provision of a 25 26 111 replacement income. The benefits for sickness-related absences vary between 40% and 65% of the 27 28 112 reference salary, depending on the family situation (Figure 1). 29 30 31 113 The SSS distinguishes between short-term and long-term work disability. Short-term work disability lasts 32 http://bmjopen.bmj.com/ 33 114 up to one year, while long-term work disability is for periods exceeding one year. The division reflects a 34 35 115 different evaluation method for assessing the worker’s eligibility for sickness absence benefits as well as 36 37 116 the calculation of the level of sickness absence benefits. 38 39 117 Entitlement to long-term sickness absence benefits begins as of the second year after stopping work (13th

40 on September 29, 2021 by guest. Protected copyright. 41 118 month) and continues until the age of retirement, with no limit of duration. This applies to employees, 42 43 44 119 self-employed and unemployed socially insured Belgian citizens. Civil servants (almost 20% of the 45 46 120 Belgian workforce) benefit from a specific social security scheme. In Belgium, more than 90% of citizens 47 48 121 are socially insured and covered by compulsory health insurance [44]. 49 50 122 51 52 123 There is an important knowledge gap in Belgium regarding a quantitative assessment of the impact of 53 54 124 55 cancer on work disability. The following aspects need to be better understood: how long the work 56 57 125 disability lasts, how the work disability ends, which workers are more at risk, etc. Our research helps to 58 Version 08 March 2017 59 60 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 5 of 39 BMJ Open

1 2 3 126 fill this gap. It is based on a recent model, developed in 2011 to study RTW after cancer [22],which 4 5 127 6 proposes a comprehensive list of influencing factors. Among these, we have been able to collect and 7 8 128 analyse data on the following: age, gender, occupational class, site of cancer and work-related outcomes 9 10 129 (ability to work, retirement, death and disability). 11 12 130 13 14 131 This study is part of the scientific approach initiated in 2012 at the request of the Federal Ministry of 15 For peer review only 16 132 Public Health and Social Security [45,46], to provide evidence and support for the decision-making 17 18 133 19 process in order to improve and facilitate the professional reintegration of cancer survivors. 20 21 134 Our work reflects research on work disability due to cancer. Work disability is defined or measured as a 22 23 135 legal status based on administrative definitions, i.e. eligibility for benefit. 24 25 136 This article describes and discusses the results of a population-based cohort study of people with long- 26 27 137 term cancer-related work disability, i.e. receiving sickness absence benefits for more than one year. We 28 29 138 will refer to this population below as ‘disabled workers’. They have been followed for three to seven 30 31

139 http://bmjopen.bmj.com/ 32 years in order to measure the outflow from work disability to either retirement, ability to work or death. 33 34 140 [Table 1 and Figure 1 can be displayed] 35 36 Group of diseases 2007 2008 2009 2010 2011 2012 2013 37 Mental heath 74,054(33%) 78,112(34%) 83,247(34%) 88,535(34%) 92,899(34%) 98,171(35%) 104,291(35%) 38 Muskuloskeletal and connective 58,032(26%) 60,595 (26%) 65,146(27%) 69,583(27%) 74,192(28%) 79,643(28%) 86,071(29%) 39 Circulatory diseases 19,372(9%) 19,216(8%) 19,427(8%) 19,571(8%) 19,549(7%) 19,772(7%) 19,963(7%) Traumatic injuries and poisoning 15,302(7%) 15,776(7%) 16,538(7%) 17,080(7%) 17,635(7%) 18,383(6%) 18,955(6%) 40 on September 29, 2021 by guest. Protected copyright. 41 Tumours* 13,592(6%) 14,266(6%) 15,103(6%) 16,083(6%) 16,742(6%) 17,591(6%) 18,462(6%) 42 Others (13 other conditions) 43,332(19%) 44,188(19%) 45,748(19%) 47,083(18%) 48,482(18%) 49,981(18%) 51,666(17%) 43 TOTAL 223,684 232,153 245,209 257,935 269,499 283,541 299,408 44 (100%) Table 1. Number of cause-specific disabled workers in Belgium. Top 5 Evolution 2007-2013. 45 Annual Report NIHDI, 2014. 46 *including cancers and benign tumours 47 141 48 49 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 6 of 39

1 2 3 4 142 Methods 5 6 143 Study population 7 8 9 144 We presented the list of data required, the objectives and the format in which we planned to publish the 10 11 145 results to the scientific board of the NIHDI. No ethical or privacy issues were identified by the Board, 12 13 146 which allowed the extraction of the required data and the transfer of the coded dataset to the Cancer 14 15 147 Centre of the ScientificFor Institute peer of Public Health.review All data are administrativeonly data collected by the 16 17 148 NIHDI. We therefore did not need informed consent from the workers. 18 19 20 149 We included all socially insured Belgian people who were recognised as work disabled due to cancer 21 22 150 between 1 January 2007 and 31 December 2011, excluding civil servants who are not included in the 23 24 151 NIHDI database. From the total of 21,701 individuals, 6,098 were excluded either due to their work 25 26 152 disability starting before 1 January 2007 (and non-equivalent follow-up time), or due to inconsistent 27 28 153 records (see Figure 2). The last update of the data was on 31st December 2013, resulting in a maximum 29 30 154 follow-up of seven years. 31 32 http://bmjopen.bmj.com/ 33 34 155 Design & statistical analysis 35 36 156 We conducted a register-based cohort study, using data from the disability register of the National 37 38 157 Institute for Health and Disability Insurance (NIHDI). Our research had three goals. Our first goal was to 39

40 158 measure the association between the duration of work disability and age, gender, year of entry in the work on September 29, 2021 by guest. Protected copyright. 41 42 159 disability system, OC and 11 cancer sites covering all types of cancer observed. To achieve this first goal, 43 44 45 160 we calculated the Kaplan-Meier estimate of time to event for all the variables. 46 47 161 Secondly, following the taxonomy set out in theories of work disability [47], our study aimed to build a 48 49 162 prognostic model to estimate the sub-distribution hazards of each event (death, ability to work and 50 51 163 retirement) in the presence of competing risks using the Fine and Gray [48] model. For each event, the 52 53 164 model was built separately for men and women, while adjusting for age, year of entry, cancer site and 54 55 165 occupational class. 56 57 58 Version 08 March 2017 59 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 7 of 39 BMJ Open

1 2 3 166 A third objective was to investigate social inequalities for ability to work among cancer survivors, paying 4 5 167 6 attention to differences in age, gender and occupational class, and adjusting for year of entry For this, we 7 8 168 also used the Fine and Gray model, replacing the cancer sites with 9 1 2 3 10 169 four categories of cancer: those with low , medium and high survival rates, according to the age- 11 12 170 standardised five-year relative survival (ASRS), calculated by the Belgian Cancer Registry [49]. The 13 14 171 missing category includes those individuals with a cancer site for which the ASRS was not available4. 15 For peer review only 16 172 The two rationales behind this approach were as follows: firstly, it generates a parsimonious model (it 17 18 173 19 avoids the lack of convergence due to the large size of the data set). Secondly, this approach makes it 20 21 174 possible to account for the severity of the disease. 22 23 175 For the two first objectives, we used the “cmprsk” package of the statistical software R which allows sub 24 25 176 distribution analysis of competing risks. For the third objective, we used the Stata’s stcrreg package. 26 27 177 28 29 30 178 Independent prognostic variables 31 32 http://bmjopen.bmj.com/ 33 179 Sociodemographic characteristics included in our study were age at entry into the work disability 34 35 180 system, gender and occupational class. The age variable was based on the date of birth and was further 36 37 181 categorised into four groups: 17-39; 40-49; 50-59; and 60+ years. Occupational classes were based on 38 39 182 four categories: blue-collar workers, white-collar workers, self-employed people and assisting spouses.

40 on September 29, 2021 by guest. Protected copyright. 41 183 They were recoded into a 3-level variable: blue-collar workers, white-collar workers and self-employed 42 43 44 184 people. 45 46 185 In total, 39 cancer sites have been identified using the ‘pathology codes’ transmitted by the NIHDI and 47 48 186 registered by their ICD-9 codes (Table 2). For the sake of comparability, we translated these into ICD-10 49 50 187 codes and gathered them into 11 cancer sites (Table 2). 51 52 53 54 55 56

57 58 Version 08 March 2017 59 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 8 of 39

1 2 3 188 The year of entry in the work disability system was a continuous variable ranging from 2007 to 2011. We 4 5 189 6 decided to recode the year of entry into a two-level variable: 2007-2010 and 2011. This decision is based 7 8 190 on an exploratory analysis that showed significant difference in survival patterns between disability 9 10 191 acquired before or after 2011 (log-rank test=502, df=1, p-value < 0.001) (Figure 3). 11 12 13 192 Outcome variables: the three competing events 14 15 193 The outcome variableFor is the peer event that causes review the end of work disability. only We defined three mutually 16 17 18 194 exclusive events, i.e. competing risks: death, retirement and ability to work. 19 20 195 The status retirement indicates that the worker is definitively out of the labour market due to age and will 21 22 196 receive social benefits until death, while able to work indicates that the cancer-disabled worker was 23 24 197 recognised by a health insurer's doctor as able to work. In practice, this might lead to a RTW, to 25 26 198 unemployment or to a decision to be a stay-at-home spouse. 27 28 199 Those long-term disabled workers who had not experienced any event by the end of follow-up, by. on 31 29 30 31 200 December 2013, were administratively censored (31%, Table 3). 32 http://bmjopen.bmj.com/ 33 201 [Table2, Figures 2 and 3 could be displayed] 34 35 202 36 203 37 204 38 205 39 206

40 207 on September 29, 2021 by guest. Protected copyright. 41 208 42 209 43 210 44 211 45 212 46 213 47 214 48 215 49 216 50 217 51 218 52 219 53 54 220 55 221 56 222 57 223 58 Version 08 March 2017 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 9 of 39 BMJ Open

1 2 3 GROUPS The 11 cancer groups by anatomical location 5-year 5-year Survival rate Frequency 4 relative relative category observed in 5 survival rate survival rate the data 6 men*(%) women*(%) 7 1 Other malignancies and undefined sites, CIS NA 1 247 8 Benign tumours 9 2 Head and neck: lip, oral cavity, nasal cavities, middle ear 50.0 57.0 medium 877 10 and accessory sinuses, pharynx, larynx 3 Digestive tract 22.8 22.7 low 2 400 11 12 Esophagus 257 13 Stomach 218 14 Colon & rectum 1 479 15 Pancreas For peer review only 209 16 Other malignant neoplasms of digestive organs and 237 17 peritoneum 18 4 Respiratory tract 14.6 19.5 low 1 417

19 Trachea and lung 1 404 20 mesothelioma 13 21 5 Haematological 1 660 22 23 Hodgkin Disease 86.1 85.0 high 263 24 Non-Hodgkin Disease 67.0 68.9 medium 711 25 Acute lymphoid leukemia and Lymphoid leukemia. Other 81.3 76.7 high 161 26 Myeloid leukemia and others 38.5 40.6 low 307 27 6 Breast 78.2 88.0 high 5 511

28 Brest female 5 494 29 Breast male 17 30 7 Female genital organs 821 31 32 Corpus uterus - cerv. ut. 69.8 medium 273http://bmjopen.bmj.com/ 33 Cervix uteri corp. ut. 79.6 high 147 34 Ovary ovary 54.1 medium 362 35 Others 36 8 Male genital organs 95.3 - high 486

37 Prostate 377 38 Testis 94 39 Others 16

40 9 Urinary tract 388 on September 29, 2021 by guest. Protected copyright. 41 Kidney 71.0 0.7 high 147 42 Bladder 56.6 49.2 medium 178 43 Others 63 44 10 Central nervous system (CNS) 22.7 25.8 low 709 45 11 Bone and connective tissue (sarcoma’s) 61.9 59.7 medium Melanoma of the skin 86.2 91.0 high 46 Malignant neoplasms of skin other than melanoma NA 47 Thyroid and other endocrine glands 89.3 94.1 high 48 Other malignancies and undefined sites, invasive 51.5 39.1 medium/low 49 Tumours of uncertain and unspecified behavior NA 50 TOTAL 15 543 51 224 Table 2. The 11 cancer groups used for the analysis. 52 225 *Reference: Belgian Cancer Registry. Cancer Survival in Belgium, 2004-2008. 53 226 54 227 55 56 57 58 Version 08 March 2017 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 10 of 39

1 2 3 4 228 Results 5 6 7 229 Description of the study population 8 9 10 230 No observed workers were lost to follow-up. Table 3 describes the main characteristics of the work- 11 12 231 disabled cancer survivors included in the study. 13 14 232 The majority (77%) of the cancer-disabled workers were aged 40–59 years. 15 For peer review only 16 233 Women were over-represented (62%), younger at entry (median age of 48 vs. 53 years for men) and 17 18 234 mostly white-collar workers (46% vs. 21% for men) or blue-collar workers (43% vs. 60%). The outcome 19 20 235 for the majority of cancer-disabled women was ability to work (44.10%), while for most men the outcome 21 22 23 236 was death (50.86%). 24 25 237 The most frequent cancer site was breast, representing 35% (n= 5,949) of disabled workers, followed by 26 27 238 15% (n=2,400) of digestive tract cancers and 9% (n=1,417) of respiratory tract cancers. 28 29 239 Regarding OC, half of the disabled workers were blue-collar workers, the majority of whom (38.2% of 30 31 240 the total) died by the end of follow-up. White-collar workers (37%) had the shortest median time of work 32 http://bmjopen.bmj.com/ 33 241 disability (1.30 years vs. 1.79 years for the others), and the majority (47.7%) were able to work by the end 34 35 36 242 of follow-up. Self-employed disabled workers represented 13% of the cohort, and the majority (38.2%) 37 38 243 died by the end of follow-up. 39

40 244 By the end of follow-up, 69% of the cohort had experienced one of the three competing events (32.2% on September 29, 2021 by guest. Protected copyright. 41 42 245 died, 2.2% retired and 34.6% were able to work). The other 31% remained disabled, distributed as 43 44 246 follows: 35% of those who entered in 2009 and 2010, 31% of those aged 40-49 years, 27% of the women, 45 46 247 28% of the blue-collar workers and 44% of those with benign or in situ tumours (Table 3). 47 48 49 248 Figures 4a-e show the non-parametric cause-specific cumulative incidences of time to ability to work in 50 51 249 the presence of competing risks. For all prognostic variables, the curves show a steep increase in ability to 52 53 250 RTW within the first two years; later, the curves virtually level off. 54 55 56 57 58 Version 08 March 2017 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 11 of 39 BMJ Open

1 2 3 251 Figures 5a-5 show the box plots of time to any event (death, ability to work or retirement) stratified by 4 5 252 6 each prognostic variable, respectively. 7 8 253 Younger workers (17-39 years) had the highest rates of ability to work at the end of follow-up (Figure 4b) 9 10 254 and relatively short periods of work disability (Figure 5b), mainly due to the ability to work. Older 11 12 255 workers presented the shortest work disability periods (Figure 5b), mainly due to death or retirement 13 14 256 (77%, Table 3). 15 For peer review only 16 257 Women had higher rates of ability to work compared to men (Figure 4c) but spent longer periods in work 17 18 258 19 disability (Figure 5c). White-collar workers had higher rates of work disability and spent less time in it 20 21 259 (Figure 5c). Regarding the cancer sites, workers with breast or haematological cancer had the highest 22 23 260 rates of ability to work by the end of follow-up (Figure 4e), but the longest periods spent on work 24 25 261 disability (Figure 5e). Those with respiratory tract, head and neck, digestive or central nervous system 26 27 262 (CNS) cancers had the lowest rates of ability to work (Figure 4e) and shorter periods of work disability 28 29 263 (Figure 5e), mainly due to death. 30 31

264 http://bmjopen.bmj.com/ 32 Those cancer survivors who entered into the work disability system in 2011 had higher rates of ability to 33 34 265 work by the end of follow-up (Figure 4a) and shorter periods on work disability (Figure 5a). 35 36 266 [Table 3, Figures 4 and 5 could be displayed] 37 38 267 39

40 268 on September 29, 2021 by guest. Protected copyright. 41 42 269 43 44 270 45 46 47 271 48 49 272 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 12 of 39

12 12 ) ) Version 08 March 2017 2017 March 08 Version

Ability to work work to Ability 41.09) (33.10- 37.09 24.54) (18.25- 21.40 28.55) (24.68- 26.61 (8.51- 11.76) 10.14 45.17) (40.18- 42.67 59.11) (49.24- 54.17 35.89) (29.19- 32.54 (35.76-45.28) 40.52 28.23) (19.41- 23.82 22.36) (16.18- 19.27 32.63) (25.40- 29.01 31.85) (29.62- 30.74 51.58) (43.80- 47.70 33.32) (28.96- 30.96 26.75) (24.39- 25.57 46.79) (41.41- 44.10 56.27) (51.82- 54.05 41.98) (36.06- 40.52 33.93) (27.55- 30.74 25.63) (20.60- 23.12 36.23) (32.02- 34.12 36.41) (33.20- 34.80 35.73) (32.41- 34.07 37.22) (32.54- 34.88 49.47) (44.04- 46.76 % (IC95%) % (IC95%) - - - Reason of ending work disability disability work ending of Reason Retirement Retirement (1.01 - 4.20) 2.61 (0.86 - 4.24) 2.55 (3.86 - 7.18) 5.52 (0.68 - 6.45) 3.57 (2.07 - 4.26) 3.16 (2.28 - 3.95) 3.12 (0.00 - 1.00) 0.47 (9.72 - 20.21) 14.96 (3.92 - 11.65) 7.78 (0.00 - 0.90) 0.42 (0.20 - 4.97) 2.58 (2.08 - 3.51) 2.80 (1.63 - 3.98) 2.80 (8.04 - 12.31) 10.25 (3.75 - 5.60) 4.68 (2.40 - 4.20) 3.30 (2.23 - 5.07) 3.65 33.63) (28.09- 30.86 (3.87 - 5.60) 4.73 (2.69 - 4.00) 3.35 (1.23 - 2.26) 1.75 (0.42 - 1.02) 0.72 (0.14 - 0.93) 0.54 http://bmjopen.bmj.com/

.30 [1.23-1.37] [1.23-1.37] .30 [3.24-6.72] 4.98 [1.41-1.93] 1.69 [1.18-1.41] 1.31 [0.69-0.83] 0.75 [1.66-2.08] 1.83 [1.95-2.30] 2.10 [1.37-1.82] 1.56 [1.52-2.24] 1.73 [1.32-2.16] 1.67 [1.18-1.89] 1.46 [1.03-1.52] 1.24 1.79 [1.63-1.97] [1.63-1.97] 1.79 [1.72-1.89] 1.79 1 (years) ( IC 95%) IC 95%) (years) ( in work disability disability work in Median time spent spent time Median 1.59 [1.52-1.66] [1.52-1.66] 1.59 sp median time (overall ent in disability) [1.12-1.26] 1.18 [1.84-2.05] 1.94 [1.415-1.69] 1.51 [1.637-1.88] 1.76 [1.580-1.79] 1.70 [0.797-1.06] 0.91 [1.780-2.048] 1.89 [1.717-1.960] 1.83 [1.621-1.889] 1.76 [1.410-1.670] 1.54 [0.413-0.487] 0.45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only n (%)n 703 (37%) (37%) 703 15543 15543 (4%) 614 (5.6%) 877 (15.4%) 2400 (9.1%) 1417 (10.7%) 1660 (35.%) 5511 (5.3%) 821 (3.1%) 486 (2.5%) 388 (4.6%) 709 (4.2%) 660 2125 (13%) (13%) 2125 individuals individuals TOTAL TOTAL (50%) 7715 5

(38%) 5874 (62%) 9669 (15.6%) 2421 (32.5%) 5052 (44.7%) 6946 ( 7.21%) 1121 (22.2%) 3454 (24.2%) 3760 (23.4%) 3630 (21.8%) 3388 (8.4 %) 1311

274 275 273 TOTAL n (%) n TOTAL Bone&Con (sarc.)/Skin/Thyroid Bone&Con Central nervous system (CNS) system nervous Central Urinary tract Urinary Male genital organs genital Male Female genital organs Female Breast Breast Haematological Haematological Respiratory tract Respiratory Digestive tract Digestive Head & Neck & Head CIS/Ben CIS/Ben Cancer site site Cancer Assisting spouse & & spouse Self-employed Assisting White collar White Blue collar collar Blue Occupational Class Occupational Women Women Men Men Gender Gender >=60 >=60 50-59 50-59 40-49 40-49 17-39 17-39 Age at entry entry Age at 2011 2010 2010 2009 2009 2008 2008 2007 2007 Year of entry Year entry of

n=15,543 n=15,543 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 13 of 39 BMJ Open

1 2 3 4 276 Prediction patterns of the end of work disability (model 1) 5 6 7 277 Results in Table 4 suggest that good prognostic factors for the ability to work for both men and women 8 9 278 are disability experienced after 2011 and white-collar OC. Regarding the 11 cancer sites, men with 10 11 279 haematological or genital organ cancers are the most likely to be able to work. Among women, the cancer 12 13 280 sites with the best chance for ability to work are haematological and breast. 14 15 281 Concerning deathsFor among men, peer disabled workers review with respiratory tract,only CNS, bone & connective tissue 16 17 18 282 cancers are most at risk. Among women, those with respiratory tract, female genital organs, digestive 19 20 283 tract and head and neck cancers are most at risk. 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 14 of 39

14 14

- - - - - 95%CI 0.45,1.24 1.13,3.02 0.03,2.44 0.16,2.05 0.03,0.93 0.21,3.34 0.21,2.42 0.01,0.51 (0.15,3.49) - (0.02,2.01) (0.14,1.17) (54.3,204.8) Retirement

Version 08 March 2017 2017 March 08 Version - - - - - SHR 1 0.75 1.84 1 0.25 0.57 0.18 0.84 0.72 0.05 0.73 0.73 0 0.20 1 0.41 1 105.5

- - 95%CI (1.32,1.54) (0.95,1.22) (0.54,1.11) (0.55,0.86) (0.19,0.38) (0.89,1.37) (1.19,1.71) (0.65,1.02) (0.48,1.13) (0.29,0.58) (0.57,1.0) (0.74,0.89) (0.58,0.71) (0.39,0.62) (1.82,2.27) WOMEN WOMEN

- - SHR Ability to workAbility 1 1.43 1.07 1 0.77 0.69 0.27 1.11 1.42 0.81 0.74 0.41 0.75 1 0.81 0.64 0.49 1 2.03

- -

95%CI (0.96,1.18) (0.80,1.12) (2.66,6.77) (3.26,7.19) (6.71,14.92) (1.40,3.28) (0.85,1.83) (3.33,7.32) (2.20,6.38) (4.26,9.88) (3.50,8.07) (0.93,1.30) (1.22,1.67) (1.70,2.71) (0.96,1.36) Death

- - SHR 1 1.06 0.94 1 4.24 4.84 10.0 2.15 1.24 4.93 3.75 6.48 5.32 1 1.10 1.42 2.15 1 1.14

------

http://bmjopen.bmj.com/ 95%CI (0.82,2.60) (2.02,5.21) (0.12,1.17) (0.19,1.31) (0.05,0.59) (0.19,1.46) (0.46,3.09) (0.21,1.87) (0.03,0.74) (0.12,1.81) (39.73,284.06) 0.07,0.69 BMJ Open Retirement

------SHR 1 106.20 1 1.46 3.24 1 0.37 0.50 0.18 0.52 1.20 0.62 0.14 0.46 1 0.21

I MEN Sub-distribution hazardat theratio (stratified by endgender) of follow-up 95%C (0.48,0.67) (0.35,0.48) (0.36,0.57) (1.31,1.71) (0.72,1.01) (0.43,0.82) (0.65,1.12) (0.24,0.48) (0.87,1.50) (0.02,0.97) - (1.04,1.85) (0.48,0.67) (0.35,0.57) (0.53,1.09) (1.30,1.84) on September 29, 2021 by guest. Protected copyright.

Ability to workAbility SHR 1 0.57 0.41 0.45 1 1.49 0.85 1 0.59 0.85 0.34 1.14 0.13 - 1.38 0.70 0.47 0.76 1 1.55

- - For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

95%CI (1.29,1.91) (1.60,2.31) (1.96,3.02) (0.92,1.14) (0.73,0.93) (1.97,3.94) (1.94,3.82) (4.14,8.17) (1.03,2.09) (0.45,3.67) (0.58,1.27) (1.78,3.85) (2.71,5.54) (2.01,4.27) (1.08,1.47) Death

For peer review only - - SHR 1 1.57 1.92 2.43 1 1.02 0.83 1 2.79 2.73 5.81 1.47 1.29 0.86 2.62 3.87 2.93 1 1.26

TableModel 4. sub-distribution1: ratio hazard the for theatstatus endof follow-up.

of entry 286 284 285 >=60 >=60 50-59 50-59 40-49 40-49 Age group Age 19-39 Self-Employed or assisting spouse spouse or assisting Self-Employed Blue Collar Blue Collar Occupational Class White Collar 11: Bone and connective 11:connective and Bone 10 Central Nervous System System 10 Nervous Central 9 Urinary Track 9 Track Urinary 8 Male Genital Organs Genital Organs 8 Male 7 Female Genital Organs Genital Organs 7 Female 6 Breast 6 Breast 5 Haematological 5 Haematological 4 Respiratory Track 4 Track Respiratory 3 Digestive track 3 Digestive track 2 Head and Neck and2 Head Neck Cancer site Benign 1 and CIS 2011 2011 2007-2010 2007-2010

Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 15 of 39 BMJ Open

1 2 3 4 287 Social inequalities in the work disability of cancer survivors (model 2) 5 6 7 288 In the second model, we stratify by age and gender and allow interactions between both these variables 8 9 289 and OC and survival categories (Table 5). The absence of individuals in certain age categories entering 10 11 290 retirement (17-49 years, Table 2) leads to a convergence issue when modelling the cause-specific hazard 12 13 291 for this type of event and this is therefore not reported. 14 15 292 Table 5 shows that,For among men, peer workers aged review 40-49, workers aged 50-59only years and blue-collar workers 16 17 18 293 were less likely (just over half as likely) to be able to work compared to white-collar workers. These 19 20 294 results translate into larger social inequalities in the 40-49 and 50-59 age groups compared to the 21 22 295 youngest cancer survivors (blue-collar workers in age category 17-39 were 22% less likely to be able to 23 24 296 work, compared to 55% less likely for age category 40-49). There were no differences between blue- 25 26 297 collar and white-collar workers for cancer survivors older than 60 years. Similar results were found for 27 28 298 women. 29 30 31 299 For both genders, self-employed people were less likely to be able to work than white-collar workers 32 http://bmjopen.bmj.com/ 33 300 from the age of 40 onwards. 34 35 301 36 37 302 38 39 303

40 on September 29, 2021 by guest. Protected copyright. 41 304 42 43 44 305 45 46 306 47 48 307 49 50 308 51 52 309 53 54 310 55 56 57 311 58 Version 08 March 2017 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 16 of 39

1 2 3 Sub-distribution hazard ratio at the end of follow-up 4 5 MEN WOMEN 6 Death Ability to work Death Ability to work 7 17-39 8 Survival rate High 1 1 1 1 9 Medium 1.16 (0.48,2.82) 0.78 (0.47,1.29) 1.69 (1.08,2.65) 0.75 (0.57,0.99) 10 Low 2.63 (1.21,5.72) 0.33 (0.19,0.60) 2.91 (1.99,4.26) 0.38 (0.27,0.53) 11 Missing 2.17 (0.88,5.40) 0.50(0.25,0.98) 1.33 (0.77,2.29) 0.58 (0.42,0.81) 12 Occupational White Collar 1 1 1 1 13 class Blue Collar 0.48(0.20,1.15) 0.86(0.58,1.27) 1.41 (1.02,1.93) 0.58(0.48,0.69) Self-employed 1.43 (0.53,3.84) 0.48 (0.26,0.87) 1.13 (0.60,2.11) 1.05 (0.81,1.38) 14 40-49 15 For peer review only 16 Survival rate High 1 1 1 1 Medium 1.53 (0.74,3.15) 1.22 (0.66,2.25) 3.19 (2.39,4.27) 0.58(0.46,0.72) 17 Low 3.32 (1.67,6.61) 0.63 (0.33,1.20) 5.84 (4.32,7.90) 0.24(0.17,0.35) 18 Missing 0.89 (0.32,2.47) 1.69 (0.82,3.50) 2.96 (2.08,4.22) 0.60 (0.46,0.78) 19 Occupational White Collar 1 1 1 1 20 class Blue Collar 1.12(0.53,2.34) 0.79 (0.42,1.47) 1.42(1.13,1.80) 0.59 (0.53,0.67) 21 Self-employed 1.27(0.47,3.42) 1.74 (0.88,3.43) 0.96(0.62,1.49) 0.84 (0.70,1) 22 50-59 23 Survial rate High 1 1 1 1 24 Medium 1.39(0.93,2.10) 0.82 (0.53,1.24) 3.21 (2.53,4.07) 0.55 (0.44,0.70) 25 Low 3.21(2.15,4.77) 0.38(0.24,0.61) 6.01 (4.73,7.64) 0.21 (0.14,0.31) 26 Missing 1.47(0.86,2.52) 0.76 (0.42,1.36) 2.22 (1.62,3.04) 0.75 (0.57,0.98) 27 Occupational White Collar 1 1 1 1 28 class Blue Collar 0.88 (0.57,1.34) 0.60 (0.39,0.92) 1.07 (0.86,1.34) 0.75 (0.65,0.86) Self-employed 0.75 (0.45,1.24) 0.57 (0.33,0.96) 1.34 (0.98,1.83) 0.69 (0.56,0.86) 29 30 ≥60 31 Survival rate High 1 1 1 1 Medium 1.13 (0.51,2.52) 2.26 (0.90,5.68) 2.85 (1.52,5.38) 0.91 (0.47,1.76) http://bmjopen.bmj.com/ 32 Low 3.16 (1.52,6.55) 0.72 (0.25,2.06) 3.92 (1.91,8.03) 0.29 (0.10,0.86) 33 Missing 2.14 (0.69,6.66 1.21 (0.29,4.99) 2.68(0.95,7.54) 0.41 (0.09,1.89) 34 White Collar 1 1 1 1 35 Occupational class Blue Collar 1.39 (0.61,3.13) 1.57 (0.60,4.08) 1.06 (0.54,2.09) 0.80 (0.46,1.39) 36 Self-employed 0.55 (0.23,1.34) 1.19 (0.46,3.10) 1.09 (0.55,2.18) 0.60 (0.32,1.12) 37 312 Table 5. Sub-distribution hazard ratio based on the Fine and Gray model stratified by age and gender, and 38 313 adjusted for year of entry. 39 314 1Low survival rates: "Esophagus", "Stomach", "Colon & rectum", "Pancreas", "Oth. Mal. Neop. of digestive organs and

40 315 peritoneum", "Mesothelioma", "Trachea and lung", "Myeloid and others", "CNS", "Oth. malignancies and undefined sites, on September 29, 2021 by guest. Protected copyright. 41 316 invasive" 42 317 2Medium survival rates: "Lip and oral cavity, nasal cavities, middle ear and accessory sinuses, pharynx, larynx","Non 43 318 Hodgkin's Disease", "Uterus", "Cervix","Ovary","Oth. mal. neop. of female genitals", "Cervix", "Ovary", "Oth. mal. neop. of 44 319 female genitals", "Bladder", "Urinary system other than bladder", "Bone and connective tissue" 320 3High survival rates: "Hodgkin Disease","Acute lymphoid leukemia and lymphoid leukemia, other", "Breast 45 321 female","Uterus","Kidney","Melanoma of the skin", "Thyroid and other endocrine glands" 46 322 4Missing survival rates:"Benign tumour", "Mal. neop. of skin other than melanoma", "Oth. malignancies and undefined sites, 47 323 CIS", "Tumours of uncertain and unspecified behavior" 48 324 49 50 325 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 17 of 39 BMJ Open

1 2 3 4 326 DISCUSSION 5 6 327 In this study, we aimed to identify the factors that influence the reason for exiting the work disability 7 8 328 system and the length of work disability among cancer survivors. 9 10 329 To achieve this, we first measured the association between the duration of work disability and age, 11 12 13 330 gender, occupational class, the year of entry into the work disability system and 11 cancer sites. Secondly, 14 15 331 we estimated theFor distribution peer of three competing review reasons for exiting only the work disability system; and 16 17 332 thirdly, we investigated social inequalities in work disability among cancer survivors. 18 19 333 20 21 334 As not many of the population-based studies in this field include several cancer sites or use competing 22 23 335 risk analysis, making comparisons is not easy. However, the impact of these determining factors on 24 25 26 336 labour market participation has been tested in previous studies [50]. Results indicate that, overall, older 27 28 337 age at entry into the work disability system and male gender are both factors that decrease the chance of 29 30 338 being economically active. Our results show that an older age (>60 years) increase the risk of dying or 31 32 339 retiring, and that workers aged 40-49 were the most likely to remain disabled for a long period (Table 3). http://bmjopen.bmj.com/ 33 34 340 Male gender did indeed reduce the likelihood of being able to work but women experienced longer 35 36 341 periods within the work disability system overall. 37 38 39 342

40 on September 29, 2021 by guest. Protected copyright. 41 343 Regarding the cancer sites, we found a strong association between respiratory tract, head and neck and 42 43 344 digestive tract cancers and death. The first two include smoking-related cancer sites [51], which represent 44 45 345 major sources of work disability and death in the working age population. 46 47 346 Other studies have compared different cancer sites to assess their association with employment status 48 49 347 after cancer diagnosis. In line with our results, workers with respiratory and female genital cancers 50 51 52 348 present smaller proportions of employment than workers with breast or haematological cancers, mainly 53 54 349 due to poor self-reported health status [26,27,52]. 55 56 350 57 58 Version 08 March 2017 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 18 of 39

1 2 3 351 In line with previous research, blue-collar and self-employed (Torp et al. 2016) workers are less likely to 4 5 352 6 be able to work after cancer compared to white-collar workers, especially for men aged 40-59 years. [26]. 7 8 353 According to other research, these social inequalities could be explained by more demanding working 9 10 354 conditions [53], later stage of cancer at diagnosis, differences in treatment [54] and lower participation in 11 12 355 rehabilitation services [55] among blue-collar workers. However, in our competing risk analysis, such 13 14 356 inequalities were found among women but not among men. OC is also strongly associated with the 15 For peer review only 16 357 income level, which may represent an incentive to RTW when it is significantly different (higher) than 17 18 358 19 sickness absence benefits [36]. 20 21 359 A different impact of the OC on the risk of work disability according to age and gender has been shown in 22 23 360 a Norwegian county, where young workers with blue-collar jobs are more at risk than older men [56]. The 24 25 361 association between age and RTW has been reported with contradictory results in the literature, but the 26 27 362 majority found higher age to be associated with later RTW or reduced chance of employment [34]. Our 28 29 363 results show that, for Belgian cancer survivors, the opposite is found, with a larger impact of OC from the 30 31

364 http://bmjopen.bmj.com/ 32 age of 40 years onwards, compared to younger counterparts. 33 34 365 35 36 366 Demographic changes and the rising retirement age will increase the number of disabled workers and the 37 38 367 length of work disability; combined with the effects of the economic crisis (i.e. greater competition and 39

40 368 emphasis on maximum performance) this will worsen the situation if we do not implement measures, on September 29, 2021 by guest. Protected copyright. 41 42 369 interventions and rehabilitation programmes to better (re)integrate disabled workers in the labour market 43 44 370 45 [5]. 46 47 371 The measure introduced by the Belgian government by the end of 2010 seems to have had an impact 48 49 372 already, as the workers who entered the work disability system in 2011 showed better outcomes than the 50 51 373 others. In 2011, a new measure was implemented, allowing disabled workers to resume work without 52 53 374 prior agreement of the health insurer's medical advisor. 54 55 56 57 58 Version 08 March 2017 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 19 of 39 BMJ Open

1 2 3 375 Further studies need to be carried out in future to confirm this trend. However, at the end of follow-up, 4 5 376 6 only 34.6% of the cancer survivors were able to work and 31% were administratively censored, remaining 7 8 377 disabled. 9 10 378 Based on our results, key features of (work) rehabilitation programmes can be drawn. The non-parametric 11 12 379 cause-specific cumulative incidence of time to ability to work (Figures 4) suggests that interventions 13 14 380 should be planned and implemented within the 2 years after the cancer diagnosis. Differences in age and 15 For peer review only 16 381 gender imply tailoring of and specific attention to the needs of young workers and women. The 17 18 382 19 association of the cancer site with the length of disability suggests that the awareness of oncologists who 20 21 383 treat breast cancer, digestive track cancers and head and neck cancers, should be raised on the RTW and 22 23 384 that they need to be involved in the assessment and management of symptoms. 24 25 385 The negative association of being blue-collar or self-employed calls for the revision of employment 26 27 386 policies for these high-risk groups, with e.g. the creation of incentives for employers to adjust the working 28 29 387 conditions of their sick-listed blue-collar workers. 30 31

388 http://bmjopen.bmj.com/ 32 33 34 389 Strengths, limitations and needs for further research 35 36 390 The main strength of this study is the representativeness of the data and the generalisability of our results. 37 38 391 We included in the analysis all Belgian workers disabled due to cancer between 2007- 2011, excluding 39 40 392 civil servants and individuals for whom we detected coding errors. on September 29, 2021 by guest. Protected copyright. 41 42 43 393 44 45 394 In most work disability studies, survival analyses are used to estimate the time to an event of interest. The 46 47 395 end of work disability is, however, more complex than this, and may be caused by multiple factors. 48 49 396 Therefore, the use of competing risks analysis becomes appropriate to avoid over- or under-estimating the 50 51 397 probability of experiencing each event [57]. This model is still rarely used in work disability studies and 52 53 398 its use should be encouraged. 54 55 56 399 57 58 Version 08 March 2017 59 60 19

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 20 of 39

1 2 3 400 Regarding the objective of predicting disability, the two models showed their capacity to and 4 5 401 6 effectiveness in predicting the length and the reasons for ending work disability among Belgian cancer 7 8 402 survivors. Our second model presents original findings, using the survival rates to identify social 9 10 403 inequalities. 11 12 404 Nevertheless, differences still exist among workers of the same age, gender, OC and cancer site; a more 13 14 405 complex model with information on treatment and symptoms [58] and on the working environment [59] 15 For peer review only 16 406 is thus called for. This is feasible in the future, e.g. by linking data from Cancer Registries to data on 17 18 407 19 employment and socioeconomic status. Results could be used to develop rehabilitation programmes for 20 21 408 cancer survivors similar to those that already exist in other countries [60-63]. 22 23 409 24 25 410 While our paper focuses on work disability among cancer survivors in Belgium, it is important to realise 26 27 411 that the methods and principles used are generic, and applicable for addressing work disability as a whole. 28 29 412 Therefore, this report is also relevant for other conditions and SSS. This paper contributes towards closing 30 31

413 http://bmjopen.bmj.com/ 32 the knowledge gap on the transition among cancer survivors from long-term work disability to ability to 33 34 414 RTW. Linking these important results to predictions of cancer incidence should make it possible to plan 35 36 415 cancer rehabilitation needs and related sickness absence benefits. 37 38 39 40 416 Contributorship statement on September 29, 2021 by guest. Protected copyright. 41 42 417 43 44 418 Régine Kiasuwa Mbengi, MPH, PhD Student 45 419 Designed the study 46 420 Requested/collected the data 47 421 48 Performed the analysis 49 422 Wrote the paper 50 423 51 424 Mioara Alina Nicolae, MSc,PhD 52 425 Provided statistical support to build the models 53 426 Provided advice on the writing of the sections on methods and results 54 55 427 56 428 Prof. Els Goetghebeur, MSc,PhD 57 429 Provided statistical support to build the models 58 Version 08 March 2017 59 60 20

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 21 of 39 BMJ Open

1 2 3 430 Provided advice on the writing of the sections on methods and results 4 5 431 6 432 Renee Otter, MD, PhD 7 433 Provided clinical expertise support to build cancer categories 8 434 Substantially contributed to the writing 9 435 10 436 11 Katrien Mortelmans, MD, PhD 12 437 Substantially contributed to the writing 13 438 14 439 Sarah Missine, MSc, PhD 15 440 Provided adviceFor and support peer to integrate review the inequalities perspective only in the paper and in the 16 441 statistical model 17 18 442 Supported the writing of the paper 19 443 20 444 Marc Arbyn, MD 21 445 Substantially contributed to the preparation of the tables and figures and the abstract section 22 446 23 24 447 Prof. Catherine Bouland, MsC, PhD 25 448 As PhD co-supervisor for Miss Kiasuwa, Prof. C. Bouland substantially contributed to the 26 449 preparation of the study, the design and the revision of the paper 27 450 28 451 Prof. Christophe de Brouwer, MD, PhD 29 452 As the PhD supervisor for Miss Kiasuwa, Prof. C. de Brouwer substantially contributed to the 30 31 453 preparation of the study, the design and the revision of the paper 32 454 http://bmjopen.bmj.com/ 33 455 34 456 Funding and competing interests 35 36 37 457 This study has been funded by the National Institute for Health and Disability Insurance. We 38 39 458 have read and understood the BMJ policy on declaration of interests and declare that we have no

40 on September 29, 2021 by guest. Protected copyright. 41 42 459 competing interests. 43 44 460 This article does not contain any studies with human participants or animals performed by any of 45 46 461 the authors. 47 48 49 462 50 51 463 Data sharing statement 52 53 54 464 The complete and anonymised dataset used for this study can be made available by the 55 56 465 corresponding author for researchers interested in comparative studies. This request would be 57 58 Version 08 March 2017 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 22 of 39

1 2 3 466 subject of approval from the Belgian National Institute for Health and Disability Insurance, the 4 5 6 467 Scientific Institute of Public Health and the Université Libre de Bruxelles. 7 8 9 10 468 Figures legends 11 12 469 Figure 1. The Belgian social security scheme related to work disability 13 470 14 15 471 Figure 2. FlowchartFor of the numberpeer of workers review disabled because ofonly cancer between 2007 and 2011 in 16 17 472 Belgium. 18 473 19 20 474 Figure 3. Kaplan-Meier estimator for the time in work disability, stratified by the year of entrance 21 475 22 into work disability. 23 476 24 25 477 Figures 4 a-e. Non-parametric cumulative incidence of ability to work by the end of follow-up, 26 478 stratified per a) year of entry, b) age at entry, c) gender, d) occupational class and e) cancer sites. 27 28 479 29 30 480 Figures 5 a-e. Box plots of time to any of the three events (death, retirement, ability to work), by 31 481 gender, age group, year of entrance and occupational class. 32 http://bmjopen.bmj.com/ 33 482 34 483 35 36 484 37 38 485 39

40 486 on September 29, 2021 by guest. Protected copyright. 41 42 487 43 44 488 45 46 489 47 490 48 49 491 50 51 492 52 493 53 54 494 55 495 56 57 496 58 Version 08 March 2017 59 60 22

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 23 of 39 BMJ Open

1 2 3 497 Reference List 4 5 6 498 7 8 499 1. Amir Z, Wilson K, Hennings J, Young A: The meaning of cancer: implications for family finances 9 500 and consequent impact on lifestyle, activities, roles and relationships. Psychooncology 2012, 21: 10 501 1167-1174. 11 12 502 2. Daniel C.Lustig: Causal Relationships Between Poverty and Disability. Rehabilitation Counseling 13 503 Bulletin 2007, 50: 194-202. 14 15 For peer review only 16 504 3. Schuring M, Robroek SJ, Otten FW, Arts CH, Burdorf A: The effect of ill health and socioeconomic 17 505 status on labor force exit and re-employment: a prospective study with ten years follow-up in 18 506 the Netherlands. Scand J Work Environ Health 2013, 39: 134-143. 19 20 507 4. Reichard A., Stransky M., Phillips K., Drum C., Mc Clain M.: Does Type of Disability Matter to 21 508 Public Health Policy Practice? Californian Journal of Health Promotion 2015, 13: 25-36. 22 23 509 5. OECD. Sickness, Disability and Work. Breaking the Barriers. 2010. 24 510 Ref Type: Report 25 26 27 511 6. Luengo-Fernandez R, Leal J, Gray A, Sullivan R: Economic burden of cancer across the European 28 512 Union: a population-based cost analysis. Lancet Oncol 2013, 14: 1165-1174. 29 30 513 7. Betsch N: Kankerpatiënten en werkhervatting: rechten en plichten. Instituut Voor Arbeidsrecht, KU 31 514 Leuven; 2013. 32 http://bmjopen.bmj.com/ 33 515 8. Desiron HA, Donceel P, de RA, Van HE: A conceptual-practice model for occupational therapy to 34 516 facilitate return to work in breast cancer patients. J Occup Rehabil 2013, 23: 516-526. 35 36 37 517 9. Desiron HA, Donceel P, Godderis L, Van HE, de RA: What is the value of occupational therapy in 38 518 return to work for breast cancer patients? A qualitative inquiry among experts. Eur J Cancer Care 39 519 (Engl ) 2015, 24: 267-280.

40 on September 29, 2021 by guest. Protected copyright. 41 520 10. Desiron HA, Crutzen R, Godderis L, Van HE, de RA: Bridging Health Care and the Workplace: 42 521 Formulation of a Return-to-Work Intervention for Breast Cancer Patients Using an Intervention 43 522 Mapping Approach. J Occup Rehabil 2016. 44 45 46 523 11. Rommel W. Werken na kanker:welke problemen ervaren (ex-)patiënten die het werk hervatten? 47 524 2012. Vlaams Liga Tegen Kanker. 48 525 Ref Type: Report 49 50 526 12. Tiedtke C, Donceel P, Knops L, Desiron H, Dierckx de CB, de RA: Supporting return-to-work in the 51 527 face of legislation: stakeholders' experiences with return-to-work after breast cancer in Belgium. 52 528 J Occup Rehabil 2012, 22: 241-251. 53 54 55 529 13. Tiedtke C, Dierckx de CB, Donceel P, de RA: Workplace support after breast cancer treatment: 56 530 recognition of vulnerability. Disabil Rehabil 2014, 1-7. 57 58 Version 08 March 2017 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 24 of 39

1 2 3 531 14. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 4 532 Chapter 12. Future need for research. Scand J Public Health Suppl 2004, 63: 256-258. 5 6 7 533 15. Alexanderson K, Hensing G: More and better research needed on sickness absence. Scand J Public 8 534 Health 2004, 32: 321-323. 9 10 535 16. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 11 536 Chapter 1. Aim, background, key concepts, regulations, and current statistics. Scand J Public 12 537 Health Suppl 2004, 63: 12-30. 13 14 538 17. INAMI. Rapport annuel. Satistiques des indemnités. 2014. 15 For peer review only 16 539 Ref Type: Report 17 18 540 18. Coleman MP, Gatta G, Verdecchia A, Esteve J, Sant M, Storm H et al.: EUROCARE-3 summary: 19 541 cancer survival in Europe at the end of the 20th century. Ann Oncol 2003, 14 Suppl 5: v128-v149. 20 21 542 19. Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, Rosso S, Coebergh JW, Comber H et al.: Cancer 22 543 incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 23 544 2013, 49: 1374-1403. 24 25 26 545 20. Pryce J, Munir F, Haslam C: Cancer survivorship and work: symptoms, supervisor response, co- 27 546 worker disclosure and work adjustment. J Occup Rehabil 2007, 17: 83-92. 28 29 547 21. Kennedy F HCMFPJ: Returning to work following cancer: a qualitative exploratory study into the 30 548 experience of returning to work following cancer. Eur J Cancer Care 2007, 16: 17-25. 31 32 549 22. Mehnert A: Employment and work-related issues in cancer survivors. Crit Rev Oncol Hematol http://bmjopen.bmj.com/ 33 550 2011, 77: 109-130. 34 35 36 551 23. Mehnert A, de BA, Feuerstein M: Employment challenges for cancer survivors. Cancer 2013, 119 37 552 Suppl 11: 2151-2159. 38 39 553 24. Verbeek J, Spelten E, Kammeijer M, Sprangers M: Return to work of cancer survivors: a

40 554 prospective cohort study into the quality of rehabilitation by occupational physicians. Occup on September 29, 2021 by guest. Protected copyright. 41 555 Environ Med 2003, 60: 352-357. 42 43 556 25. Dal ML, Guzzinati S, Buzzoni C, Capocaccia R, Serraino D, Caldarella A et al.: Long-term survival, 44 45 557 prevalence, and cure of cancer: a population-based estimation for 818 902 Italian patients and 46 558 26 cancer types. Ann Oncol 2014, 25: 2251-2260. 47 48 559 26. Kiasuwa MR, Otter R, Mortelmans K, Arbyn M, Van OH, Bouland C et al.: Barriers and 49 560 opportunities for return-to-work of cancer survivors: time for action-rapid review and expert 50 561 consultation. Syst Rev 2016, 5: 35. 51 52 562 27. Hewitt M, Rowland JH, Yancik R: Cancer survivors in the United States: age, health, and disability. 53 54 563 J Gerontol A Biol Sci Med Sci 2003, 58: 82-91. 55 56 564 28. Amir Z, Brocky J: Cancer survivorship and employment: epidemiology. Occup Med (Lond) 2009, 57 565 59: 373-377. 58 Version 08 March 2017 59 60 24

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 25 of 39 BMJ Open

1 2 3 566 29. Collins C.G, Ottati A., Feuerstein M: Cancer Survivorship. In Handbook of Work Disability. 4 567 Prevention and Management. Springer edition. Edited by Loisel P, Anema J.R. New York: 2013:289- 5 6 568 302. 7 8 569 30. Tiedtke C, de RA, Dierckx de CB, Christiaens MR, Donceel P: Experiences and concerns about 9 570 'returning to work' for women breast cancer survivors: a literature review. Psychooncology 2010, 10 571 19: 677-683. 11 12 572 31. Tiedtke C, Dierckx de CB, de RA, Christiaens MR, Donceel P: Breast cancer treatment and work 13 573 disability: patient perspectives. Breast 2011, 20: 534-538. 14 15 For peer review only 16 574 32. van Muijen P., Duijts SF, van der Beek AJ, Anema JR: Prognostic factors of work disability in sick- 17 575 listed cancer survivors. J Cancer Surviv 2013, 7: 582-591. 18 19 576 33. Grunfeld EA, Cooper AF: A longitudinal qualitative study of the experience of working following 20 577 treatment for gynaecological cancer. Psychooncology 2012, 21: 82-89. 21 22 578 34. van Muijen P., Weevers NL, Snels IA, Duijts SF, Bruinvels DJ, Schellart AJ et al.: Predictors of return 23 579 to work and employment in cancer survivors: a systematic review. Eur J Cancer Care (Engl ) 2013, 24 25 580 22: 144-160. 26 27 581 35. Sterud T, Johannessen HA: Do work-related mechanical and psychosocial factors contribute to 28 582 the social gradient in long-term sick leave: a prospective study of the general working 29 583 population in Norway. Scand J Public Health 2014, 42: 329-334. 30 31 584 36. Piha K, Laaksonen M, Martikainen P, Rahkonen O, Lahelma E: Interrelationships between http://bmjopen.bmj.com/ 32 585 education, occupational class, income and sickness absence. Eur J Public Health 2010, 20: 276- 33 34 586 280. 35 36 587 37. Melchior M., Krieger N, Kawachi I., et al.: Work Factors and Occupational Class Disaprities in 37 588 Sickness Absence: Findings From the GAZEL Cohort Study. American Journal of Public Health 38 589 2005, 95: 1206-1212. 39

40 590 38. Melchior M, Goldberg M, Krieger N, Kawachi I, Menvielle G, Zins M et al.: Occupational class, on September 29, 2021 by guest. Protected copyright. 41 591 occupational mobility and cancer incidence among middle-aged men and women: a prospective 42 43 592 study of the French GAZEL cohort*. Cancer Causes Control 2005, 16: 515-524. 44 45 593 39. Melchior M, Krieger N, Kawachi I, Berkman LF, Niedhammer I, Goldberg M: Work factors and 46 594 occupational class disparities in sickness absence: findings from the GAZEL cohort study. Am J 47 595 Public Health 2005, 95: 1206-1212. 48 49 596 40. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 50 597 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 51 52 53 598 41. Lindbohm ML, Kuosma E, Taskila T, Hietanen P, Carlsen K, Gudbergsson S et al.: Early retirement 54 599 and non-employment after breast cancer. Psychooncology 2014, 23: 634-641. 55 56 57 58 Version 08 March 2017 59 60 25

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 26 of 39

1 2 3 600 42. Carlsen K, Hoybye MT, Dalton SO, Tjonneland A: Social inequality and incidence of and survival 4 601 from breast cancer in a population-based study in Denmark, 1994-2003. Eur J Cancer 2008, 44: 5 6 602 1996-2002. 7 8 603 43. Carlsen K, Dalton SO, Diderichsen F, Johansen C: Risk for unemployment of cancer survivors: A 9 604 Danish cohort study. Eur J Cancer 2008, 44: 1866-1874. 10 11 605 44. Gerkens S, Merkur S: Belgium: Health system review. Health Syst Transit 2010, 12: 1-266, xxv. 12 13 606 45. Kiasuwa Mbengi R. La réinsertion socioprofessionnelle des patients atteints de cancer. Exposé des 14 607 positions. 2014. 15 For peer review only 16 608 Ref Type: Online Source 17 18 609 46. Onkelinx L. Toespraak van de Minister van Volksgezondheid en Sociale Zaken op het Symposium 19 610 Evaluatie van het Kankerplan. 2012. 20 611 Ref Type: Report 21 22 612 47. de Rijk A.: Work Disability Theories: A Taxonomy for Researchers. In Handbook of Work 23 613 Disability. Prevention and Managment. Edited by Loisel P., Anema J.R. New York: Springer; 24 25 614 2013:475-499. 26 27 615 48. Jason P.Fine and Robert J.Gray: A Proportional Hazards Model for the Subdistribution of a 28 616 Competing Risk. Journal of the American Statistical Association 1999, 94: 496-509. 29 30 617 49. Belgian Cancer Registry. Cancer Survival in Belgium. 2004-2008. 1-109. 2012. 31 618 Ref Type: Report 32 http://bmjopen.bmj.com/ 33 619 50. Rijken M, Spreeuwenberg P, Schippers J, Groenewegen PP: The importance of illness duration, 34 35 620 age at diagnosis and the year of diagnosis for labour participation chances of people with 36 621 chronic illness: results of a nationwide panel-study in The Netherlands. BMC Public Health 2013, 37 622 13: 803. 38 39 623 51. Regional variation in incidence for smoking and alcohol related cancers in Belgium. Edited by Kris

40 624 Henau, Elizabeth Van Eycken, Geert Silverman, Eero Pukkala. Cancer Epidemiology 39[1], 55-65. on September 29, 2021 by guest. Protected copyright. 41 625 2016. 42 43 626 Ref Type: Generic 44 45 627 52. Mehnert A, Koch U: Predictors of employment among cancer survivors after medical 46 628 rehabilitation--a prospective study. Scand J Work Environ Health 2013, 39: 76-87. 47 48 629 53. Ervasti J, Kivimaki M, Dray-Spira R, Head J, Goldberg M, Pentti J et al.: Socioeconomic gradient in 49 630 work disability in diabetes: evidence from three occupational cohorts. J Epidemiol Community 50 631 Health 2016, 70: 125-131. 51 52 53 632 54. Woods LM, Rachet B, Coleman MP: Origins of socio-economic inequalities in cancer survival: a 54 633 review. Ann Oncol 2006, 17: 5-19. 55 56 634 55. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 57 635 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 58 Version 08 March 2017 59 60 26

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 27 of 39 BMJ Open

1 2 3 636 56. Krokstad S, Johnsen R, Westin S: Social determinants of disability pension: a 10-year follow-up of 4 637 62 000 people in a Norwegian county population. Int J Epidemiol 2002, 31: 1183-1191. 5 6 7 638 57. Noordzij M, Leffondre K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ: When do we need 8 639 competing risks methods for survival analysis in nephrology? Nephrol Dial Transplant 2013, 28: 9 640 2670-2677. 10 11 641 58. Glimelius I, Ekberg S, Linderoth J, Jerkeman M, Chang ET, Neovius M et al.: Sick leave and 12 642 disability pension in Hodgkin lymphoma survivors by stage, treatment, and follow-up time-a 13 643 population-based comparative study. J Cancer Surviv 2015. 14 15 For peer review only 16 644 59. Bradley CJ, Bednarek HL: Employment patterns of long-term cancer survivors. Psychooncology 17 645 2002, 11: 188-198. 18 19 646 60. Scott DA Mills M, Black A, Cantwell M, Campbell A, Cardwell CR, Porter S et al.. Multidimensional 20 647 rehabilitation programmes for adult cancer survivors (Review). Cochrane.Database.Syst.Rev. [3]. 21 648 2016. 22 649 Ref Type: Generic 23 24 25 650 61. Mewes JC, Steuten LM, Groeneveld IF, de Boer AG, Frings-Dresen MH, IJzerman MJ et al.: Return- 26 651 to-work intervention for cancer survivors: budget impact and allocation of costs and returns in 27 652 the Netherlands and six major EU-countries. BMC Cancer 2015, 15: 899. 28 29 653 62. de Boer AG, Taskila TK, Tamminga SJ, Feuerstein M, Frings-Dresen MH, Verbeek JH: Interventions 30 654 to enhance return-to-work for cancer patients. Cochrane Database Syst Rev 2015, 9: CD007569. 31 http://bmjopen.bmj.com/ 32 655 63. Zaman AC, Bruinvels DJ, de Boer AG, Frings-Dresen MH: Supporting cancer patients with work- 33 34 656 related problems through an oncological occupational physician: a feasibility study. Eur J Cancer 35 657 Care (Engl ) 2015. 36 658 37 659 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 27

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 28 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 Figure 1. The Belgian social security scheme related to work disability 34 35 279x215mm (300 x 300 DPI) 36 37 38 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 Figure2. Flowchart of the number of workers disabled because of cancer between 2007 and 2011 in Belgium. 48 49 143x186mm (300 x 300 DPI) 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 30 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 3. Kaplan-Meier estimator for the time in work disability, stratified by the year of entrance in work on September 29, 2021 by guest. Protected copyright. 41 disability 42 43 127x127mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 31 of 39 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 4a. Non-parametric cumulative incidence of ability to work by the end of follow-up stratified per year on September 29, 2021 by guest. Protected copyright. 41 of entry 42 43 127x127mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 32 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 4b. Non-parametric cumulative incidence of ability to work by the end of follow-up stratified per age on September 29, 2021 by guest. Protected copyright. 41 42 127x127mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 33 of 39 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 4c. Non-parametric cumulative incidence of ability to work by the end of follow-up stratified per on September 29, 2021 by guest. Protected copyright. 41 gender 42 43 127x127mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 34 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 4d. Non-parametric cumulative incidence of ability to work by the end of follow-up stratified per on September 29, 2021 by guest. Protected copyright. 41 occupational class 42 43 127x127mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 35 of 39 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 4e. Non-parametric cumulative incidence of ability to work by the end of follow-up stratified per on September 29, 2021 by guest. Protected copyright. 41 cancer site 42 43 127x127mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 36 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figures 5 a-e. Box plots of time to any of the three events (death, retirement, ability to work), stratified by on September 29, 2021 by guest. Protected copyright. 41 year of entrance, age, gender and occupational class 42 43 177x177mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 37 of 39 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 5e. Box plot of time to any of the three event (death, retirement, ability to work) by cancer site on September 29, 2021 by guest. Protected copyright. 41 42 118x118mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 38 of 39

Assessing Factors Associated with Long-Term Work Disability After Cancer in Belgium. A Population-Based Cohort Study Using Competing Risks Analysis with a 7-year Follow-Up

For peer review only Journal: BMJ Open

Manuscript ID bmjopen-2016-014094.R2

Article Type: Research

Date Submitted by the Author: 30-May-2017

Centre http://bmjopen.bmj.com/ Arbyn, Marc; Scientific Insitute of Public Health, Unit Cancer Epidemiology Bouland, Catherine; Universite Libre de Bruxelles, Centre de Recherche en Santé Environnement et Santé au travail de Brouwer, Christophe; Universite Libre de Bruxelles, Centre de Recherche en Santé Environnement et Santé au travail

Primary Subject Epidemiology Heading:

Secondary Subject Heading: Public health on September 29, 2021 by guest. Protected copyright.

Epidemiology < ONCOLOGY, ONCOLOGY, Organisation of health services < Keywords: HEALTH SERVICES ADMINISTRATION & MANAGEMENT

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 1 of 39 BMJ Open

1 2 3 4 1 Assessing Factors Associated with Long-Term Work Disability After 5 2 Cancer in Belgium. A Population-Based Cohort Study Using 6 7 3 Competing Risks Analysis with a 7-year Follow-Up 8 9 4 R. Kiasuwa Mbengia,c, A.M Nicolaieb , Els Goetghebeurb, R. Ottera, K. Mortlemansd, S. Missinnea, 10 e c c 11 5 M. Arbyn , C. Bouland and C. de Brouwer 12 6 a Belgian Cancer Centre, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 13 7 b Stat-Gent CRESCENDO, University of Ghent 14 8 c Research Centre for Environmental and Occupational Health, Brussels School of Public Health, Université Libre de Bruxelles (ESP-ULB) 15 9 d Dr. Katrien Mortelmans,For PhD Consulting peer (KaMoCo) review only 16 10 e Unit Cancer Epidemiology, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 17 11 18 12 Corresponding author: Régine Kiasuwa Mbengi 19 13 [email protected] 14 20 14 rue Juliette Wytsman, 1050 Brusss - Belgium 15 Office: +32 2 642 57 65 21 16 Mobile: +32 479 3926 58 22 23 24 25 17 ABSTRACT 26 18 Objectives 27 19 The number of workers with cancer is increasing dramatically worldwide. One of the main priorities is to 28 20 preserve their quality of life and the sustainability of social security systems. We have carried out this 29 21 study in order to assess factors associated with the ability to work after cancer. Such insight should help 30 31 22 with the planning of rehabilitation needs and tailored programmes. 32 23 Participants http://bmjopen.bmj.com/ 33 24 We conducted this register-based cohort study using individual data from the Belgian Disability 34 25 Insurance. Data on 15,543 socially insured Belgian people who entered into the long-term work disability 35 26 between 2007 and 2011 due to cancer were used. 36 27 Primary and secondary outcome measures 37 28 We estimated the duration of work disability using Kaplan Meier and the cause-specific cumulative 38 29 incidence of ability to work stratified by age, gender, occupational class and year of entering the work 39 30 disability system for 11 cancer sites using the Fine and Gray model allowing for competing risks. 40 31 Results on September 29, 2021 by guest. Protected copyright. 41 32 The overall median time of work disability was 1.59 years (95%CI [1.52-1.66]), ranging from 0.75 to 42 33 4.98 years. By the end of follow-up, more than one-third of the disabled cancer survivors were able to 43 44 34 work (35%). While a large proportion of the women were able to work at the end of follow-up, the men 45 35 who were able to work could do so sooner. Women, white-collar, younger and having haematological, 46 36 men genital or breast cancers were the most likely to be able to RTW. 47 37 Conclusion 48 38 Good prognostic factors for the ability to work were youth, woman, white-collar, and having breast, men 49 39 genital or haematological cancers. 50 40 Reviewing our results together with the cancer incidence predictions up to 2025 offers a high value for 51 41 social security and rehabilitation planning, and for ascertaining patients’ perspectives. 52 42 53 43 Key words: sickness absence, cancer survivors, competing risks, predictive model, social inequalities 54 44 55 45 Strengths and limitations of this study: 56 57 46 Strengths and limitations of the study methods: 58 Version 08 March 2017 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 2 of 39

1 2 3 47 • (Good) External validity: we used a population-based dataset without loss of follow-up; the 4 48 external validity is therefore largely not limited and the study offers high value (when linked with 5 6 49 cancer incidence predictions) for the planning of rehabilitation needs for cancer patients, up to the 7 50 year 2025; 8 51 • Use of competing risk analysis: competing risks were added to the traditional survival analysis in 9 52 order to respect the complexity of the outcomes. This is still rarely done in disability studies; 10 53 • Incomplete model: the lack of information on treatments and job demands: limited the capacity of 11 54 our model to (1) support the identification of a precise risk profile and (2) to tailor return-to-work 12 55 (RTW) interventions 13 56 14 57 15 For peer review only 16 58 BACKGROUND 17 59 18 The direct and indirect effects of work disability represent a significant burden for people who are absent 19 20 60 due to sickness, and to their families and their employers [1]. Long-term work disability may lead to 21 22 61 social exclusion, deprivation or economic insecurity [2], as well as poor health [3]. The negative impact 23 24 62 of work disability on both social and health status is of high importance for public health [4] but studies 25 26 63 identifying those cancer survivors who are at risk of experiencing long-term work disability and 27 28 64 identifying the avoidable proportion of work disability are lagging behind. 29 30 31 65 32 66 Work disability imposes significant costs on society [5,6] with up to 5% of GDP in Organisation for http://bmjopen.bmj.com/ 33 34 67 Economic Co-operation and Development (OECD) countries being spent on disability benefits [5]. In 35 36 68 37 2010, the OECD published a report describing the barriers to (re)integration in the labour market for 38 39 69 people with disability (i.e. greater competition, more demanding workload and work pressure) [5]. The

40 on September 29, 2021 by guest. Protected copyright. 41 70 report also describes the underlying social and economic tragedies. As the results for Belgium were poor, 42 43 71 with a decrease in the number of people with disabilities employed over the past decade, authorities and 44 45 72 social security administrators have been looking for measures or interventions to reverse the trend. A 46 47 73 number of studies have been performed to support the authorities, but these are mainly qualitative and are 48 49 74 50 based on small samples of cancer survivors [7-13]. 51 52 75 53 54 55 56 57 58 Version 08 March 2017 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 3 of 39 BMJ Open

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 4 of 39

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 5 of 39 BMJ Open

139 http://bmjopen.bmj.com/ 32 years in order to measure the outflow from work disability to either retirement, ability to work or death. 33 34 140 [Table 1 and Figure 1 can be displayed] 35 36 Group of diseases 2007 2008 2009 2010 2011 2012 2013 37 Mental heath 74,054(33%) 78,112(34%) 83,247(34%) 88,535(34%) 92,899(34%) 98,171(35%) 104,291(35%) 38 Muskuloskeletal and connective 58,032(26%) 60,595 (26%) 65,146(27%) 69,583(27%) 74,192(28%) 79,643(28%) 86,071(29%) 39 Circulatory diseases 19,372(9%) 19,216(8%) 19,427(8%) 19,571(8%) 19,549(7%) 19,772(7%) 19,963(7%) Traumatic injuries and poisoning 15,302(7%) 15,776(7%) 16,538(7%) 17,080(7%) 17,635(7%) 18,383(6%) 18,955(6%) 40 on September 29, 2021 by guest. Protected copyright. 41 Tumours* 13,592(6%) 14,266(6%) 15,103(6%) 16,083(6%) 16,742(6%) 17,591(6%) 18,462(6%) 42 Others (13 other conditions) 43,332(19%) 44,188(19%) 45,748(19%) 47,083(18%) 48,482(18%) 49,981(18%) 51,666(17%) 43 TOTAL 223,684 232,153 245,209 257,935 269,499 283,541 299,408 44 (100%) Table 1. Number of cause-specific disabled workers in Belgium. Top 5 Evolution 2007-2013. 45 Annual Report NIHDI, 2014. 46 *including cancers and benign tumours 47 48 49 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 6 of 39

1 2 3 141 4 5 6 7 8 142 Methods 9 10 11 143 Study population 12 13 144 We presented the list of data required, the objectives and the format in which we planned to publish the 14 15 145 results to the scientificFor board peerof the NIHDI. Noreview ethical or privacy issuesonly were identified by the Board, 16 17 146 which allowed the extraction of the required data and the transfer of the coded dataset to the Cancer 18 19 147 Centre of the Scientific Institute of Public Health. All data are administrative data collected by the 20 21 22 148 NIHDI. We therefore did not need informed consent from the workers. 23 24 149 We included all socially insured Belgian people who entered into long-term work disability ursddue to 25 26 150 cancer between 1 January 2007 and 31 December 2011, excluding civil servants who are not included in 27 28 151 the NIHDI database. From the total of 21,701 individuals, 6,098 were excluded either due to their work 29 30 152 disability starting before 1 January 2007 (and non-equivalent follow-up time), or due to inconsistent 31 32 http://bmjopen.bmj.com/ 153 records (see Figure 2). The last update of the data was on 31st December 2013, resulting in a maximum 33 34 35 154 follow-up of seven years. 36 37 38 155 Design & statistical analysis 39

40 156 We conducted a register-based cohort study, using data from the disability register of the National on September 29, 2021 by guest. Protected copyright. 41 42 157 Institute for Health and Disability Insurance (NIHDI). Our research had three goals. Our first goal was to 43 44 45 158 measure the duration of work disability by the year of entry in the work disability system. To achieve this 46 47 159 first goal, we calculated the Kaplan-Meier estimate. 48 49 160 Secondly, following the taxonomy set out in theories of work disability [47], our study aimed to build a 50 51 161 prognostic model to estimate the sub-distribution hazards of each event (death, ability to work and 52 53 162 retirement) in the presence of competing risks using the Fine and Gray [48] model. For each event, the 54 55 56 57 58 Version 08 March 2017 59 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 7 of 39 BMJ Open

1 2 3 163 model was built separately for men and women, while adjusting for age, year of entry, cancer site and 4 5 164 6 occupational class. 7 8 165 A third objective was to investigate social inequalities for ability to work among cancer survivors, paying 9 10 166 attention to differences in age, gender and occupational class, and adjusting for year of entry For this, we 11 12 167 also used the Fine and Gray model, replacing the cancer sites with 13 14 168 four categories of cancer: those with low1, medium2 and high3 survival rates, according to the age- 15 For peer review only 16 169 standardised five-year relative survival (ASRS), calculated by the Belgian Cancer Registry [49]. The 17 18 170 4 19 missing category includes those individuals with a cancer site for which the ASRS was not available . 20 21 171 The two rationales behind this approach were as follows: firstly, it generates a parsimonious model (it 22 23 172 avoids the lack of convergence due to the large size of the data set). Secondly, this approach makes it 24 25 173 possible to account for the severity of the disease. 26 27 174 For the two first objectives, we used the “cmprsk” package of the statistical software R which allows sub 28 29 175 distribution analysis of competing risks. For the third objective, we used the Stata’s stcrreg package. 30 31

176 http://bmjopen.bmj.com/ 32 33 34 35 177 Independent prognostic variables 36 37 178 Sociodemographic characteristics included in our study were age at entry into the work disability 38 39 179 system, gender and occupational class. The age variable was based on the date of birth and was further

40 on September 29, 2021 by guest. Protected copyright. 41 180 categorised into four groups: 17-39; 40-49; 50-59; and 60+ years. Occupational classes were based on 42 43 44 181 four categories: blue-collar workers, white-collar workers, self-employed people and assisting spouses. 45 46 182 They were recoded into a 3-level variable: blue-collar workers, white-collar workers and self-employed 47 48 183 people. 49 50 51 52 53 54 55 56

57 58 Version 08 March 2017 59 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 8 of 39

1 2 3 184 In total, 39 cancer sites have been identified using the ‘pathology codes’ transmitted by the NIHDI and 4 5 185 6 registered by their ICD-9 codes (Table 2). For the sake of comparability, we translated these into ICD-10 7 8 186 codes and gathered them into 11 cancer sites (Table 2). 9 10 187 The year of entry in the work disability system was a continuous variable ranging from 2007 to 2011. We 11 12 188 decided to recode the year of entry into a two-level variable: 2007-2010 and 2011. This decision is based 13 14 189 on an exploratory analysis that showed significant difference in survival patterns between disability 15 For peer review only 16 190 acquired before or after 2011 (log-rank test=502, df=1, p-value < 0.001) (Figure 3). 17 18 19 20 191 Outcome variables: the three competing events 21 22 192 The outcome variable is the event that causes the end of work disability. We defined three mutually 23 24 193 exclusive events, i.e. competing risks: death, retirement and ability to work. 25 26 194 The status retirement indicates that the worker is definitively out of the labour market due to age and will 27 28 195 receive social benefits until death, while able to work indicates that the cancer-disabled worker was 29 30 31 196 recognised by a health insurer's doctor as able to work. In practice, this might lead to a RTW, to 32 http://bmjopen.bmj.com/ 33 197 unemployment or to a decision to be a stay-at-home spouse. 34 35 198 Those long-term disabled workers who had not experienced any event by the end of follow-up, by. on 31 36 37 199 December 2013, were administratively censored (31%, Table 3). 38 39 200 [Table2, Figures 2 and 3 could be displayed]

40 on September 29, 2021 by guest. Protected copyright. 41 201 42 202 43 203 44 204 45 205 46 206 47 207 48 49 208 50 209 51 210 52 211 53 212 54 213 55 214 56 215 57 216 58 Version 08 March 2017 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 9 of 39 BMJ Open

1 2 3 217 4 218 5 219 6 GROUPS The 11 cancer groups by anatomical location 5-year 5-year Survival rate Frequency 7 relative relative category observed in 8 survival rate survival rate the data 9 men*(%) women*(%) 10 1 Other malignancies and undefined sites, CIS NA 1 247 11 Benign tumours 12 2 Head and neck: lip, oral cavity, nasal cavities, middle ear 50.0 57.0 medium 877 13 and accessory sinuses, pharynx, larynx 14 3 Digestive tract 22.8 22.7 low 2 400

15 EsophagusFor peer review only 257 16 Stomach 218 17 Colon & rectum 1 479 18 Pancreas 209 19 Other malignant neoplasms of digestive organs and 237 20 peritoneum 21 4 Respiratory tract 14.6 19.5 low 1 417 22 23 Trachea and lung 1 404 mesothelioma 13 24 5 Haematological 1 660 25 26 Hodgkin Disease 86.1 85.0 high 263 27 Non-Hodgkin Disease 67.0 68.9 medium 711 28 Acute lymphoid leukemia and Lymphoid leukemia. Other 81.3 76.7 high 161 29 Myeloid leukemia and others 38.5 40.6 low 307 30 6 Breast 78.2 88.0 high 5 511 31 Brest women 5 494 http://bmjopen.bmj.com/ 32 Breast men 17 33 7 Women genital organs 821 34 35 Corpus uterus - cerv. ut. 69.8 medium 273 36 Cervix uteri corp. ut. 79.6 high 147 37 Ovary ovary 54.1 medium 362 38 Others 39 8 Men genital organs 95.3 - high 486

40 on September 29, 2021 by guest. Protected copyright. Prostate 377 41 Testis 94 42 Others 16 43 9 Urinary tract 388 44 Kidney 71.0 0.7 high 147 45 Bladder 56.6 49.2 medium 178 46 Others 63 47 10 Central nervous system (CNS) 22.7 25.8 low 709 48 11 Bone and connective tissue (sarcoma’s) 61.9 59.7 medium 49 Melanoma of the skin 86.2 91.0 high Malignant neoplasms of skin other than melanoma NA 50 Thyroid and other endocrine glands 89.3 94.1 high 51 Other malignancies and undefined sites, invasive 51.5 39.1 medium/low 52 Tumours of uncertain and unspecified behavior NA 53 TOTAL 15 543 54 220 Table 2. The 11 cancer groups used for the analysis. 55 221 *Reference: Belgian Cancer Registry. Cancer Survival in Belgium, 2004-2008. 56 222 57 223 58 Version 08 March 2017 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 10 of 39

1 2 3 4 224 Results 5 6 7 225 Description of the study population 8 9 10 226 No observed workers were lost to follow-up. Table 3 describes the main characteristics of the work- 11 12 227 disabled cancer survivors included in the study. 13 14 228 The majority (77%) of the cancer-disabled workers were aged 40–59 years. 15 For peer review only 16 229 Women were over-represented (62%), younger at entry (median age of 48 vs. 53 years for men) and 17 18 230 mostly white-collar workers (46% vs. 21% for men) or blue-collar workers (43% vs. 60%). The outcome 19 20 231 for the majority of cancer-disabled women was ability to work (44.10%), while for most men the outcome 21 22 23 232 was death (50.86%). 24 25 233 The most frequent cancer site was breast, representing 35% (n= 5,949) of disabled workers, followed by 26 27 234 15% (n=2,400) of digestive tract cancers and 9% (n=1,417) of respiratory tract cancers. 28 29 235 Regarding OC, half of the disabled workers were blue-collar workers, the majority of whom (38.2% of 30 31 236 the total) died by the end of follow-up. White-collar workers (37%) had the shortest median time of work 32 http://bmjopen.bmj.com/ 33 237 disability (1.30 years vs. 1.79 years for the others), and the majority (47.7%) were able to work by the end 34 35 36 238 of follow-up. Self-employed disabled workers represented 13% of the cohort, and the majority (38.2%) 37 38 239 died by the end of follow-up. 39

40 240 By the end of follow-up, 69% of the cohort had experienced one of the three competing events (32.2% on September 29, 2021 by guest. Protected copyright. 41 42 241 died, 2.2% retired and 34.6% were able to work). The other 31% remained disabled, distributed as 43 44 242 follows: 35% of those who entered in 2009 and 2010, 31% of those aged 40-49 years, 27% of the women, 45 46 243 28% of the blue-collar workers and 44% of those with benign or in situ tumours (Table 3). 47 48 49 244 Figures 4a-e show the non-parametric cause-specific cumulative incidences of time to ability to work in 50 51 245 the presence of competing risks. For all prognostic variables, the curves show a steep increase in ability to 52 53 246 RTW within the first two years; later, the curves virtually level off. 54 55 56 57 58 Version 08 March 2017 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 11 of 39 BMJ Open

1 2 3 247 Figures 5a-5 show the box plots of time to any event (death, ability to work or retirement) stratified by 4 5 248 6 each prognostic variable, respectively. 7 8 249 Younger workers (17-39 years) had the highest rates of ability to work at the end of follow-up (Figure 4b) 9 10 250 and relatively short periods of work disability (Figure 5b), mainly due to the ability to work. Older 11 12 251 workers presented the shortest work disability periods (Figure 5b), mainly due to death or retirement 13 14 252 (77%, Table 3). 15 For peer review only 16 253 Women had higher rates of ability to work compared to men (Figure 4c) but spent longer periods in work 17 18 254 19 disability (Figure 5c). White-collar workers had higher rates of work disability and spent less time in it 20 21 255 (Figure 5c). Regarding the cancer sites, workers with breast or haematological cancer had the highest 22 23 256 rates of ability to work by the end of follow-up (Figure 4e), but the longest periods spent on work 24 25 257 disability (Figure 5e). Those with respiratory tract, head and neck, digestive or central nervous system 26 27 258 (CNS) cancers had the lowest rates of ability to work (Figure 4e) and shorter periods of work disability 28 29 259 (Figure 5e), mainly due to death. 30 31

260 http://bmjopen.bmj.com/ 32 Those cancer survivors who entered into the work disability system in 2011 had higher rates of ability to 33 34 261 work by the end of follow-up (Figure 4a) and shorter periods on work disability (Figure 5a). 35 36 262 [Table 3, Figures 4 and 5 could be displayed] 37 38 263 39

40 264 on September 29, 2021 by guest. Protected copyright. 41 42 265 43 44 266 45 46 47 267 48 49 268 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 12 of 39

12 12 ) ) Version 08 March 2017 2017 March 08 Version

(38%) 5874 (62%) 9669 (15.6%) 2421 (32.5%) 5052 (44.7%) 6946 ( 7.21%) 1121 (22.2%) 3454 (24.2%) 3760 (23.4%) 3630 (21.8%) 3388 (8.4 %) 1311

270 271 269 TOTAL n (%) n TOTAL Bone&Con (sarc.)/Skin/Thyroid Bone&Con Central nervous system (CNS) system nervous Central Urinary tract Urinary Men genital organs genital Men Women genital organs genital Women Breast Breast Haematological Haematological Respiratory tract Respiratory Digestive tract Digestive Head & Neck & Head CIS/Ben CIS/Ben Cancer site site Cancer Assisting spouse & & spouse Self-employed Assisting White collar White Blue collar collar Blue Occupational Class Occupational Women Women Men Men Gender Gender >=60 >=60 50-59 50-59 40-49 40-49 17-39 17-39 Age at entry entry Age at 2011 2010 2010 2009 2009 2008 2008 2007 2007 Year of entry Year entry of

1 2 3 4 272 Prediction patterns of the end of work disability (model 1) 5 6 7 273 Results in Table 4 suggest that good prognostic factors for the ability to work for both men and women 8 9 274 are disability experienced after 2011 and white-collar OC. Regarding the 11 cancer sites, men with 10 11 275 haematological or genital organ cancers are the most likely to be able to work. Among women, the cancer 12 13 276 sites with the best chance for ability to work are haematological and breast. 14 15 277 Concerning deathsFor among men, peer disabled workers review with respiratory tract,only CNS, bone & connective tissue 16 17 18 278 cancers are most at risk. Among women, those with respiratory tract, women genital organs, digestive 19 20 279 tract and head and neck cancers are most at risk. 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 14 of 39

14 14

- - - - - 95%CI 0.45,1.24 1.13,3.02 0.03,2.44 0.16,2.05 0.03,0.93 0.21,3.34 0.21,2.42 0.01,0.51 (0.15,3.49) - (0.02,2.01) (0.14,1.17) (54.3,204.8) Retirement

Version 08 March 2017 2017 March 08 Version - - - - - SHR 1 0.75 1.84 1 0.25 0.57 0.18 0.84 0.72 0.05 0.73 0.73 0 0.20 1 0.41 1 105.5

- - SHR Ability to workAbility 1 1.43 1.07 1 0.77 0.69 0.27 1.11 1.42 0.81 0.74 0.41 0.75 1 0.81 0.64 0.49 1 2.03

- -

95%CI (0.96,1.18) (0.80,1.12) (2.66,6.77) (3.26,7.19) (6.71,14.92) (1.40,3.28) (0.85,1.83) (3.33,7.32) (2.20,6.38) (4.26,9.88) (3.50,8.07) (0.93,1.30) (1.22,1.67) (1.70,2.71) (0.96,1.36) Death

- - SHR 1 1.06 0.94 1 4.24 4.84 10.0 2.15 1.24 4.93 3.75 6.48 5.32 1 1.10 1.42 2.15 1 1.14

------

http://bmjopen.bmj.com/ 95%CI (0.82,2.60) (2.02,5.21) (0.12,1.17) (0.19,1.31) (0.05,0.59) (0.19,1.46) (0.46,3.09) (0.21,1.87) (0.03,0.74) (0.12,1.81) (39.73,284.06) 0.07,0.69 BMJ Open Retirement

------SHR 1 106.20 1 1.46 3.24 1 0.37 0.50 0.18 0.52 1.20 0.62 0.14 0.46 1 0.21

Ability to workAbility SHR 1 0.57 0.41 0.45 1 1.49 0.85 1 0.59 0.85 0.34 1.14 0.13 - 1.38 0.70 0.47 0.76 1 1.55

- - For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

95%CI (1.29,1.91) (1.60,2.31) (1.96,3.02) (0.92,1.14) (0.73,0.93) (1.97,3.94) (1.94,3.82) (4.14,8.17) (1.03,2.09) (0.45,3.67) (0.58,1.27) (1.78,3.85) (2.71,5.54) (2.01,4.27) (1.08,1.47) Death

For peer review only - - SHR 1 1.57 1.92 2.43 1 1.02 0.83 1 2.79 2.73 5.81 1.47 1.29 0.86 2.62 3.87 2.93 1 1.26

TableModel 4. sub-distribution1: ratio hazard the for theatstatus endof follow-up.

of entry 282 280 281 >=60 >=60 50-59 50-59 40-49 40-49 Age group Age 19-39 Self-Employed or assisting spouse spouse or assisting Self-Employed Blue Collar Blue Collar Occupational Class White Collar 11: Bone and connective 11:connective and Bone 10 Central Nervous System System 10 Nervous Central 9 Urinary Track 9 Track Urinary 8 Men Genital Organs Genital8 Organs Men 7 Women Genital Organs Women Organs 7 Genital 6 Breast 6 Breast 5 Haematological 5 Haematological 4 Respiratory Track 4 Track Respiratory 3 Digestive track 3 Digestive track 2 Head and Neck and2 Head Neck Cancer site Benign 1 and CIS 2011 2011 2007-2010 2007-2010

1 2 3 4 283 Social inequalities in the work disability of cancer survivors (model 2) 5 6 7 284 In the second model, we stratify by age and gender and allow interactions between both these variables 8 9 285 and OC and survival categories (Table 5). The absence of individuals in certain age categories entering 10 11 286 retirement (17-49 years, Table 2) leads to a convergence issue when modelling the cause-specific hazard 12 13 287 for this type of event and this is therefore not reported. 14 15 288 Table 5 shows that,For among men, peer blue-collar andreview self-employed workers only aged 50-59 years were less likely 16 17 18 289 to be able to work compared to white-collar workers. Similar results were found for blue-collar women 19 20 290 aged 17-39, 40-49 and 50-59. These results translate into larger social inequalities in the 50-59 age group 21 22 291 for both men and women. 23 24 292 Men self-employed were less likely to be able to work than white-collar workers when aged 17-39 or 50- 25 26 293 59 and similarly for women aged 50-59. 27 28 294 29 30 31 295 32 http://bmjopen.bmj.com/ 33 296 34 35 297 36 37 298 38 39 299

40 on September 29, 2021 by guest. Protected copyright. 41 300 42 43 44 301 45 46 302 47 48 303 49 50 304 51 52 305 53 54 306 55 56 57 307 58 Version 08 March 2017 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 16 of 39

1 2 3 Sub-distribution hazard ratio at the end of follow-up 4 5 MEN WOMEN 6 Death Ability to work Death Ability to work 7 17-39 8 Survival rate High 1 1 1 1 9 Medium 1.16 (0.48,2.82) 0.78 (0.47,1.29) 1.69 (1.08,2.65) 0.75 (0.57,0.99) 10 Low 2.63 (1.21,5.72) 0.33 (0.19,0.60) 2.91 (1.99,4.26) 0.38 (0.27,0.53) 11 Missing 2.17 (0.88,5.40) 0.50(0.25,0.98) 1.33 (0.77,2.29) 0.58 (0.42,0.81) 12 Occupational White Collar 1 1 1 1 13 class Blue Collar 0.48(0.20,1.15) 0.86(0.58,1.27) 1.41 (1.02,1.93) 0.58(0.48,0.69) Self-employed 1.43 (0.53,3.84) 0.48 (0.26,0.87) 1.13 (0.60,2.11) 1.05 (0.81,1.38) 14 40-49 15 For peer review only 16 Survival rate High 1 1 1 1 Medium 1.53 (0.74,3.15) 1.22 (0.66,2.25) 3.19 (2.39,4.27) 0.58(0.46,0.72) 17 Low 3.32 (1.67,6.61) 0.63 (0.33,1.20) 5.84 (4.32,7.90) 0.24(0.17,0.35) 18 Missing 0.89 (0.32,2.47) 1.69 (0.82,3.50) 2.96 (2.08,4.22) 0.60 (0.46,0.78) 19 Occupational White Collar 1 1 1 1 20 class Blue Collar 1.12(0.53,2.34) 0.79 (0.42,1.47) 1.42(1.13,1.80) 0.59 (0.53,0.67) 21 Self-employed 1.27(0.47,3.42) 1.74 (0.88,3.43) 0.96(0.62,1.49) 0.84 (0.70,1) 22 50-59 23 Survial rate High 1 1 1 1 24 Medium 1.39(0.93,2.10) 0.82 (0.53,1.24) 3.21 (2.53,4.07) 0.55 (0.44,0.70) 25 Low 3.21(2.15,4.77) 0.38(0.24,0.61) 6.01 (4.73,7.64) 0.21 (0.14,0.31) 26 Missing 1.47(0.86,2.52) 0.76 (0.42,1.36) 2.22 (1.62,3.04) 0.75 (0.57,0.98) 27 Occupational White Collar 1 1 1 1 28 class Blue Collar 0.88 (0.57,1.34) 0.60 (0.39,0.92) 1.07 (0.86,1.34) 0.75 (0.65,0.86) Self-employed 0.75 (0.45,1.24) 0.57 (0.33,0.96) 1.34 (0.98,1.83) 0.69 (0.56,0.86) 29 30 ≥60 31 Survival rate High 1 1 1 1 Medium 1.13 (0.51,2.52) 2.26 (0.90,5.68) 2.85 (1.52,5.38) 0.91 (0.47,1.76) http://bmjopen.bmj.com/ 32 Low 3.16 (1.52,6.55) 0.72 (0.25,2.06) 3.92 (1.91,8.03) 0.29 (0.10,0.86) 33 Missing 2.14 (0.69,6.66 1.21 (0.29,4.99) 2.68(0.95,7.54) 0.41 (0.09,1.89) 34 White Collar 1 1 1 1 35 Occupational class Blue Collar 1.39 (0.61,3.13) 1.57 (0.60,4.08) 1.06 (0.54,2.09) 0.80 (0.46,1.39) 36 Self-employed 0.55 (0.23,1.34) 1.19 (0.46,3.10) 1.09 (0.55,2.18) 0.60 (0.32,1.12) 37 308 Table 5. Sub-distribution hazard ratio based on the Fine and Gray model stratified by age and gender, and 38 309 adjusted for year of entry. 39 310 1Low survival rates: "Esophagus", "Stomach", "Colon & rectum", "Pancreas", "Oth. Mal. Neop. of digestive organs and

40 311 peritoneum", "Mesothelioma", "Trachea and lung", "Myeloid and others", "CNS", "Oth. malignancies and undefined sites, on September 29, 2021 by guest. Protected copyright. 41 312 invasive" 42 313 2Medium survival rates: "Lip and oral cavity, nasal cavities, middle ear and accessory sinuses, pharynx, larynx","Non 43 314 Hodgkin's Disease", "Uterus", "Cervix","Ovary","Oth. mal. neop. Women genitals", "Cervix", "Ovary", "Oth. mal. neop. of 44 315 women genitals", "Bladder", "Urinary system other than bladder", "Bone and connective tissue" 316 3High survival rates: "Hodgkin Disease","Acute lymphoid leukemia and lymphoid leukemia, other", "Breast 45 317 women","Uterus","Kidney","Melanoma of the skin", "Thyroid and other endocrine glands" 46 318 4Missing survival rates:"Benign tumour", "Mal. neop. of skin other than melanoma", "Oth. malignancies and undefined sites, 47 319 CIS", "Tumours of uncertain and unspecified behavior" 48 320 49 50 321 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 17 of 39 BMJ Open

1 2 3 4 322 DISCUSSION 5 6 323 In this study, we aimed to identify the factors that influence the reason for exiting the long-term work 7 8 324 disability system and the length of work disability among cancer survivors. 9 10 325 To achieve this, we first measured the association between the duration of work disability and age, 11 12 13 326 gender, occupational class, the year of entry into the work disability system and 11 cancer sites. Secondly, 14 15 327 we estimated theFor distribution peer of three competing review reasons for exiting only the work disability system; and 16 17 328 thirdly, we investigated social inequalities in work disability among cancer survivors. 18 19 329 20 21 330 As not many of the population-based studies in this field include several cancer sites or use competing 22 23 331 risk analysis, making comparisons is not easy. Moreover, our follow-up starts one year after the first day 24 25 26 332 of sickness absence, i.e. that we only include long-term disabled workers. However, the impact of these 27 28 333 determining factors on labour market participation has been tested in previous studies [50]. Results 29 30 334 indicate that, overall, older age at entry into the work disability system and man gender are both factors 31 32 335 that decrease the chance of being economically active. Our results show that an older age (>60 years) http://bmjopen.bmj.com/ 33 34 336 increase the risk of dying or retiring, and that workers aged 40-49 were the most likely to remain disabled 35 36 337 for a long period (Table 3). Man gender did indeed reduce the likelihood of being able to work but 37 38 39 338 women experienced longer periods within the work disability system overall.

40 on September 29, 2021 by guest. Protected copyright. 41 339 42 43 340 Regarding the cancer sites, we found a strong association between respiratory tract, head and neck and 44 45 341 digestive tract cancers and death. The first two include smoking-related cancer sites [51], which represent 46 47 342 major sources of work disability and death in the working age population. 48 49 343 Other studies have compared different cancer sites to assess their association with employment status 50 51 52 344 after cancer diagnosis. In line with our results, workers with respiratory and women genital cancers 53 54 345 present smaller proportions of employment than workers with breast or haematological cancers, mainly 55 56 346 due to poor self-reported health status [26,27,52]. 57 58 Version 08 March 2017 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 18 of 39

1 2 3 347 4 5 348 6 In line with previous research, blue-collar and self-employed (Torp et al. 2016) workers are less likely to 7 8 349 be able to work after cancer compared to white-collar workers, especially those aged 50-59 years. [26]. 9 10 350 According to other research, these social inequalities could be explained by more demanding working 11 12 351 conditions [53], later stage of cancer at diagnosis, differences in treatment [54] and lower participation in 13 14 352 rehabilitation services [55] among blue-collar workers. OC is also strongly associated with the income 15 For peer review only 16 353 level, which may represent an incentive to RTW when it is significantly different (higher) than sickness 17 18 354 19 absence benefits [36]. 20 21 355 A different impact of the OC on the risk of work disability according to age and gender has been shown in 22 23 356 a Norwegian county, where young workers with blue-collar jobs are more at risk than older men [56]. The 24 25 357 association between age and RTW has been reported with contradictory results in the literature, but the 26 27 358 majority found higher age to be associated with later RTW or reduced chance of employment [34]. Our 28 29 359 results show that, for Belgian cancer survivors, the opposite is found, with a larger impact of OC from the 30 31

360 http://bmjopen.bmj.com/ 32 age of 40 years onwards, compared to younger counterparts. 33 34 361 35 36 362 Demographic changes and the rising retirement age will increase the number of disabled workers and the 37 38 363 length of work disability; combined with the effects of the economic crisis (i.e. greater competition and 39

40 364 emphasis on maximum performance) this will worsen the situation if we do not implement measures, on September 29, 2021 by guest. Protected copyright. 41 42 365 interventions and rehabilitation programmes to better (re)integrate disabled workers in the labour market 43 44 366 45 [5]. 46 47 367 The measure introduced by the Belgian government by the end of 2010 seems to have had an impact 48 49 368 already, as the workers who entered the work disability system in 2011 showed better outcomes than the 50 51 369 others. In 2011, a new measure was implemented, allowing disabled workers to resume work without 52 53 370 prior agreement of the health insurer's medical advisor. 54 55 56 57 58 Version 08 March 2017 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 19 of 39 BMJ Open

1 2 3 371 Further studies need to be carried out in future to confirm this trend. However, at the end of follow-up, 4 5 372 6 only 34.6% of the cancer survivors were able to work and 31% were administratively censored, remaining 7 8 373 disabled. 9 10 374 Based on our results, key features of (work) rehabilitation programmes can be drawn. The non-parametric 11 12 375 cause-specific cumulative incidence of time to ability to work (Figures 4) suggests that interventions 13 14 376 should be planned and implemented within the 2 years after the cancer diagnosis. Differences in age and 15 For peer review only 16 377 gender imply tailoring of and specific attention to the needs of young workers and women. The 17 18 378 19 association of the cancer site with the length of disability suggests that the awareness of oncologists who 20 21 379 treat breast cancer, digestive track cancers and head and neck cancers, should be raised on the RTW and 22 23 380 that they need to be involved in the assessment and management of symptoms. 24 25 381 The negative association of being blue-collar or self-employed calls for the revision of employment 26 27 382 policies for these high-risk groups, with e.g. the creation of incentives for employers to adjust the working 28 29 383 conditions of their sick-listed blue-collar workers. 30 31

384 http://bmjopen.bmj.com/ 32 33 34 385 Strengths, limitations and needs for further research 35 36 386 The main strength of this study is the representativeness of the data and the generalisability of our results. 37 38 387 We included in the analysis all Belgian workers disabled due to cancer between 2007- 2011, excluding 39 40 388 civil servants and individuals for whom we detected coding errors. on September 29, 2021 by guest. Protected copyright. 41 42 43 389 44 45 390 In most work disability studies, survival analyses are used to estimate the time to an event of interest. The 46 47 391 end of work disability is, however, more complex than this, and may be caused by multiple factors. 48 49 392 Therefore, the use of competing risks analysis becomes appropriate to avoid over- or under-estimating the 50 51 393 probability of experiencing each event [57]. This model is still rarely used in work disability studies and 52 53 394 its use should be encouraged. 54 55 56 395 57 58 Version 08 March 2017 59 60 19

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 20 of 39

1 2 3 396 Regarding the objective of predicting disability, the two models showed their capacity to and 4 5 397 6 effectiveness in predicting the length and the reasons for ending work disability among Belgian cancer 7 8 398 survivors. Our second model presents original findings, using the survival rates to identify social 9 10 399 inequalities. 11 12 400 Nevertheless, differences still exist among workers of the same age, gender, OC and cancer site; a more 13 14 401 complex model with information on treatment and symptoms [58] and on the working environment [59] 15 For peer review only 16 402 is thus called for. This is feasible in the future, e.g. by linking data from Cancer Registries to data on 17 18 403 19 employment and socioeconomic status. Results could be used to develop rehabilitation programmes for 20 21 404 cancer survivors similar to those that already exist in other countries [60-63]. 22 23 405 24 25 406 While our paper focuses on work disability among cancer survivors in Belgium, it is important to realise 26 27 407 that the methods and principles used are generic, and applicable for addressing work disability as a whole. 28 29 408 Therefore, this report is also relevant for other conditions and SSS. This paper contributes towards closing 30 31

409 http://bmjopen.bmj.com/ 32 the knowledge gap on the transition among cancer survivors from long-term work disability to ability to 33 34 410 RTW. Linking these important results to predictions of cancer incidence should make it possible to plan 35 36 411 cancer rehabilitation needs and related sickness absence benefits. 37 38 39 40 412 Contributorship statement on September 29, 2021 by guest. Protected copyright. 41 42 413 43 44 414 Régine Kiasuwa Mbengi, MPH, PhD Student 45 415 Designed the study 46 416 Requested/collected the data 47 417 48 Performed the analysis 49 418 Wrote the paper 50 419 51 420 Mioara Alina Nicolae, MSc,PhD 52 421 Provided statistical support to build the models 53 422 Provided advice on the writing of the sections on methods and results 54 55 423 56 424 Prof. Els Goetghebeur, MSc,PhD 57 425 Provided statistical support to build the models 58 Version 08 March 2017 59 60 20

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 21 of 39 BMJ Open

1 2 3 426 Provided advice on the writing of the sections on methods and results 4 5 427 6 428 Renee Otter, MD, PhD 7 429 Provided clinical expertise support to build cancer categories 8 430 Substantially contributed to the writing 9 431 10 432 11 Katrien Mortelmans, MD, PhD 12 433 Substantially contributed to the writing 13 434 14 435 Sarah Missine, MSc, PhD 15 436 Provided adviceFor and support peer to integrate review the inequalities perspective only in the paper and in the 16 437 statistical model 17 18 438 Supported the writing of the paper 19 439 20 440 Marc Arbyn, MD 21 441 Substantially contributed to the preparation of the tables and figures and the abstract section 22 442 23 24 443 Prof. Catherine Bouland, MsC, PhD 25 444 As PhD co-supervisor for Miss Kiasuwa, Prof. C. Bouland substantially contributed to the 26 445 preparation of the study, the design and the revision of the paper 27 446 28 447 Prof. Christophe de Brouwer, MD, PhD 29 448 As the PhD supervisor for Miss Kiasuwa, Prof. C. de Brouwer substantially contributed to the 30 31 449 preparation of the study, the design and the revision of the paper 32 450 http://bmjopen.bmj.com/ 33 451 34 452 Funding and competing interests 35 36 37 453 This study has been funded by the National Institute for Health and Disability Insurance. We 38 39 454 have read and understood the BMJ policy on declaration of interests and declare that we have no

40 on September 29, 2021 by guest. Protected copyright. 41 42 455 competing interests. 43 44 456 This article does not contain any studies with human participants or animals performed by any of 45 46 457 the authors. 47 48 49 458 50 51 459 Data sharing statement 52 53 54 460 The complete and anonymised dataset used for this study can be made available by the 55 56 461 corresponding author for researchers interested in comparative studies. This request would be 57 58 Version 08 March 2017 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 22 of 39

1 2 3 462 subject of approval from the Belgian National Institute for Health and Disability Insurance, the 4 5 6 463 Scientific Institute of Public Health and the Université Libre de Bruxelles. 7 8 9 10 464 Figures legends 11 12 465 Figure 1. The Belgian social security scheme related to work disability 13 466 14 15 467 Figure 2. FlowchartFor of the numberpeer of workers review disabled because ofonly cancer between 2007 and 2011 in 16 17 468 Belgium. 18 469 19 20 470 Figure 3. Kaplan-Meier estimator for the time in work disability, stratified by the year of entrance 21 471 22 into work disability. 23 472 24 25 473 Figures 4 a-e. Non-parametric cumulative incidence of ability to work by the end of follow-up, 26 474 stratified per a) year of entry, b) age at entry, c) gender, d) occupational class and e) cancer sites. 27 28 475 29 30 476 Figures 5 a-e. Box plots of time to any of the three events (death, retirement, ability to work), by 31 477 gender, age group, year of entrance and occupational class. 32 http://bmjopen.bmj.com/ 33 478 34 479 35 36 480 37 38 481 39

40 482 on September 29, 2021 by guest. Protected copyright. 41 42 483 43 44 484 45 46 485 47 486 48 49 487 50 51 488 52 489 53 54 490 55 491 56 57 492 58 Version 08 March 2017 59 60 22

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 23 of 39 BMJ Open

1 2 3 493 Reference List 4 5 6 494 7 8 495 1. Amir Z, Wilson K, Hennings J, Young A: The meaning of cancer: implications for family finances 9 496 and consequent impact on lifestyle, activities, roles and relationships. Psychooncology 2012, 21: 10 497 1167-1174. 11 12 498 2. Daniel C.Lustig: Causal Relationships Between Poverty and Disability. Rehabilitation Counseling 13 499 Bulletin 2007, 50: 194-202. 14 15 For peer review only 16 500 3. Schuring M, Robroek SJ, Otten FW, Arts CH, Burdorf A: The effect of ill health and socioeconomic 17 501 status on labor force exit and re-employment: a prospective study with ten years follow-up in 18 502 the Netherlands. Scand J Work Environ Health 2013, 39: 134-143. 19 20 503 4. Reichard A., Stransky M., Phillips K., Drum C., Mc Clain M.: Does Type of Disability Matter to 21 504 Public Health Policy Practice? Californian Journal of Health Promotion 2015, 13: 25-36. 22 23 505 5. OECD. Sickness, Disability and Work. Breaking the Barriers. 2010. 24 506 Ref Type: Report 25 26 27 507 6. Luengo-Fernandez R, Leal J, Gray A, Sullivan R: Economic burden of cancer across the European 28 508 Union: a population-based cost analysis. Lancet Oncol 2013, 14: 1165-1174. 29 30 509 7. Betsch N: Kankerpatiënten en werkhervatting: rechten en plichten. Instituut Voor Arbeidsrecht, KU 31 510 Leuven; 2013. 32 http://bmjopen.bmj.com/ 33 511 8. Desiron HA, Donceel P, de RA, Van HE: A conceptual-practice model for occupational therapy to 34 512 facilitate return to work in breast cancer patients. J Occup Rehabil 2013, 23: 516-526. 35 36 37 513 9. Desiron HA, Donceel P, Godderis L, Van HE, de RA: What is the value of occupational therapy in 38 514 return to work for breast cancer patients? A qualitative inquiry among experts. Eur J Cancer Care 39 515 (Engl ) 2015, 24: 267-280.

40 on September 29, 2021 by guest. Protected copyright. 41 516 10. Desiron HA, Crutzen R, Godderis L, Van HE, de RA: Bridging Health Care and the Workplace: 42 517 Formulation of a Return-to-Work Intervention for Breast Cancer Patients Using an Intervention 43 518 Mapping Approach. J Occup Rehabil 2016. 44 45 46 519 11. Rommel W. Werken na kanker:welke problemen ervaren (ex-)patiënten die het werk hervatten? 47 520 2012. Vlaams Liga Tegen Kanker. 48 521 Ref Type: Report 49 50 522 12. Tiedtke C, Donceel P, Knops L, Desiron H, Dierckx de CB, de RA: Supporting return-to-work in the 51 523 face of legislation: stakeholders' experiences with return-to-work after breast cancer in Belgium. 52 524 J Occup Rehabil 2012, 22: 241-251. 53 54 55 525 13. Tiedtke C, Dierckx de CB, Donceel P, de RA: Workplace support after breast cancer treatment: 56 526 recognition of vulnerability. Disabil Rehabil 2014, 1-7. 57 58 Version 08 March 2017 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 24 of 39

1 2 3 527 14. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 4 528 Chapter 12. Future need for research. Scand J Public Health Suppl 2004, 63: 256-258. 5 6 7 529 15. Alexanderson K, Hensing G: More and better research needed on sickness absence. Scand J Public 8 530 Health 2004, 32: 321-323. 9 10 531 16. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 11 532 Chapter 1. Aim, background, key concepts, regulations, and current statistics. Scand J Public 12 533 Health Suppl 2004, 63: 12-30. 13 14 534 17. INAMI. Rapport annuel. Satistiques des indemnités. 2014. 15 For peer review only 16 535 Ref Type: Report 17 18 536 18. Coleman MP, Gatta G, Verdecchia A, Esteve J, Sant M, Storm H et al.: EUROCARE-3 summary: 19 537 cancer survival in Europe at the end of the 20th century. Ann Oncol 2003, 14 Suppl 5: v128-v149. 20 21 538 19. Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, Rosso S, Coebergh JW, Comber H et al.: Cancer 22 539 incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 23 540 2013, 49: 1374-1403. 24 25 26 541 20. Pryce J, Munir F, Haslam C: Cancer survivorship and work: symptoms, supervisor response, co- 27 542 worker disclosure and work adjustment. J Occup Rehabil 2007, 17: 83-92. 28 29 543 21. Kennedy F HCMFPJ: Returning to work following cancer: a qualitative exploratory study into the 30 544 experience of returning to work following cancer. Eur J Cancer Care 2007, 16: 17-25. 31 32 545 22. Mehnert A: Employment and work-related issues in cancer survivors. Crit Rev Oncol Hematol http://bmjopen.bmj.com/ 33 546 2011, 77: 109-130. 34 35 36 547 23. Mehnert A, de BA, Feuerstein M: Employment challenges for cancer survivors. Cancer 2013, 119 37 548 Suppl 11: 2151-2159. 38 39 549 24. Verbeek J, Spelten E, Kammeijer M, Sprangers M: Return to work of cancer survivors: a

40 550 prospective cohort study into the quality of rehabilitation by occupational physicians. Occup on September 29, 2021 by guest. Protected copyright. 41 551 Environ Med 2003, 60: 352-357. 42 43 552 25. Dal ML, Guzzinati S, Buzzoni C, Capocaccia R, Serraino D, Caldarella A et al.: Long-term survival, 44 45 553 prevalence, and cure of cancer: a population-based estimation for 818 902 Italian patients and 46 554 26 cancer types. Ann Oncol 2014, 25: 2251-2260. 47 48 555 26. Kiasuwa MR, Otter R, Mortelmans K, Arbyn M, Van OH, Bouland C et al.: Barriers and 49 556 opportunities for return-to-work of cancer survivors: time for action-rapid review and expert 50 557 consultation. Syst Rev 2016, 5: 35. 51 52 558 27. Hewitt M, Rowland JH, Yancik R: Cancer survivors in the United States: age, health, and disability. 53 54 559 J Gerontol A Biol Sci Med Sci 2003, 58: 82-91. 55 56 560 28. Amir Z, Brocky J: Cancer survivorship and employment: epidemiology. Occup Med (Lond) 2009, 57 561 59: 373-377. 58 Version 08 March 2017 59 60 24

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 25 of 39 BMJ Open

1 2 3 562 29. Collins C.G, Ottati A., Feuerstein M: Cancer Survivorship. In Handbook of Work Disability. 4 563 Prevention and Management. Springer edition. Edited by Loisel P, Anema J.R. New York: 2013:289- 5 6 564 302. 7 8 565 30. Tiedtke C, de RA, Dierckx de CB, Christiaens MR, Donceel P: Experiences and concerns about 9 566 'returning to work' for women breast cancer survivors: a literature review. Psychooncology 2010, 10 567 19: 677-683. 11 12 568 31. Tiedtke C, Dierckx de CB, de RA, Christiaens MR, Donceel P: Breast cancer treatment and work 13 569 disability: patient perspectives. Breast 2011, 20: 534-538. 14 15 For peer review only 16 570 32. van Muijen P., Duijts SF, van der Beek AJ, Anema JR: Prognostic factors of work disability in sick- 17 571 listed cancer survivors. J Cancer Surviv 2013, 7: 582-591. 18 19 572 33. Grunfeld EA, Cooper AF: A longitudinal qualitative study of the experience of working following 20 573 treatment for gynaecological cancer. Psychooncology 2012, 21: 82-89. 21 22 574 34. van Muijen P., Weevers NL, Snels IA, Duijts SF, Bruinvels DJ, Schellart AJ et al.: Predictors of return 23 575 to work and employment in cancer survivors: a systematic review. Eur J Cancer Care (Engl ) 2013, 24 25 576 22: 144-160. 26 27 577 35. Sterud T, Johannessen HA: Do work-related mechanical and psychosocial factors contribute to 28 578 the social gradient in long-term sick leave: a prospective study of the general working 29 579 population in Norway. Scand J Public Health 2014, 42: 329-334. 30 31 580 36. Piha K, Laaksonen M, Martikainen P, Rahkonen O, Lahelma E: Interrelationships between http://bmjopen.bmj.com/ 32 581 education, occupational class, income and sickness absence. Eur J Public Health 2010, 20: 276- 33 34 582 280. 35 36 583 37. Melchior M., Krieger N, Kawachi I., et al.: Work Factors and Occupational Class Disaprities in 37 584 Sickness Absence: Findings From the GAZEL Cohort Study. American Journal of Public Health 38 585 2005, 95: 1206-1212. 39

40 586 38. Melchior M, Goldberg M, Krieger N, Kawachi I, Menvielle G, Zins M et al.: Occupational class, on September 29, 2021 by guest. Protected copyright. 41 587 occupational mobility and cancer incidence among middle-aged men and women: a prospective 42 43 588 study of the French GAZEL cohort*. Cancer Causes Control 2005, 16: 515-524. 44 45 589 39. Melchior M, Krieger N, Kawachi I, Berkman LF, Niedhammer I, Goldberg M: Work factors and 46 590 occupational class disparities in sickness absence: findings from the GAZEL cohort study. Am J 47 591 Public Health 2005, 95: 1206-1212. 48 49 592 40. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 50 593 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 51 52 53 594 41. Lindbohm ML, Kuosma E, Taskila T, Hietanen P, Carlsen K, Gudbergsson S et al.: Early retirement 54 595 and non-employment after breast cancer. Psychooncology 2014, 23: 634-641. 55 56 57 58 Version 08 March 2017 59 60 25

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 26 of 39

1 2 3 596 42. Carlsen K, Hoybye MT, Dalton SO, Tjonneland A: Social inequality and incidence of and survival 4 597 from breast cancer in a population-based study in Denmark, 1994-2003. Eur J Cancer 2008, 44: 5 6 598 1996-2002. 7 8 599 43. Carlsen K, Dalton SO, Diderichsen F, Johansen C: Risk for unemployment of cancer survivors: A 9 600 Danish cohort study. Eur J Cancer 2008, 44: 1866-1874. 10 11 601 44. Gerkens S, Merkur S: Belgium: Health system review. Health Syst Transit 2010, 12: 1-266, xxv. 12 13 602 45. Kiasuwa Mbengi R. La réinsertion socioprofessionnelle des patients atteints de cancer. Exposé des 14 603 positions. 2014. 15 For peer review only 16 604 Ref Type: Online Source 17 18 605 46. Onkelinx L. Toespraak van de Minister van Volksgezondheid en Sociale Zaken op het Symposium 19 606 Evaluatie van het Kankerplan. 2012. 20 607 Ref Type: Report 21 22 608 47. de Rijk A.: Work Disability Theories: A Taxonomy for Researchers. In Handbook of Work 23 609 Disability. Prevention and Managment. Edited by Loisel P., Anema J.R. New York: Springer; 24 25 610 2013:475-499. 26 27 611 48. Jason P.Fine and Robert J.Gray: A Proportional Hazards Model for the Subdistribution of a 28 612 Competing Risk. Journal of the American Statistical Association 1999, 94: 496-509. 29 30 613 49. Belgian Cancer Registry. Cancer Survival in Belgium. 2004-2008. 1-109. 2012. 31 614 Ref Type: Report 32 http://bmjopen.bmj.com/ 33 615 50. Rijken M, Spreeuwenberg P, Schippers J, Groenewegen PP: The importance of illness duration, 34 35 616 age at diagnosis and the year of diagnosis for labour participation chances of people with 36 617 chronic illness: results of a nationwide panel-study in The Netherlands. BMC Public Health 2013, 37 618 13: 803. 38 39 619 51. Regional variation in incidence for smoking and alcohol related cancers in Belgium. Edited by Kris

40 620 Henau, Elizabeth Van Eycken, Geert Silverman, Eero Pukkala. Cancer Epidemiology 39[1], 55-65. on September 29, 2021 by guest. Protected copyright. 41 621 2016. 42 43 622 Ref Type: Generic 44 45 623 52. Mehnert A, Koch U: Predictors of employment among cancer survivors after medical 46 624 rehabilitation--a prospective study. Scand J Work Environ Health 2013, 39: 76-87. 47 48 625 53. Ervasti J, Kivimaki M, Dray-Spira R, Head J, Goldberg M, Pentti J et al.: Socioeconomic gradient in 49 626 work disability in diabetes: evidence from three occupational cohorts. J Epidemiol Community 50 627 Health 2016, 70: 125-131. 51 52 53 628 54. Woods LM, Rachet B, Coleman MP: Origins of socio-economic inequalities in cancer survival: a 54 629 review. Ann Oncol 2006, 17: 5-19. 55 56 630 55. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 57 631 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 58 Version 08 March 2017 59 60 26

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 27 of 39 BMJ Open

1 2 3 632 56. Krokstad S, Johnsen R, Westin S: Social determinants of disability pension: a 10-year follow-up of 4 633 62 000 people in a Norwegian county population. Int J Epidemiol 2002, 31: 1183-1191. 5 6 7 634 57. Noordzij M, Leffondre K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ: When do we need 8 635 competing risks methods for survival analysis in nephrology? Nephrol Dial Transplant 2013, 28: 9 636 2670-2677. 10 11 637 58. Glimelius I, Ekberg S, Linderoth J, Jerkeman M, Chang ET, Neovius M et al.: Sick leave and 12 638 disability pension in Hodgkin lymphoma survivors by stage, treatment, and follow-up time-a 13 639 population-based comparative study. J Cancer Surviv 2015. 14 15 For peer review only 16 640 59. Bradley CJ, Bednarek HL: Employment patterns of long-term cancer survivors. Psychooncology 17 641 2002, 11: 188-198. 18 19 642 60. Scott DA Mills M, Black A, Cantwell M, Campbell A, Cardwell CR, Porter S et al.. Multidimensional 20 643 rehabilitation programmes for adult cancer survivors (Review). Cochrane.Database.Syst.Rev. [3]. 21 644 2016. 22 645 Ref Type: Generic 23 24 25 646 61. Mewes JC, Steuten LM, Groeneveld IF, de Boer AG, Frings-Dresen MH, IJzerman MJ et al.: Return- 26 647 to-work intervention for cancer survivors: budget impact and allocation of costs and returns in 27 648 the Netherlands and six major EU-countries. BMC Cancer 2015, 15: 899. 28 29 649 62. de Boer AG, Taskila TK, Tamminga SJ, Feuerstein M, Frings-Dresen MH, Verbeek JH: Interventions 30 650 to enhance return-to-work for cancer patients. Cochrane Database Syst Rev 2015, 9: CD007569. 31 http://bmjopen.bmj.com/ 32 651 63. Zaman AC, Bruinvels DJ, de Boer AG, Frings-Dresen MH: Supporting cancer patients with work- 33 34 652 related problems through an oncological occupational physician: a feasibility study. Eur J Cancer 35 653 Care (Engl ) 2015. 36 654 37 655 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 08 March 2017 59 60 27

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 28 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure3_The Kaplan-Meier estimate for the time into work disability on September 29, 2021 by guest. Protected copyright. 41 42 127x127mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 31 of 39 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 4c_Cumulative incidence of ability to work stratified by gender.jpg on September 29, 2021 by guest. Protected copyright. 41 42 45x45mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 34 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 Figure 5a-d_Boxplot_for_time_to_event_by_year, age, gender and occupational class 38 39 59x53mm (300 x 300 DPI)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

Assessing Factors Associated with Long-Term Work Disability After Cancer in Belgium. A Population-Based Cohort Study Using Competing Risks Analysis with a 7-year Follow-Up

For peer review only Journal: BMJ Open

Manuscript ID bmjopen-2016-014094.R3

Article Type: Research

Date Submitted by the Author: 18-Aug-2017

Primary Subject Epidemiology Heading:

Secondary Subject Heading: Public health on September 29, 2021 by guest. Protected copyright.

Epidemiology < ONCOLOGY, ONCOLOGY, Organisation of health services < Keywords: HEALTH SERVICES ADMINISTRATION & MANAGEMENT

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 1 of 42 BMJ Open

1 2 3 4 1 Assessing Factors Associated with Long-Term Work Disability After 5 2 Cancer in Belgium. A Population-Based Cohort Study Using 6 7 3 Competing Risks Analysis with a 7-year Follow-Up 8 9 4 R. Kiasuwa Mbengia,c, A.M Nicolaieb , Els Goetghebeurb, R. Ottera, K. Mortlemansd, S. Missinnea, 10 e c c 11 5 M. Arbyn , C. Bouland and C. de Brouwer 12 6 a Belgian Cancer Centre, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 13 7 b Stat-Gent CRESCENDO, University of Ghent 14 8 c Research Centre for Environmental and Occupational Health, Brussels School of Public Health, Université Libre de Bruxelles (ESP-ULB) 15 9 d Dr. Katrien Mortelmans,For PhD Consulting peer (KaMoCo) review only 16 10 e Unit Cancer Epidemiology, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 17 11 18 12 Corresponding author: Régine Kiasuwa Mbengi 19 13 [email protected] 14 20 14 rue Juliette Wytsman, 1050 Brusss - Belgium 15 Office: +32 2 642 57 65 21 16 Mobile: +32 479 3926 58 22 23 24 25 17 ABSTRACT 26 18 Objectives 27 19 The number of workers with cancer is increasing dramatically worldwide. One of the main priorities is to 28 20 preserve their quality of life and the sustainability of social security systems. We have carried out this 29 21 study in order to assess factors associated with the ability to work after cancer. Such insight should help 30 31 22 with the planning of rehabilitation needs and tailored programmes. 32 23 Participants http://bmjopen.bmj.com/ 33 24 We conducted this register-based cohort study using individual data from the Belgian Disability 34 25 Insurance. Data on 15,543 socially insured Belgian people who entered into the long-term work disability 35 26 between 2007 and 2011 due to cancer were used. 36 27 Primary and secondary outcome measures 37 28 We estimated the duration of work disability using Kaplan Meier and the cause-specific cumulative 38 29 incidence of ability to work stratified by age, gender, occupational class and year of entering the work 39 30 disability system for 11 cancer sites using the Fine and Gray model allowing for competing risks. 40 31 Results on September 29, 2021 by guest. Protected copyright. 41 32 The overall median time of work disability was 1.59 years (95%CI [1.52-1.66]), ranging from 0.75 to 42 33 4.98 years. By the end of follow-up, more than one-third of the disabled cancer survivors were able to 43 44 34 work (35%). While a large proportion of the women were able to work at the end of follow-up, the men 45 35 who were able to work could do so sooner. Women, white-collar, younger and having haematological, 46 36 men genital or breast cancers were the most likely to be able to RTW. 47 37 Conclusion 48 38 Good prognostic factors for the ability to work were youth, woman, white-collar, and having breast, men 49 39 genital or haematological cancers. 50 40 Reviewing our results together with the cancer incidence predictions up to 2025 offers a high value for 51 41 social security and rehabilitation planning, and for ascertaining patients’ perspectives. 52 42 53 43 Key words: sickness absence, cancer survivors, competing risks, predictive model, social inequalities 54 44 55 45 Strengths and limitations of the study methods: 56 57 58 Version 07 July 2017 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 2 of 42

1 2 3 46 • (Good) External validity: we used a population-based dataset without loss of follow-up; the 4 47 external validity is therefore largely not limited and the study offers high value (when linked with 5 6 48 cancer incidence predictions) for the planning of rehabilitation needs for cancer patients, up to the 7 49 year 2025; 8 50 • Use of competing risk analysis: competing risks were added to the traditional survival analysis in 9 51 order to respect the complexity of the outcomes. This is still rarely done in disability studies; 10 52 • Incomplete model: the lack of information on treatments and job demands: limited the capacity of 11 53 our model to (1) support the identification of a precise risk profile and (2) to tailor return-to-work 12 54 (RTW) interventions 13 55 14 56 15 For peer review only 16 57 BACKGROUND 17 58 18 The direct and indirect effects of work disability represent a significant burden for people who are absent 19 20 59 due to sickness, and to their families and their employers [1]. Long-term work disability may lead to 21 22 60 social exclusion, deprivation or economic insecurity [2], as well as poor health [3]. The negative impact 23 24 61 of work disability on both social and health status is of high importance for public health [4] but studies 25 26 62 identifying those cancer survivors who are at risk of experiencing long-term work disability and 27 28 63 identifying the avoidable proportion of work disability are lagging behind. 29 30 31 64 32 65 Work disability imposes significant costs on society [5,6] with up to 5% of GDP in Organisation for http://bmjopen.bmj.com/ 33 34 66 Economic Co-operation and Development (OECD) countries being spent on disability benefits [5]. In 35 36 67 37 2010, the OECD published a report describing the barriers to (re)integration in the labour market for 38 39 68 people with disability (i.e. greater competition, more demanding workload and work pressure) [5]. The

40 on September 29, 2021 by guest. Protected copyright. 41 69 report also describes the underlying social and economic tragedies. As the results for Belgium were poor, 42 43 70 with a decrease in the number of people with disabilities employed over the past decade, authorities and 44 45 71 social security administrators have been looking for measures or interventions to reverse the trend. A 46 47 72 number of studies have been performed to support the authorities, but these are mainly qualitative and are 48 49 73 50 based on small samples of cancer survivors [7-13]. 51 52 74 53 54 55 56 57 58 Version 07 July 2017 59 60 2

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 3 of 42 BMJ Open

1 2 3 75 Insurance medicine researchers and epidemiologists acknowledge differences between diagnoses in terms 4 5 76 6 of the duration of work disability [14,15]. Overall, the leading causes of work disability are 7 8 77 musculoskeletal disorders and mental health problems, which have been widely studied [16]. 9 10 78 In Belgium, cancer is the fifth greatest cause of work disability, with 18,462 people on work disability 11 12 79 due to cancer in 2013 (6.2% of all workers on work disability in Belgium) [17] (Table 1). Each year, more 13 14 80 than 25,000 Belgian inhabitants of working age (20-64 years) are diagnosed with cancer. 15 For peer review only 16 81 Over the last decade, cancer treatments in middle and high income countries have greatly improved, 17 18 82 19 leading to increased rates of cancer survival [18,19]. Despite these improved survival rates, a cancer 20 21 83 diagnosis still causes great distress among individuals and their relatives [20], and is associated with work 22 23 84 disability or death by their colleagues and supervisors [7,21-24]. 24 25 85 This automatic association of cancer with death is becoming less and less accurate, however, as was 26 27 86 notably demonstrated in the study by Dal Maso et al. [25] that a quarter of Italian cancer survivors have 28 29 87 reached a death rate similar to that of the general population. 30 31

88 http://bmjopen.bmj.com/ 32 Cancer survivors can experience physiological and/or psychosocial symptoms, due to side effects or long- 33 34 89 term effects of treatment [26] and are more likely to report fair or poor health overall in all age groups 35 36 90 [27]. For these survivors, work can represent a return to health or normality; a safeguard of their financial 37 38 91 security, self-esteem and social contacts [28-33]. 39

40 92 on September 29, 2021 by guest. Protected copyright. 41 42 93 Many studies have highlighted social inequalities in relation to return to work (RTW) among cancer 43 44 94 45 survivors [34] The well-established relationship between socioeconomic position (SEP) and long-term 46 47 95 sickness absence predicts that returning to work will be more difficult for cancer survivors in manual 48 49 96 occupations [35,36]. Previous research has shown that working conditions and psychosocial conditions in 50 51 97 manual occupations act as additional barriers [35,37-39]. Alongside the impact of working conditions, the 52 53 98 unequal use of cancer rehabilitation services [40] may also lead to social inequalities in terms of RTW. It 54 55 99 has also been shown that cancer survivors with a low SEP more commonly become unemployed [41] or 56 57 58 Version 07 July 2017 59 60 3

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 4 of 42

1 2 3 100 take early retirement, which can act as a substitute for sickness absence benefits or unemployment [41- 4 5 101 6 43]. 7 8 102 9 10 11 103 The Belgian context 12 13 104 In Belgium, cessation of work due to sickness must be reported to the employer immediately. The 14 15 105 employer pays theFor guaranteed peer salary for 14 workingreview days for blue-collar only workers (manual workers) and 16 17 18 106 28 working days for white-collar workers (intellectual workers). For self-employed or unemployed 19 20 107 individuals, the social security system (SSS) covers salary replacement after 28 working days. The 21 22 108 absence due to sickness must be confirmed by a general practitioner or a specialist doctor. 23 24 109 After the period of guaranteed income from the employer, the SSS takes over the provision of a 25 26 110 replacement income. The benefits for sickness-related absences vary between 40% and 65% of the 27 28 111 reference salary, depending on the family situation (Figure 1). 29 30 31 112 The SSS distinguishes between short-term and long-term work disability. Short-term work disability lasts 32 http://bmjopen.bmj.com/ 33 113 up to one year, while long-term work disability is for periods exceeding one year. The division reflects a 34 35 114 different evaluation method for assessing the worker’s eligibility for sickness absence benefits as well as 36 37 115 the calculation of the level of sickness absence benefits. 38 39 116 Entitlement to long-term sickness absence benefits begins as of the second year after stopping work (13th

40 on September 29, 2021 by guest. Protected copyright. 41 117 month) and continues until the age of retirement, with no limit of duration. This applies to employees, 42 43 44 118 self-employed and unemployed socially insured Belgian citizens. Civil servants (almost 20% of the 45 46 119 Belgian workforce) benefit from a specific social security scheme. In Belgium, more than 90% of citizens 47 48 120 are socially insured and covered by compulsory health insurance [44]. 49 50 121 51 52 122 There is an important knowledge gap in Belgium regarding a quantitative assessment of the impact of 53 54 123 55 cancer on work disability. The following aspects need to be better understood: how long the work 56 57 124 disability lasts, how the work disability ends, which workers are more at risk, etc. Our research helps to 58 Version 07 July 2017 59 60 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 5 of 42 BMJ Open

1 2 3 125 fill this gap. It is based on a recent model, developed in 2011 to study RTW after cancer [22],which 4 5 126 6 proposes a comprehensive list of influencing factors. Among these, we have been able to collect and 7 8 127 analyse data on the following: age, gender, occupational class, site of cancer and work-related outcomes 9 10 128 (ability to work, retirement, death and disability). 11 12 129 13 14 130 This study is part of the scientific approach initiated in 2012 at the request of the Federal Ministry of 15 For peer review only 16 131 Public Health and Social Security [45,46], to provide evidence and support for the decision-making 17 18 132 19 process in order to improve and facilitate the professional reintegration of cancer survivors. 20 21 133 Our work reflects research on work disability due to cancer. Work disability is defined or measured as a 22 23 134 legal status based on administrative definitions, i.e. eligibility for benefit. 24 25 135 This article describes and discusses the results of a population-based cohort study of people with long- 26 27 136 term cancer-related work disability, i.e. receiving sickness absence benefits for more than one year. We 28 29 137 will refer to this population below as ‘disabled workers’. They have been followed for three to seven 30 31

138 http://bmjopen.bmj.com/ 32 years in order to measure the outflow from work disability to either retirement, ability to work or death. 33 34 139 [Table 1 and Figure 1 can be displayed] 35 36 Group of diseases 2007 2008 2009 2010 2011 2012 2013 37 Mental heath 74,054(33%) 78,112(34%) 83,247(34%) 88,535(34%) 92,899(34%) 98,171(35%) 104,291(35%) 38 Muskuloskeletal and connective 58,032(26%) 60,595 (26%) 65,146(27%) 69,583(27%) 74,192(28%) 79,643(28%) 86,071(29%) 39 Circulatory diseases 19,372(9%) 19,216(8%) 19,427(8%) 19,571(8%) 19,549(7%) 19,772(7%) 19,963(7%) Traumatic injuries and poisoning 15,302(7%) 15,776(7%) 16,538(7%) 17,080(7%) 17,635(7%) 18,383(6%) 18,955(6%) 40 on September 29, 2021 by guest. Protected copyright. 41 Tumours* 13,592(6%) 14,266(6%) 15,103(6%) 16,083(6%) 16,742(6%) 17,591(6%) 18,462(6%) 42 Others (13 other conditions) 43,332(19%) 44,188(19%) 45,748(19%) 47,083(18%) 48,482(18%) 49,981(18%) 51,666(17%) 43 TOTAL 223,684 232,153 245,209 257,935 269,499 283,541 299,408 44 (100%) Table 1. Number of cause-specific disabled workers in Belgium. Top 5 Evolution 2007-2013. 45 Annual Report NIHDI, 2014. 46 *including cancers and benign tumours 47 140 48 49 50 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 5

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 6 of 42

1 2 3 4 141 Methods 5 6 142 Study population 7 8 9 143 We presented the list of data required, the objectives and the format in which we planned to publish the 10 11 144 results to the scientific board of the NIHDI. No ethical or privacy issues were identified by the Board, 12 13 145 which allowed the extraction of the required data and the transfer of the coded dataset to the Cancer 14 15 146 Centre of the ScientificFor Institute peer of Public Health review (IPH). All data areonly administrative data automatically 16 17 147 collected by the NIHDI. We therefore did not need informed consent from the workers. The coded data 18 19 20 148 were transferred to the IPH trough Outlook and are stored on thfe local server of the IPH that meets data 21 22 149 safety and protection standards. 23 24 150 We included all socially insured Belgian people who entered into long-term work disability due to cancer 25 26 151 between 1 January 2007 and 31 December 2011, excluding civil servants who are not included in the 27 28 152 NIHDI database. From the total of 21,701 individuals, 6,098 were excluded either due to their work 29 30 153 disability starting before 1 January 2007 (and non-equivalent follow-up time), or due to inconsistent 31 http://bmjopen.bmj.com/ 32 st 33 154 records (see Figure 2). The last update of the data was on 31 December 2013, resulting in a maximum 34 35 155 follow-up of seven years. 36 37 38 156 Design & statistical analysis 39

40 157 We conducted a register-based cohort study, using data from the disability register of the National on September 29, 2021 by guest. Protected copyright. 41 42 158 Institute for Health and Disability Insurance (NIHDI). Our research had three goals. Our first goal was to 43 44 45 159 measure the duration of work disability by the year of entry in the work disability system. To achieve this 46 47 160 first goal, we calculated the Kaplan-Meier estimate. 48 49 161 Secondly, following the taxonomy set out in theories of work disability [47], our study aimed to build a 50 51 162 prognostic model to estimate the sub-distribution hazards of each event (death, ability to work and 52 53 163 retirement) in the presence of competing risks using the Fine and Gray [48] model. For each event, the 54 55 56 57 58 Version 07 July 2017 59 60 6

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 7 of 42 BMJ Open

1 2 3 164 model was built separately for men and women, while adjusting for age, year of entry, cancer site and 4 5 165 6 occupational class. 7 8 166 A third objective was to investigate social inequalities for ability to work among cancer survivors, paying 9 10 167 attention to differences in age, gender and occupational class, and adjusting for year of entry For this, we 11 12 168 also used the Fine and Gray model, replacing the cancer sites with 13 14 169 four categories of cancer: those with low1, medium2 and high3 survival rates, according to the age- 15 For peer review only 16 170 standardised five-year relative survival (ASRS), calculated by the Belgian Cancer Registry [49]. The 17 18 171 4 19 missing category includes those individuals with a cancer site for which the ASRS was not available . 20 21 172 The two rationales behind this approach were as follows: firstly, it generates a parsimonious model (it 22 23 173 avoids the lack of convergence due to the large size of the data set). Secondly, this approach makes it 24 25 174 possible to account for the severity of the disease. 26 27 175 For the two first objectives, we used the “cmprsk” package of the statistical software R which allows sub 28 29 176 distribution analysis of competing risks. For the third objective, we used the Stata’s stcrreg package. 30 31

177 http://bmjopen.bmj.com/ 32 33 34 35 178 Independent prognostic variables 36 37 179 Sociodemographic characteristics included in our study were age at entry into the work disability 38 39 180 system, gender and occupational class. The age variable was based on the date of birth and was further

40 on September 29, 2021 by guest. Protected copyright. 41 181 categorised into four groups: 17-39; 40-49; 50-59; and 60+ years. Occupational classes were based on 42 43 44 182 four categories: blue-collar workers, white-collar workers, self-employed people and assisting spouses. 45 46 183 They were recoded into a 3-level variable: blue-collar workers, white-collar workers and self-employed 47 48 184 people. 49 50 51 52 53 54 55 56

57 58 Version 07 July 2017 59 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 8 of 42

1 2 3 185 In total, 39 cancer sites have been identified using the ‘pathology codes’ transmitted by the NIHDI and 4 5 186 6 registered by their ICD-9 codes (Table 2). For the sake of comparability, we translated these into ICD-10 7 8 187 codes and gathered them into 11 cancer sites (Table 2). 9 10 188 The year of entry in the work disability system was a continuous variable ranging from 2007 to 2011. We 11 12 189 decided to recode the year of entry into a two-level variable: 2007-2010 and 2011. This decision is based 13 14 190 on an exploratory analysis that showed significant difference in survival patterns between disability 15 For peer review only 16 191 acquired before or after 2011 (log-rank test=502, df=1, p-value < 0.001) (Figure 3). 17 18 19 20 192 Outcome variables: the three competing events 21 22 193 The outcome variable is the event that causes the end of work disability. We defined three mutually 23 24 194 exclusive events, i.e. competing risks: death, retirement and ability to work. 25 26 195 The status retirement indicates that the worker is definitively out of the labour market due to age and will 27 28 196 receive social benefits until death, while able to work indicates that the cancer-disabled worker was 29 30 31 197 recognised by a health insurer's doctor as able to work. In practice, this might lead to a RTW, to 32 http://bmjopen.bmj.com/ 33 198 unemployment or to a decision to be a stay-at-home spouse. 34 35 199 Those long-term disabled workers who had not experienced any event by the end of follow-up, on 31 36 37 200 December 2013, were administratively censored (31%, Table 3). 38 39 201 [Table2, Figures 2 and 3* could be displayed]

40 on September 29, 2021 by guest. Protected copyright. 41 202 *Remark for Fig 3: ending long-term work disability happens by death, retirement or ability to work, 42 43 44 203 whichever occurs first. 45 46 204 47 205 48 206 49 207 50 208 51 209 52 210 53 211 54 212 55 213 56 214 57 215 58 Version 07 July 2017 59 60 8

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 9 of 42 BMJ Open

1 2 3 216 4 217 5 218 6 GROUPS The 11 cancer groups by anatomical location 5-year 5-year Survival rate Frequency 7 relative relative category observed in 8 survival rate survival rate the data 9 men*(%) women*(%) 10 1 Other malignancies and undefined sites, CIS NA 1 247 11 Benign tumours 12 2 Head and neck: lip, oral cavity, nasal cavities, middle ear 50.0 57.0 medium 877 13 and accessory sinuses, pharynx, larynx 14 3 Digestive tract 22.8 22.7 low 2 400

40 on September 29, 2021 by guest. Protected copyright. Prostate 377 41 Testis 94 42 Others 16 43 9 Urinary tract 388 44 Kidney 71.0 0.7 high 147 45 Bladder 56.6 49.2 medium 178 46 Others 63 47 10 Central nervous system (CNS) 22.7 25.8 low 709 48 11 Bone and connective tissue (sarcoma’s) 61.9 59.7 medium 49 Melanoma of the skin 86.2 91.0 high Malignant neoplasms of skin other than melanoma NA 50 Thyroid and other endocrine glands 89.3 94.1 high 51 Other malignancies and undefined sites, invasive 51.5 39.1 medium/low 52 Tumours of uncertain and unspecified behavior NA 53 TOTAL 15 543 54 219 Table 2. The 11 cancer groups used for the analysis. 55 220 *Reference: Belgian Cancer Registry. Cancer Survival in Belgium, 2004-2008. 56 221 57 222 58 Version 07 July 2017 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 10 of 42

1 2 3 4 223 Results 5 6 7 224 Description of the study population 8 9 10 225 No observed workers were lost to follow-up. Table 3 describes the main characteristics of the work- 11 12 226 disabled cancer survivors included in the study. 13 14 227 The majority (77%) of the cancer-disabled workers were aged 40–59 years. 15 For peer review only 16 228 Women were over-represented (62%), younger at entry (median age of 48 vs. 53 years for men) and 17 18 229 mostly white-collar workers (46% vs. 21% for men) or blue-collar workers (43% vs. 60%). The outcome 19 20 230 for the majority of cancer-disabled women was ability to work (44.10%), while for most men the outcome 21 22 23 231 was death (50.86%). 24 25 232 The most frequent cancer site was breast, representing 35% (n= 5,949) of disabled workers, followed by 26 27 233 15% (n=2,400) of digestive tract cancers and 9% (n=1,417) of respiratory tract cancers. 28 29 234 Regarding OC, half of the disabled workers were blue-collar workers, the majority of whom (38.2% of 30 31 235 the total) died by the end of follow-up. White-collar workers (37%) had the shortest median time of work 32 http://bmjopen.bmj.com/ 33 236 disability (1.30 years vs. 1.79 years for the others), and the majority (47.7%) were able to work by the end 34 35 36 237 of follow-up. Self-employed disabled workers represented 13% of the cohort, and the majority (38.2%) 37 38 238 died by the end of follow-up. 39

40 239 By the end of follow-up, 69% of the cohort had experienced one of the three competing events (32.2% on September 29, 2021 by guest. Protected copyright. 41 42 240 died, 2.2% retired and 34.6% were able to work). The other 31% remained disabled, distributed as 43 44 241 follows: 35% of those who entered in 2009 and 2010, 31% of those aged 40-49 years, 27% of the women, 45 46 242 28% of the blue-collar workers and 44% of those with benign or in situ tumours (Table 3). 47 48 49 243 Figures 4-8 show the non-parametric cause-specific cumulative incidences of time to ability to work in 50 51 244 the presence of competing risks. For all prognostic variables, the curves show a steep increase in ability to 52 53 245 RTW within the first two years; later, the curves virtually level off. 54 55 56 57 58 Version 07 July 2017 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 11 of 42 BMJ Open

1 2 3 246 Figures 9-13 show the box plots of time to any event (death, ability to work or retirement) stratified by 4 5 247 6 each prognostic variable, respectively. 7 8 248 Younger workers (17-39 years) had the highest rates of ability to work at the end of follow-up (Figure 5) 9 10 249 and relatively short periods of work disability (Figure 10), mainly due to the ability to work. Older 11 12 250 workers presented the shortest work disability periods (Figure 10), mainly due to death or retirement 13 14 251 (77%, Table 3). 15 For peer review only 16 252 Women had higher rates of ability to work compared to men (Figure 6) but spent longer periods in work 17 18 253 19 disability (Figure 11). White-collar workers had higher rates of work disability and spent less time in it 20 21 254 (Figure 11). Regarding the cancer sites, workers with breast or haematological cancer had the highest 22 23 255 rates of ability to work by the end of follow-up (Figure 8), but the longest periods spent on work 24 25 256 disability (Figure 13). Those with respiratory tract, head and neck, digestive or central nervous system 26 27 257 (CNS) cancers had the lowest rates of ability to work (Figure 8) and shorter periods of work disability 28 29 258 (Figure 13), mainly due to death. 30 31

259 http://bmjopen.bmj.com/ 32 Those cancer survivors who entered into the work disability system in 2011 had higher cumulative 33 34 260 incidence of ability to work after three years in long-term disability (Figure 4) compared to those having 35 36 261 entered in earlier years (Figure 9). 37 38 262 [Table 3, Figures 4 and 5 could be displayed] 39

40 263 on September 29, 2021 by guest. Protected copyright. 41 42 264 43 44 265 45 46 47 266 48 49 267 50 51 268 52 53 54 55 56 57 58 Version 07 July 2017 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 12 of 42

12 12 ) ) Version 07 July 2017 2017 07 July Version (censored) (censored) Work disability disability Work 7.46 (17.76- 37.17) 37.17) (17.76- 7.46 27.75 (9.83- 45.68) (9.83- 45.68) 27.75 (7.11- 40.51) 23.81 (0.00- 48.43) 24.05 (0.00- 48.49) 21.33 41.29) (14.28- 27.78 (0.00- 41.44) 20.42 57.68) (31.57- 44.63 (0.00- 48.53) 19.96 (2.65- 32.13 17.39 11.36 (0.00- 40.58) (0.00- 40.58) 11.36 2 33.25) (23.11- 28.18 (0.00- 39.94) 19.49 (5.78- 36.86) 20.50 27.49) (10.29- 18.89 37.47) (16.68- 27.07 31.53) (12.07- 21.80 36.24) (25.95- 31.09 (8.60- 38.44) 23.51 (NA) 0.01 < 33.29) (16.47- 24.88 33.44) (21.40- 27.41 37.31) (26.13- 35.02 41.90) (28.15- 35.03 32.00) (10.23- 21.11 Ability to work work to Ability 41.09) (33.10- 37.09 24.54) (18.25- 21.40 28.55) (24.68- 26.61 (8.51- 11.76) 10.14 45.17) (40.18- 42.67 59.11) (49.24- 54.17 35.89) (29.19- 32.54 (35.76-45.28) 40.52 28.23) (19.41- 23.82 22.36) (16.18- 19.27 32.63) (25.40- 29.01 31.85) (29.62- 30.74 51.58) (43.80- 47.70 33.32) (28.96- 30.96 26.75) (24.39- 25.57 46.79) (41.41- 44.10 56.27) (51.82- 54.05 41.98) (36.06- 40.52 33.93) (27.55- 30.74 25.63) (20.60- 23.12 36.23) (32.02- 34.12 36.41) (33.20- 34.80 35.73) (32.41- 34.07 37.22) (32.54- 34.88 49.47) (44.04- 46.76

- - - % (IC95%) % (IC95%) Retirement Retirement (1.01 - 4.20) 2.61 (0.86 - 4.24) 2.55 (3.86 - 7.18) 5.52 (0.68 - 6.45) 3.57 (2.07 - 4.26) 3.16 (2.28 - 3.95) 3.12 (0.00 - 1.00) 0.47 (9.72 - 20.21) 14.96 (3.92 - 11.65) 7.78 (0.00 - 0.90) 0.42 (0.20 - 4.97) 2.58 (2.08 - 3.51) 2.80 (1.63 - 3.98) 2.80 (8.04 - 12.31) 10.25 (3.75 - 5.60) 4.68 (2.40 - 4.20) 3.30 (2.23 - 5.07) 3.65 33.63) (28.09- 30.86 (3.87 - 5.60) 4.73 (2.69 - 4.00) 3.35 (1.23 - 2.26) 1.75 (0.42 - 1.02) 0.72 (0.14 - 0.93) 0.54 Reason of ending work disability disability work ending of Reason http://bmjopen.bmj.com/

(38%) 5874 (62%) 9669 (15.6%) 2421 (32.5%) 5052 (44.7%) 6946 ( 7.21%) 1121 (22.2%) 3454 (24.2%) 3760 (23.4%) 3630 (21.8%) 3388 (8.4 %) 1311

Table 3. DescriptionTable 3. of the characteristics of the individuals included in the cohort, mediantheir spenttime the in disability, rate of those one experiencing of the three competing events and the of rate remainingthose work disabled at the end of follow-up.

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 14 of 42

14 14

- - - - - 95%CI 0.45,1.24 1.13,3.02 0.03,2.44 0.16,2.05 0.03,0.93 0.21,3.34 0.21,2.42 0.01,0.51 (0.15,3.49) - (0.02,2.01) (0.14,1.17) (54.3,204.8) Version 07 July 2017 2017 07 July Version Retirement

- - - - - SHR 1 0.75 1.84 1 0.25 0.57 0.18 0.84 0.72 0.05 0.73 0.73 0 0.20 1 0.41 1 105.5

- - SHR Ability to workAbility 1 1.43 1.07 1 0.77 0.69 0.27 1.11 1.42 0.81 0.74 0.41 0.75 1 0.81 0.64 0.49 1 2.03

- -

95%CI (0.96,1.18) (0.80,1.12) (2.66,6.77) (3.26,7.19) (6.71,14.92) (1.40,3.28) (0.85,1.83) (3.33,7.32) (2.20,6.38) (4.26,9.88) (3.50,8.07) (0.93,1.30) (1.22,1.67) (1.70,2.71) (0.96,1.36) Death

- - SHR 1 1.06 0.94 1 4.24 4.84 10.0 2.15 1.24 4.93 3.75 6.48 5.32 1 1.10 1.42 2.15 1 1.14

------

http://bmjopen.bmj.com/ 95%CI (0.82,2.60) (2.02,5.21) (0.12,1.17) (0.19,1.31) (0.05,0.59) (0.19,1.46) (0.46,3.09) (0.21,1.87) (0.03,0.74) (0.12,1.81) (39.73,284.06) 0.07,0.69 BMJ Open Retirement

------SHR 1 106.20 1 1.46 3.24 1 0.37 0.50 0.18 0.52 1.20 0.62 0.14 0.46 1 0.21

Ability to workAbility SHR 1 0.57 0.41 0.45 1 1.49 0.85 1 0.59 0.85 0.34 1.14 0.13 - 1.38 0.70 0.47 0.76 1 1.55

- - For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

95%CI (1.29,1.91) (1.60,2.31) (1.96,3.02) (0.92,1.14) (0.73,0.93) (1.97,3.94) (1.94,3.82) (4.14,8.17) (1.03,2.09) (0.45,3.67) (0.58,1.27) (1.78,3.85) (2.71,5.54) (2.01,4.27) (1.08,1.47) Death

For peer review only - - SHR 1 1.57 1.92 2.43 1 1.02 0.83 1 2.79 2.73 5.81 1.47 1.29 0.86 2.62 3.87 2.93 1 1.26

TableModel 4. sub-distribution1: ratio hazard the for theatstatus endof follow-up.

of entry 281 280 >=60 >=60 50-59 50-59 40-49 40-49 Age group Age 19-39 Self-Employed or assisting spouse spouse or assisting Self-Employed Blue Collar Blue Collar Occupational Class White Collar 11: Bone and connective 11:connective and Bone 10 Central Nervous System System 10 Nervous Central 9 Urinary Track 9 Track Urinary 8 Men Genital Organs Genital8 Organs Men 7 Women Genital Organs Women Organs 7 Genital 6 Breast 6 Breast 5 Haematological 5 Haematological 4 Respiratory Track 4 Track Respiratory 3 Digestive track 3 Digestive track 2 Head and Neck and2 Head Neck Cancer site Benign 1 and CIS 2011 2011 2007-2010 2007-2010

1 2 3 4 282 Social inequalities in the work disability of cancer survivors (model 2) 5 6 7 283 In the second model, we stratify by age and gender and allow interactions between both these variables 8 9 284 and OC and survival categories (Table 5). The absence of individuals in certain age categories entering 10 11 285 retirement (17-49 years, Table 2) leads to a convergence issue when modelling the cause-specific hazard 12 13 286 for this type of event and this is therefore not reported. 14 15 287 Table 5 shows that,For among men, peer blue-collar andreview self-employed workers only aged 50-59 years were less likely 16 17 18 288 to be able to work compared to white-collar workers. Similar results were found for blue-collar women 19 20 289 aged 17-39, 40-49 and 50-59. These results translate into larger social inequalities in the 50-59 age group 21 22 290 for both men and women. 23 24 291 Men self-employed were less likely to be able to work than white-collar workers when aged 17-39 or 50- 25 26 292 59 and similarly for women aged 50-59. 27 28 293 29 30 31 294 32 http://bmjopen.bmj.com/ 33 295 34 35 296 36 37 297 38 39 298

40 on September 29, 2021 by guest. Protected copyright. 41 299 42 43 44 300 45 46 301 47 48 302 49 50 303 51 52 304 53 54 305 55 56 57 306 58 Version 07 July 2017 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 16 of 42

1 2 3 Sub-distribution hazard ratio at the end of follow-up 4 5 MEN WOMEN 6 Death Ability to work Death Ability to work 7 17-39 8 Survival rate High 1 1 1 1 9 Medium 1.16 (0.48,2.82) 0.78 (0.47,1.29) 1.69 (1.08,2.65) 0.75 (0.57,0.99) 10 Low 2.63 (1.21,5.72) 0.33 (0.19,0.60) 2.91 (1.99,4.26) 0.38 (0.27,0.53) 11 Missing 2.17 (0.88,5.40) 0.50(0.25,0.98) 1.33 (0.77,2.29) 0.58 (0.42,0.81) 12 Occupational White Collar 1 1 1 1 13 class Blue Collar 0.48(0.20,1.15) 0.86(0.58,1.27) 1.41 (1.02,1.93) 0.58(0.48,0.69) Self-employed 1.43 (0.53,3.84) 0.48 (0.26,0.87) 1.13 (0.60,2.11) 1.05 (0.81,1.38) 14 40-49 15 For peer review only 16 Survival rate High 1 1 1 1 Medium 1.53 (0.74,3.15) 1.22 (0.66,2.25) 3.19 (2.39,4.27) 0.58(0.46,0.72) 17 Low 3.32 (1.67,6.61) 0.63 (0.33,1.20) 5.84 (4.32,7.90) 0.24(0.17,0.35) 18 Missing 0.89 (0.32,2.47) 1.69 (0.82,3.50) 2.96 (2.08,4.22) 0.60 (0.46,0.78) 19 Occupational White Collar 1 1 1 1 20 class Blue Collar 1.12(0.53,2.34) 0.79 (0.42,1.47) 1.42(1.13,1.80) 0.59 (0.53,0.67) 21 Self-employed 1.27(0.47,3.42) 1.74 (0.88,3.43) 0.96(0.62,1.49) 0.84 (0.70,1) 22 50-59 23 Survial rate High 1 1 1 1 24 Medium 1.39(0.93,2.10) 0.82 (0.53,1.24) 3.21 (2.53,4.07) 0.55 (0.44,0.70) 25 Low 3.21(2.15,4.77) 0.38(0.24,0.61) 6.01 (4.73,7.64) 0.21 (0.14,0.31) 26 Missing 1.47(0.86,2.52) 0.76 (0.42,1.36) 2.22 (1.62,3.04) 0.75 (0.57,0.98) 27 Occupational White Collar 1 1 1 1 28 class Blue Collar 0.88 (0.57,1.34) 0.60 (0.39,0.92) 1.07 (0.86,1.34) 0.75 (0.65,0.86) Self-employed 0.75 (0.45,1.24) 0.57 (0.33,0.96) 1.34 (0.98,1.83) 0.69 (0.56,0.86) 29 30 ≥60 31 Survival rate High 1 1 1 1 Medium 1.13 (0.51,2.52) 2.26 (0.90,5.68) 2.85 (1.52,5.38) 0.91 (0.47,1.76) http://bmjopen.bmj.com/ 32 Low 3.16 (1.52,6.55) 0.72 (0.25,2.06) 3.92 (1.91,8.03) 0.29 (0.10,0.86) 33 Missing 2.14 (0.69,6.66 1.21 (0.29,4.99) 2.68(0.95,7.54) 0.41 (0.09,1.89) 34 White Collar 1 1 1 1 35 Occupational class Blue Collar 1.39 (0.61,3.13) 1.57 (0.60,4.08) 1.06 (0.54,2.09) 0.80 (0.46,1.39) 36 Self-employed 0.55 (0.23,1.34) 1.19 (0.46,3.10) 1.09 (0.55,2.18) 0.60 (0.32,1.12) 37 307 Table 5. Sub-distribution hazard ratio based on the Fine and Gray model stratified by age and gender, and 38 308 adjusted for year of entry. 39 309 1Low survival rates: "Esophagus", "Stomach", "Colon & rectum", "Pancreas", "Oth. Mal. Neop. of digestive organs and

40 310 peritoneum", "Mesothelioma", "Trachea and lung", "Myeloid and others", "CNS", "Oth. malignancies and undefined sites, on September 29, 2021 by guest. Protected copyright. 41 311 invasive" 42 312 2Medium survival rates: "Lip and oral cavity, nasal cavities, middle ear and accessory sinuses, pharynx, larynx","Non 43 313 Hodgkin's Disease", "Uterus", "Cervix","Ovary","Oth. mal. neop. Women genitals", "Cervix", "Ovary", "Oth. mal. neop. of 44 314 women genitals", "Bladder", "Urinary system other than bladder", "Bone and connective tissue" 315 3High survival rates: "Hodgkin Disease","Acute lymphoid leukemia and lymphoid leukemia, other", "Breast 45 316 women","Uterus","Kidney","Melanoma of the skin", "Thyroid and other endocrine glands" 46 317 4Missing survival rates:"Benign tumour", "Mal. neop. of skin other than melanoma", "Oth. malignancies and undefined sites, 47 318 CIS", "Tumours of uncertain and unspecified behavior" 48 319 49 50 320 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 17 of 42 BMJ Open

1 2 3 4 321 DISCUSSION 5 6 322 In this study, we aimed to identify the factors that influence the reason for exiting the long-term work 7 8 323 disability system and the length of work disability among cancer survivors. 9 10 324 To achieve this, we first measured the association between the duration of work disability and age, 11 12 13 325 gender, occupational class, the year of entry into the work disability system and 11 cancer sites. Secondly, 14 15 326 we estimated theFor distribution peer of three competing review reasons for exiting only the work disability system; and 16 17 327 thirdly, we investigated social inequalities in work disability among cancer survivors. 18 19 328 20 21 329 As not many of the population-based studies in this field include several cancer sites or use competing 22 23 330 risk analysis, making comparisons is not easy. Moreover, our follow-up starts one year after the first day 24 25 26 331 of sickness absence, i.e. that we only include long-term disabled workers. However, the impact of these 27 28 332 determining factors on labour market participation has been tested in previous studies [50]. Results 29 30 333 indicate that, overall, older age at entry into the work disability system and man gender are both factors 31 32 334 that decrease the chance of being economically active. Our results show that an older age (>60 years) http://bmjopen.bmj.com/ 33 34 335 increase the risk of dying or retiring, and that workers aged 40-49 were the most likely to remain disabled 35 36 336 for a long period (Table 3). Man gender did indeed reduce the likelihood of being able to work but 37 38 39 337 women experienced longer periods within the work disability system overall.

40 on September 29, 2021 by guest. Protected copyright. 41 338 42 43 339 Regarding the cancer sites, we found a strong association between respiratory tract, head and neck and 44 45 340 digestive tract cancers and death. The first two include smoking-related cancer sites [51], which represent 46 47 341 major sources of work disability and death in the working age population. 48 49 342 Other studies have compared different cancer sites to assess their association with employment status 50 51 52 343 after cancer diagnosis. In line with our results, workers with respiratory and women genital cancers 53 54 344 present smaller proportions of employment than workers with breast or haematological cancers, mainly 55 56 345 due to poor self-reported health status [26,27,52]. 57 58 Version 07 July 2017 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 18 of 42

1 2 3 346 4 5 347 6 In line with previous research, blue-collar and self-employed workers are less likely to be able to work 7 8 348 after cancer compared to white-collar workers, especially those aged 50-59 years. [26]. According to 9 10 349 other research, these social inequalities could be explained by more demanding working conditions [53], 11 12 350 later stage of cancer at diagnosis, differences in treatment [54] and lower participation in rehabilitation 13 14 351 services [55] among blue-collar workers. OC is also strongly associated with the income level, which may 15 For peer review only 16 352 represent an incentive to RTW when it is significantly different (higher) than sickness absence benefits 17 18 353 19 [36]. 20 21 354 A different impact of the OC on the risk of work disability according to age and gender has been shown in 22 23 355 a Norwegian county, where young workers with blue-collar jobs are more at risk than older men [56]. The 24 25 356 association between age and RTW has been reported with contradictory results in the literature, but the 26 27 357 majority found higher age to be associated with later RTW or reduced chance of employment [34]. Our 28 29 358 results show that, for Belgian cancer survivors, the opposite is found, with a larger impact of OC from the 30 31

359 http://bmjopen.bmj.com/ 32 age of 40 years onwards, compared to younger counterparts. 33 34 360 35 36 361 Demographic changes and the rising retirement age will increase the number of disabled workers and the 37 38 362 length of work disability; combined with the effects of the economic crisis (i.e. greater competition and 39

40 363 emphasis on maximum performance) this will worsen the situation if we do not implement measures, on September 29, 2021 by guest. Protected copyright. 41 42 364 interventions and rehabilitation programmes to better (re)integrate disabled workers in the labour market 43 44 365 45 [5]. 46 47 366 The measure introduced by the Belgian government by the end of 2010 seems to have had an impact 48 49 367 already, as the workers who entered the work disability system in 2011 showed better outcomes than the 50 51 368 others. In 2011, a new measure was implemented, allowing disabled workers to resume work without 52 53 369 prior agreement of the health insurer's medical advisor. 54 55 56 57 58 Version 07 July 2017 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 19 of 42 BMJ Open

1 2 3 370 Further studies need to be carried out in future to confirm this trend. However, at the end of follow-up, 4 5 371 6 only 34.6% of the cancer survivors were able to work and 31% were administratively censored, remaining 7 8 372 disabled. 9 10 373 Based on our results, key features of (work) rehabilitation programmes can be drawn. The non-parametric 11 12 374 cause-specific cumulative incidence of time to ability to work (Figures 4) suggests that interventions 13 14 375 should be planned and implemented within the 2 years after the cancer diagnosis. Differences in age and 15 For peer review only 16 376 gender imply tailoring of and specific attention to the needs of young workers and women. The 17 18 377 19 association of the cancer site with the length of disability suggests that the awareness of oncologists who 20 21 378 treat breast cancer, digestive track cancers and head and neck cancers, should be raised on the RTW and 22 23 379 that they need to be involved in the assessment and management of symptoms. 24 25 380 The negative association of being blue-collar or self-employed calls for the revision of employment 26 27 381 policies for these high-risk groups, with e.g. the creation of incentives for employers to adjust the working 28 29 382 conditions of their sick-listed blue-collar workers. 30 31

383 http://bmjopen.bmj.com/ 32 33 34 384 Strengths, limitations and needs for further research 35 36 385 The main strength of this study is the representativeness of the data and the generalizability of our results. 37 38 386 We included in the analysis all Belgian workers disabled due to cancer between 2007- 2011, excluding 39 40 387 civil servants and individuals for whom we detected coding errors. on September 29, 2021 by guest. Protected copyright. 41 42 43 388 44 45 389 In most work disability studies, survival analyses are used to estimate the time to an event of interest. The 46 47 390 end of work disability is, however, more complex than this, and may be caused by multiple factors. 48 49 391 Therefore, the use of competing risks analysis becomes appropriate to avoid over- or under-estimating the 50 51 392 probability of experiencing each event [57]. This model is still rarely used in work disability studies and 52 53 393 its use should be encouraged. 54 55 56 394 57 58 Version 07 July 2017 59 60 19

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 20 of 42

1 2 3 395 Regarding the objective of predicting disability, the two models showed their capacity to and 4 5 396 6 effectiveness in predicting the length and the reasons for ending work disability among Belgian cancer 7 8 397 survivors. Our second model presents original findings, using the survival rates to identify social 9 10 398 inequalities. 11 12 399 Nevertheless, regarding the objective of providing insights on the content of work rehabilitation 13 14 400 programmes, crucial information is lacking: the stage at diagnosis, the type of treatment received and the 15 For peer review only 16 401 side and long-term effects of the treatment [58]; the specific occupation and the working environment 17 18 402 19 [59]. The inclusion of these variables in our models would allow a more complex but efficient model, 20 21 403 explaining the remaining differences that still exist among workers of the same age, gender, OC and 22 23 404 cancer site. This is feasible in the future, e.g. by linking data from Cancer Registries to data on 24 25 405 employment and socioeconomic status. Results could be used to develop rehabilitation programmes for 26 27 406 cancer survivors similar to those that already exist in other countries [60-63]. 28 29 407 30 31

408 http://bmjopen.bmj.com/ 32 While our paper focuses on work disability among cancer survivors in Belgium, it is important to realise 33 34 409 that the methods and principles used are generic, and applicable for addressing work disability as a whole. 35 36 410 Therefore, this report is also relevant for other conditions and SSS. This paper contributes towards closing 37 38 411 the knowledge gap on the transition among cancer survivors from long-term work disability to ability to 39

40 412 RTW. Linking these important results to predictions of cancer incidence should make it possible to plan on September 29, 2021 by guest. Protected copyright. 41 42 413 cancer rehabilitation needs and related sickness absence benefits. 43 44 45 414 46 47 415 48 49 416 50 51 417 52 53 418 54 55 56 57 58 Version 07 July 2017 59 60 20

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 21 of 42 BMJ Open

1 2 3 4 419 Contributorship statement 5 420 6 7 421 Régine Kiasuwa Mbengi, MPH, PhD Student 8 422 Designed the study 9 10 423 Requested/collected the data 11 424 Performed the analysis 12 425 Wrote the paper 13 426 14 427 Mioara Alina Nicolae, MSc,PhD 15 428 For peer review only 16 Provided statistical support to build the models 17 429 Provided advice on the writing of the sections on methods and results 18 430 19 431 Prof. Els Goetghebeur, MSc,PhD 20 432 Provided statistical support to build the models 21 433 Provided advice on the writing of the sections on methods and results 22 23 434 24 435 Renee Otter, MD, PhD 25 436 Provided clinical expertise support to build cancer categories 26 437 Substantially contributed to the writing 27 438 28 439 29 Katrien Mortelmans, MD, PhD 30 440 Substantially contributed to the writing 31 441 32 442 Sarah Missine, MSc, PhD http://bmjopen.bmj.com/ 33 443 Provided advice and support to integrate the inequalities perspective in the paper and in the 34 444 statistical model 35 36 445 Supported the writing of the paper 37 446 38 447 Marc Arbyn, MD 39 448 Substantially contributed to the preparation of the tables and figures and the abstract section 40 449 on September 29, 2021 by guest. Protected copyright. 41 450 42 Prof. Catherine Bouland, MsC, PhD 43 451 As PhD co-supervisor for Miss Kiasuwa, Prof. C. Bouland substantially contributed to the 44 452 preparation of the study, the design and the revision of the paper 45 453 46 454 Prof. Christophe de Brouwer, MD, PhD 47 455 As the PhD supervisor for Miss Kiasuwa, Prof. C. de Brouwer substantially contributed to the 48 49 456 preparation of the study, the design and the revision of the paper 50 457 51 458 52 53 54 55 56 57 58 Version 07 July 2017 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 22 of 42

1 2 3 4 459 Funding and competing interests 5 6 460 This study has been funded by the National Institute for Health and Disability Insurance. We 7 8 9 461 have read and understood the BMJ policy on declaration of interests and declare that we have no 10 11 462 competing interests. 12 13 463 This article does not contain any studies with human participants or animals performed by any of 14 15 For peer review only 16 464 the authors. 17 18 465 19 20 21 466 Data sharing statement 22 23 467 The complete and anonymised dataset used for this study can be made available by the 24 25 26 468 corresponding author for researchers interested in comparative studies. This request would be 27 28 469 subject of approval from the Belgian National Institute for Health and Disability Insurance, the 29 30 470 Scientific Institute of Public Health and the Université Libre de Bruxelles. 31 32 http://bmjopen.bmj.com/ 33 34 35 471 Figures legends 36 472 Figure 1. The Belgian social security scheme related to work disability 37 38 473 39 474 Figure 2. Flowchart of the number of workers disabled because of cancer between 2007 and 2011 in 40 on September 29, 2021 by guest. Protected copyright. 41 475 Belgium. 42 43 476 44 477 Figure 3. Kaplan-Meier estimator for the time in work disability, stratified by the year of entrance 45 46 478 into work disability. 47 48 479 49 480 Figures 4 -8. Non-parametric cumulative incidence of ability to work by the end of follow-up, 50 51 481 stratified per a) year of entry, b) age at entry, c) gender, d) occupational class and e) cancer sites. 52 53 482 54 483 Figures 9-13. Box plots of time to any of the three events (death, retirement, ability to work), by 55 56 484 gender, age group, year of entrance and occupational class. 57 58 Version 07 July 2017 59 60 22

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 23 of 42 BMJ Open

1 2 3 485 4 5 486 6 7 487 Reference List 8 9 488 10 11 489 1. Amir Z, Wilson K, Hennings J, Young A: The meaning of cancer: implications for family finances 12 490 and consequent impact on lifestyle, activities, roles and relationships. Psychooncology 2012, 21: 13 491 1167-1174. 14 15 492 2. Daniel C.Lustig:For Causal peerRelationships Betweenreview Poverty and Disability.only Rehabilitation Counseling 16 17 493 Bulletin 2007, 50: 194-202. 18 19 494 3. Schuring M, Robroek SJ, Otten FW, Arts CH, Burdorf A: The effect of ill health and socioeconomic 20 495 status on labor force exit and re-employment: a prospective study with ten years follow-up in 21 496 the Netherlands. Scand J Work Environ Health 2013, 39: 134-143. 22 23 497 4. Reichard A., Stransky M., Phillips K., Drum C., Mc Clain M.: Does Type of Disability Matter to 24 498 Public Health Policy Practice? Californian Journal of Health Promotion 2015, 13: 25-36. 25 26 27 499 5. OECD. Sickness, Disability and Work. Breaking the Barriers. 2010. 28 500 Ref Type: Report 29 30 501 6. Luengo-Fernandez R, Leal J, Gray A, Sullivan R: Economic burden of cancer across the European 31 502 Union: a population-based cost analysis. Lancet Oncol 2013, 14: 1165-1174. 32 http://bmjopen.bmj.com/ 33 503 7. Betsch N: Kankerpatiënten en werkhervatting: rechten en plichten. Instituut Voor Arbeidsrecht, KU 34 504 Leuven; 2013. 35 36 37 505 8. Desiron HA, Donceel P, de RA, Van HE: A conceptual-practice model for occupational therapy to 38 506 facilitate return to work in breast cancer patients. J Occup Rehabil 2013, 23: 516-526. 39

40 507 9. Desiron HA, Donceel P, Godderis L, Van HE, de RA: What is the value of occupational therapy in on September 29, 2021 by guest. Protected copyright. 41 508 return to work for breast cancer patients? A qualitative inquiry among experts. Eur J Cancer Care 42 509 (Engl ) 2015, 24: 267-280. 43 44 510 10. Desiron HA, Crutzen R, Godderis L, Van HE, de RA: Bridging Health Care and the Workplace: 45 46 511 Formulation of a Return-to-Work Intervention for Breast Cancer Patients Using an Intervention 47 512 Mapping Approach. J Occup Rehabil 2016. 48 49 513 11. Rommel W. Werken na kanker:welke problemen ervaren (ex-)patiënten die het werk hervatten? 50 514 2012. Vlaams Liga Tegen Kanker. 51 515 Ref Type: Report 52 53 516 12. Tiedtke C, Donceel P, Knops L, Desiron H, Dierckx de CB, de RA: Supporting return-to-work in the 54 517 face of legislation: stakeholders' experiences with return-to-work after breast cancer in Belgium. 55 56 518 J Occup Rehabil 2012, 22: 241-251. 57 58 Version 07 July 2017 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 24 of 42

1 2 3 519 13. Tiedtke C, Dierckx de CB, Donceel P, de RA: Workplace support after breast cancer treatment: 4 520 recognition of vulnerability. Disabil Rehabil 2014, 1-7. 5 6 7 521 14. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 8 522 Chapter 12. Future need for research. Scand J Public Health Suppl 2004, 63: 256-258. 9 10 523 15. Alexanderson K, Hensing G: More and better research needed on sickness absence. Scand J Public 11 524 Health 2004, 32: 321-323. 12 13 525 16. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 14 526 Chapter 1. Aim, background, key concepts, regulations, and current statistics. Scand J Public 15 For peer review only 16 527 Health Suppl 2004, 63: 12-30. 17 18 528 17. INAMI. Rapport annuel. Satistiques des indemnités. 2014. 19 529 Ref Type: Report 20 21 530 18. Coleman MP, Gatta G, Verdecchia A, Esteve J, Sant M, Storm H et al.: EUROCARE-3 summary: 22 531 cancer survival in Europe at the end of the 20th century. Ann Oncol 2003, 14 Suppl 5: v128-v149. 23 24 532 19. Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, Rosso S, Coebergh JW, Comber H et al.: Cancer 25 26 533 incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 27 534 2013, 49: 1374-1403. 28 29 535 20. Pryce J, Munir F, Haslam C: Cancer survivorship and work: symptoms, supervisor response, co- 30 536 worker disclosure and work adjustment. J Occup Rehabil 2007, 17: 83-92. 31 32 537 21. Kennedy F HCMFPJ: Returning to work following cancer: a qualitative exploratory study into the http://bmjopen.bmj.com/ 33 538 experience of returning to work following cancer. Eur J Cancer Care 2007, 16: 17-25. 34 35 36 539 22. Mehnert A: Employment and work-related issues in cancer survivors. Crit Rev Oncol Hematol 37 540 2011, 77: 109-130. 38 39 541 23. Mehnert A, de BA, Feuerstein M: Employment challenges for cancer survivors. Cancer 2013, 119

40 542 Suppl 11: 2151-2159. on September 29, 2021 by guest. Protected copyright. 41 42 543 24. Verbeek J, Spelten E, Kammeijer M, Sprangers M: Return to work of cancer survivors: a 43 544 prospective cohort study into the quality of rehabilitation by occupational physicians. Occup 44 45 545 Environ Med 2003, 60: 352-357. 46 47 546 25. Dal ML, Guzzinati S, Buzzoni C, Capocaccia R, Serraino D, Caldarella A et al.: Long-term survival, 48 547 prevalence, and cure of cancer: a population-based estimation for 818 902 Italian patients and 49 548 26 cancer types. Ann Oncol 2014, 25: 2251-2260. 50 51 549 26. Kiasuwa MR, Otter R, Mortelmans K, Arbyn M, Van OH, Bouland C et al.: Barriers and 52 550 opportunities for return-to-work of cancer survivors: time for action-rapid review and expert 53 54 551 consultation. Syst Rev 2016, 5: 35. 55 56 552 27. Hewitt M, Rowland JH, Yancik R: Cancer survivors in the United States: age, health, and disability. 57 553 J Gerontol A Biol Sci Med Sci 2003, 58: 82-91. 58 Version 07 July 2017 59 60 24

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 25 of 42 BMJ Open

1 2 3 554 28. Amir Z, Brocky J: Cancer survivorship and employment: epidemiology. Occup Med (Lond) 2009, 4 555 59: 373-377. 5 6 7 556 29. Collins C.G, Ottati A., Feuerstein M: Cancer Survivorship. In Handbook of Work Disability. 8 557 Prevention and Management. Springer edition. Edited by Loisel P, Anema J.R. New York: 2013:289- 9 558 302. 10 11 559 30. Tiedtke C, de RA, Dierckx de CB, Christiaens MR, Donceel P: Experiences and concerns about 12 560 'returning to work' for women breast cancer survivors: a literature review. Psychooncology 2010, 13 561 19: 677-683. 14 15 For peer review only 16 562 31. Tiedtke C, Dierckx de CB, de RA, Christiaens MR, Donceel P: Breast cancer treatment and work 17 563 disability: patient perspectives. Breast 2011, 20: 534-538. 18 19 564 32. van Muijen P., Duijts SF, van der Beek AJ, Anema JR: Prognostic factors of work disability in sick- 20 565 listed cancer survivors. J Cancer Surviv 2013, 7: 582-591. 21 22 566 33. Grunfeld EA, Cooper AF: A longitudinal qualitative study of the experience of working following 23 567 treatment for gynaecological cancer. Psychooncology 2012, 21: 82-89. 24 25 26 568 34. van Muijen P., Weevers NL, Snels IA, Duijts SF, Bruinvels DJ, Schellart AJ et al.: Predictors of return 27 569 to work and employment in cancer survivors: a systematic review. Eur J Cancer Care (Engl ) 2013, 28 570 22: 144-160. 29 30 571 35. Sterud T, Johannessen HA: Do work-related mechanical and psychosocial factors contribute to 31 572 the social gradient in long-term sick leave: a prospective study of the general working http://bmjopen.bmj.com/ 32 573 population in Norway. Scand J Public Health 2014, 42: 329-334. 33 34 35 574 36. Piha K, Laaksonen M, Martikainen P, Rahkonen O, Lahelma E: Interrelationships between 36 575 education, occupational class, income and sickness absence. Eur J Public Health 2010, 20: 276- 37 576 280. 38 39 577 37. Melchior M., Krieger N, Kawachi I., et al.: Work Factors and Occupational Class Disaprities in

40 578 Sickness Absence: Findings From the GAZEL Cohort Study. American Journal of Public Health on September 29, 2021 by guest. Protected copyright. 41 579 2005, 95: 1206-1212. 42 43 44 580 38. Melchior M, Goldberg M, Krieger N, Kawachi I, Menvielle G, Zins M et al.: Occupational class, 45 581 occupational mobility and cancer incidence among middle-aged men and women: a prospective 46 582 study of the French GAZEL cohort*. Cancer Causes Control 2005, 16: 515-524. 47 48 583 39. Melchior M, Krieger N, Kawachi I, Berkman LF, Niedhammer I, Goldberg M: Work factors and 49 584 occupational class disparities in sickness absence: findings from the GAZEL cohort study. Am J 50 585 Public Health 2005, 95: 1206-1212. 51 52 53 586 40. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 54 587 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 55 56 588 41. Lindbohm ML, Kuosma E, Taskila T, Hietanen P, Carlsen K, Gudbergsson S et al.: Early retirement 57 589 and non-employment after breast cancer. Psychooncology 2014, 23: 634-641. 58 Version 07 July 2017 59 60 25

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 26 of 42

1 2 3 590 42. Carlsen K, Hoybye MT, Dalton SO, Tjonneland A: Social inequality and incidence of and survival 4 591 from breast cancer in a population-based study in Denmark, 1994-2003. Eur J Cancer 2008, 44: 5 6 592 1996-2002. 7 8 593 43. Carlsen K, Dalton SO, Diderichsen F, Johansen C: Risk for unemployment of cancer survivors: A 9 594 Danish cohort study. Eur J Cancer 2008, 44: 1866-1874. 10 11 595 44. Gerkens S, Merkur S: Belgium: Health system review. Health Syst Transit 2010, 12: 1-266, xxv. 12 13 596 45. Kiasuwa Mbengi R. La réinsertion socioprofessionnelle des patients atteints de cancer. Exposé des 14 597 positions. 2014. 15 For peer review only 16 598 Ref Type: Online Source 17 18 599 46. Onkelinx L. Toespraak van de Minister van Volksgezondheid en Sociale Zaken op het Symposium 19 600 Evaluatie van het Kankerplan. 2012. 20 601 Ref Type: Report 21 22 602 47. de Rijk A.: Work Disability Theories: A Taxonomy for Researchers. In Handbook of Work 23 603 Disability. Prevention and Managment. Edited by Loisel P., Anema J.R. New York: Springer; 24 25 604 2013:475-499. 26 27 605 48. Jason P.Fine and Robert J.Gray: A Proportional Hazards Model for the Subdistribution of a 28 606 Competing Risk. Journal of the American Statistical Association 1999, 94: 496-509. 29 30 607 49. Belgian Cancer Registry. Cancer Survival in Belgium. 2004-2008. 1-109. 2012. 31 608 Ref Type: Report 32 http://bmjopen.bmj.com/ 33 609 50. Rijken M, Spreeuwenberg P, Schippers J, Groenewegen PP: The importance of illness duration, 34 35 610 age at diagnosis and the year of diagnosis for labour participation chances of people with 36 611 chronic illness: results of a nationwide panel-study in The Netherlands. BMC Public Health 2013, 37 612 13: 803. 38 39 613 51. Regional variation in incidence for smoking and alcohol related cancers in Belgium. Edited by Kris

40 614 Henau, Elizabeth Van Eycken, Geert Silverman, Eero Pukkala. Cancer Epidemiology 39[1], 55-65. on September 29, 2021 by guest. Protected copyright. 41 615 2016. 42 43 616 Ref Type: Generic 44 45 617 52. Mehnert A, Koch U: Predictors of employment among cancer survivors after medical 46 618 rehabilitation--a prospective study. Scand J Work Environ Health 2013, 39: 76-87. 47 48 619 53. Ervasti J, Kivimaki M, Dray-Spira R, Head J, Goldberg M, Pentti J et al.: Socioeconomic gradient in 49 620 work disability in diabetes: evidence from three occupational cohorts. J Epidemiol Community 50 621 Health 2016, 70: 125-131. 51 52 53 622 54. Woods LM, Rachet B, Coleman MP: Origins of socio-economic inequalities in cancer survival: a 54 623 review. Ann Oncol 2006, 17: 5-19. 55 56 624 55. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 57 625 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 58 Version 07 July 2017 59 60 26

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 27 of 42 BMJ Open

1 2 3 626 56. Krokstad S, Johnsen R, Westin S: Social determinants of disability pension: a 10-year follow-up of 4 627 62 000 people in a Norwegian county population. Int J Epidemiol 2002, 31: 1183-1191. 5 6 7 628 57. Noordzij M, Leffondre K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ: When do we need 8 629 competing risks methods for survival analysis in nephrology? Nephrol Dial Transplant 2013, 28: 9 630 2670-2677. 10 11 631 58. Glimelius I, Ekberg S, Linderoth J, Jerkeman M, Chang ET, Neovius M et al.: Sick leave and 12 632 disability pension in Hodgkin lymphoma survivors by stage, treatment, and follow-up time-a 13 633 population-based comparative study. J Cancer Surviv 2015. 14 15 For peer review only 16 634 59. Bradley CJ, Bednarek HL: Employment patterns of long-term cancer survivors. Psychooncology 17 635 2002, 11: 188-198. 18 19 636 60. Scott DA Mills M, Black A, Cantwell M, Campbell A, Cardwell CR, Porter S et al.. Multidimensional 20 637 rehabilitation programmes for adult cancer survivors (Review). Cochrane.Database.Syst.Rev. [3]. 21 638 2016. 22 639 Ref Type: Generic 23 24 25 640 61. Mewes JC, Steuten LM, Groeneveld IF, de Boer AG, Frings-Dresen MH, IJzerman MJ et al.: Return- 26 641 to-work intervention for cancer survivors: budget impact and allocation of costs and returns in 27 642 the Netherlands and six major EU-countries. BMC Cancer 2015, 15: 899. 28 29 643 62. de Boer AG, Taskila TK, Tamminga SJ, Feuerstein M, Frings-Dresen MH, Verbeek JH: Interventions 30 644 to enhance return-to-work for cancer patients. Cochrane Database Syst Rev 2015, 9: CD007569. 31 http://bmjopen.bmj.com/ 32 645 63. Zaman AC, Bruinvels DJ, de Boer AG, Frings-Dresen MH: Supporting cancer patients with work- 33 34 646 related problems through an oncological occupational physician: a feasibility study. Eur J Cancer 35 647 Care (Engl ) 2015. 36 648 37 649 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 27

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 28 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 3. Kaplan-Meier estimator for remaining in long-term disability stratified by the year of entrance on September 29, 2021 by guest. Protected copyright. 41 42 152x152mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 31 of 42 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 4. Cumulative incidence of ability to work stratified by the year of entrance into long-term disability on September 29, 2021 by guest. Protected copyright. 41 42 152x152mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 32 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39 40 on September 29, 2021 by guest. Protected copyright. 41 42 Figure 5. Cumulative incidence of ability to work stratified by the age at entry into long-term disability 43 44 381x407mm (120 x 120 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 33 of 42 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 Figure 6. Cumulative incidence of ability to work stratified by gender 44 45 381x417mm (120 x 120 DPI) 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 34 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 Figure 7. Cumulative incidence of ability to work stratified by occupational class 44 381x411mm (120 x 120 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 35 of 42 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 Figure 8. Cumulative incidence of ability to work stratified by cancer site 43 44 381x404mm (120 x 120 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 36 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 9. Box plot of time to any event (death, ability to work, retirement) by the year of entry into long- on September 29, 2021 by guest. Protected copyright. 41 term disability 42 43 381x381mm (120 x 120 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 37 of 42 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 10. Box plot of time to any event (death, ability to work, retirement) by age at entry into long-term on September 29, 2021 by guest. Protected copyright. 41 disability 42 43 381x381mm (120 x 120 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 38 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 11. Box plot of time to any event (death, ability to work, retirement) by gender on September 29, 2021 by guest. Protected copyright. 41 42 381x381mm (120 x 120 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 39 of 42 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 12. Box plot of time to any event (death, ability to work, retirement) by occupational class on September 29, 2021 by guest. Protected copyright. 41 42 381x381mm (120 x 120 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 40 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 Figure 13. Box plot of time to any event (death, ability to work, retirement) by cancer site 42 370x375mm (96 x 96 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 41 of 42 BMJ Open

Assessing Factors Associated with Long-Term Work Disability After Cancer in Belgium. A Population-Based Cohort Study Using Competing Risks Analysis with a 7-year Follow-Up

For peer review only

Journal: BMJ Open

Manuscript ID bmjopen-2016-014094.R4

Article Type: Research

Date Submitted by the Author: 17-Oct-2017

Primary Subject Epidemiology Heading:

Secondary Subject Heading: Public health on September 29, 2021 by guest. Protected copyright.

Epidemiology < ONCOLOGY, ONCOLOGY, Organisation of health services < Keywords: HEALTH SERVICES ADMINISTRATION & MANAGEMENT

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 1 of 41 BMJ Open

1 2 3 4 1 Assessing Factors Associated with Long-Term Work Disability After 5 2 Cancer in Belgium. A Population-Based Cohort Study Using 6 7 3 Competing Risks Analysis with a 7-year Follow-Up 8 9 4 R. Kiasuwa Mbengia,c, A.M Nicolaieb , Els Goetghebeurb, R. Ottera, K. Mortlemansd, S. Missinnea, 10 e c c 11 5 M. Arbyn , C. Bouland and C. de Brouwer 12 6 a Belgian Cancer Centre, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 13 7 b Stat-Gent CRESCENDO, University of Ghent 14 8 c Research Centre for Environmental and Occupational Health, Brussels School of Public Health, Université Libre de Bruxelles (ESP-ULB) 15 9 d Dr. Katrien Mortelmans,For PhD Consulting peer (KaMoCo) review only 16 10 e Unit Cancer Epidemiology, Scientific Institute of Public Health, Brussels (WIV-ISP), Belgium 17 11 18 12 Corresponding author: Régine Kiasuwa Mbengi 19 13 [email protected] 14 20 14 rue Juliette Wytsman, 1050 Brusss - Belgium 15 Office: +32 2 642 57 65 21 16 Mobile: +32 479 3926 58 22 23 24 25 17 ABSTRACT 26 18 Objectives 27 19 The number of workers with cancer is increasing dramatically worldwide. One of the main priorities is to 28 20 preserve their quality of life and the sustainability of social security systems. We have carried out this 29 21 study in order to assess factors associated with the ability to work after cancer. Such insight should help 30 31 22 with the planning of rehabilitation needs and tailored programmes. 32 23 Participants http://bmjopen.bmj.com/ 33 24 We conducted this register-based cohort study using individual data from the Belgian Disability 34 25 Insurance. Data on 15,543 socially insured Belgian people who entered into the long-term work disability 35 26 between 2007 and 2011 due to cancer were used. 36 27 Primary and secondary outcome measures 37 28 We estimated the duration of work disability using Kaplan Meier and the cause-specific cumulative 38 29 incidence of ability to work stratified by age, gender, occupational class and year of entering the work 39 30 disability system for 11 cancer sites using the Fine and Gray model allowing for competing risks. 40 31 Results on September 29, 2021 by guest. Protected copyright. 41 32 The overall median time of work disability was 1.59 years (95%CI [1.52-1.66]), ranging from 0.75 to 42 33 4.98 years. By the end of follow-up, more than one-third of the disabled cancer survivors were able to 43 44 34 work (35%). While a large proportion of the women were able to work at the end of follow-up, the men 45 35 who were able to work could do so sooner. Women, white-collar, younger and having haematological, 46 36 men genital or breast cancers were the most likely to be able to RTW. 47 37 Conclusion 48 38 Good prognostic factors for the ability to work were youth, woman, white-collar, and having breast, men 49 39 genital or haematological cancers. 50 40 Reviewing our results together with the cancer incidence predictions up to 2025 offers a high value for 51 41 social security and rehabilitation planning, and for ascertaining patients’ perspectives. 52 42 53 43 Key words: sickness absence, cancer survivors, competing risks, predictive model, social inequalities 54 44 55 45 Strengths and limitations of the study methods: 56 57 58 Version 07 July 2017 59 60 1

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 2 of 41

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 3 of 41 BMJ Open

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 4 of 41

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 5 of 41 BMJ Open

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 6 of 41

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 7 of 41 BMJ Open

57 58 Version 07 July 2017 59 60 7

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 8 of 41

1 2 3 185 In total, 39 cancer sites have been identified using the ‘pathology codes’ transmitted by the NIHDI and 4 5 186 6 registered by their ICD-9 codes (Table 2). For the sake of comparability, we translated these into ICD-10 7 8 187 codes and gathered them into 11 cancer sites (Table 2). 9 10 188 The year of entry in the work disability system was a continuous variable ranging from 2007 to 2011. We 11 12 189 decided to recode the year of entry into a two-level variable: 2007-2010 and 2011. This decision is based 13 14 190 on an exploratory analysis that showed significant difference in survival patterns between disability 15 For peer review only 16 191 acquired before or after 2011 (log-rank test=502, df=1, p-value < 0.001) (Figure 3). 17 18 19 20 192 Outcome variables: the three competing events 21 22 193 The outcome variable is the event that causes the end of work disability. We defined three mutually 23 24 194 exclusive events, i.e. competing risks: death, retirement and ability to work. 25 26 195 The status retirement indicates that the worker is definitively out of the labour market due to age and will 27 28 196 receive social benefits until death, while able to work indicates that the cancer-disabled worker was 29 30 31 197 recognised by a health insurer's doctor as able to work. In practice, this might lead to a RTW, to 32 http://bmjopen.bmj.com/ 33 198 unemployment or to a decision to be a stay-at-home spouse. 34 35 199 Those long-term disabled workers who had not experienced any event by the end of follow-up, on 31 36 37 200 December 2013, were administratively censored (38%, Table 3). 38 39 201 [Table2, Figures 2 and 3* could be displayed]

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 9 of 41 BMJ Open

40 on September 29, 2021 by guest. Protected copyright. Prostate 377 41 Testis 94 42 Others 16 43 9 Urinary tract 388 44 Kidney 71.0 0.7 high 147 45 Bladder 56.6 49.2 medium 178 46 Others 63 47 10 Central nervous system (CNS) 22.7 25.8 low 709 48 11 Bone and connective tissue (sarcoma’s) 61.9 59.7 medium 49 Melanoma of the skin 86.2 91.0 high Malignant neoplasms of skin other than melanoma NA 50 Thyroid and other endocrine glands 89.3 94.1 high 51 Other malignancies and undefined sites, invasive 51.5 39.1 medium/low 52 Tumours of uncertain and unspecified behavior NA 53 TOTAL 15 543 54 219 Table 2. The 11 cancer groups used for the analysis. 55 220 *Reference: Belgian Cancer Registry. Cancer Survival in Belgium, 2004-2008. 56 221 57 222 58 Version 07 July 2017 59 60 9

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 10 of 41

1 2 3 4 223 Results 5 6 7 224 Description of the study population 8 9 10 225 No observed workers were lost to follow-up. Table 3 describes the main characteristics of the work- 11 12 226 disabled cancer survivors included in the study. 13 14 227 The majority (77%) of the cancer-disabled workers were aged 40–59 years. 15 For peer review only 16 228 Women were over-represented (62%), younger at entry (median age of 48 vs. 53 years for men) and 17 18 229 mostly white-collar workers (46% vs. 21% for men) or blue-collar workers (43% vs. 60%). After 3 years 19 20 230 of follow-up, the outcome for the majority of cancer-disabled women (irrespectively from their year of 21 22 23 231 entry) was disability (42.35%), while for most men the outcome was death (43.52%). 24 25 232 The most frequent cancer site was breast, representing 35% (n= 5,949) of disabled workers, followed by 26 27 233 15% (n=2,400) of digestive tract cancers and 9% (n=1,417) of respiratory tract cancers. 28 29 234 Regarding OC, half of the disabled workers were blue-collar workers, the majority of whom (41.34% of 30 31 235 the total) was still disabled after 3 years of follow-up. White-collar workers (37%) had the shortest 32 http://bmjopen.bmj.com/ 33 236 median time of work disability (1.30 years vs. 1.79 years for the others), and the majority (40.74%) were 34 35 36 237 able to work after 3 years of follow-up. Self-employed disabled workers represented 13% of the cohort, 37 38 238 and the majority (38.82%) was still in disability after 3 years of follow-up. 39

40 239 After 3 years of follow-up, 62% of the cohort had experienced one of the three competing events (29% on September 29, 2021 by guest. Protected copyright. 41 42 240 died, 1% retired and 32% were able to work). The other 38% remained disabled, (Table 3). 43 44 241 Figures 4-8 show the non-parametric cause-specific cumulative incidences of time to ability to work in 45 46 242 the presence of competing risks. For all prognostic variables, the curves show a steep increase in ability to 47 48 49 243 RTW within the first two years; later, the curves virtually level off. 50 51 244 Figures 9-12 show the box plots of time to any event (death, ability to work or retirement) stratified by 52 53 245 each prognostic variable, respectively. 54 55 56 57 58 Version 07 July 2017 59 60 10

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 11 of 41 BMJ Open

1 2 3 246 Younger workers (17-39 years) had the highest rates of ability to work at the end of follow-up (Figure 5) 4 5 247 6 and relatively short periods of work disability (Figure 9), mainly due to the ability to work. Older workers 7 8 248 presented the shortest work disability periods (Figure 9), mainly due to death or retirement (59.14%, 9 10 249 Table 3). 11 12 250 Women had higher rates of ability to work compared to men (Figure 6) but spent longer periods in work 13 14 251 disability (Figure 10). White-collar workers had higher rates of work disability and spent less time in it 15 For peer review only 16 252 (Figure 11). Regarding the cancer sites, workers with breast or haematological cancer had the highest 17 18 253 19 rates of ability to work by the end of follow-up (Figure 8), but the longest periods spent on work 20 21 254 disability (Figure 12). Those with respiratory tract, head and neck, digestive or central nervous system 22 23 255 (CNS) cancers had the lowest rates of ability to work (Figure 8) and shorter periods of work disability 24 25 256 (Figure 12), mainly due to death. 26 27 257 [Table 3, Figures 4-12 could be displayed] 28 29 258 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 11

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 12 of 41

12 12 Version 07 July 2017 2017 07 July Version

(censored) (censored) Work disability disability Work (47.87, 62.18) 55.03 (28.97, 46.08) 37.53 (26.29, 37.94) 32.11 (9.26, 30.28) 19.77 (35.38,46.90) 41.14 (41.15, 47.10) 44.13 (27.60,45.62) 36.61 (28.60, 50.61) 39.60 (26.37,51.36) 38.86 (27.88, 46.96) 37.42 (22.81, 44.31) 33.56 (38.68,44.00) 41.34 (30.56,37.78) 34.17 (33.47, 44.18) 38.82 (28.07,35.57) 31.82 (40.02,44.67) 42.35 (30.69, 41.33) 36.01 (37.78, 44.39) 41.09 (37.41, 43.15) 40.28 (7.36,31.33) 19.34 (36.74,44.78) 40.76 (35.62,43.52) 39.57 (35.90,43.88) 39.89 (35.45,43.77) 39.61 (10.23,32.00) 21.12 Ability to work work to Ability (29.35, 36.79) 33.07 (15.17, 20.22) 17.69 (22.27,25.69) 23.98 (7.94,9.47 10.99) (35.06,39.72) 37.39 (43.49,46.12) 44.80 (25.67,31.86) 28.76 (31.97, 40.53) 36.25 (17.31, 25.47) 21.39 (13.52, 18.95) 16.23 (23.18, 29.92) 26.55 (25.66,27.63) 26.65 (39.47, 42.02) 40.74 (24.66, 28.41) 26.53 (22.08,24.23) 23.15 (36.09,38.02) 37.06 (43.54, 47.52) 45.53 (34.84, 37.49) 36.17 (24.48,26.53) 25.51 (19.10, 23.92) 21.51 (27.18,30.20) 28.69 (28.67,31.60) 30.13 (29.11,32.11) 30.61 (30.72,33.86) 32.29 (44.04,49.47) 46.76

Retirement Retirement (0.30,1.98) 1.14 (0.00,0.74) 0.35 (1.23,2.28) 1.76 (0.37,0.86 1.34) (0.45,1.36) 0.91 (0.55,1.02) 0.78 (0.00,0.36) 0.12 (4.06,8.36) 6.21 (0.32,1.55 2.78) (0.00,0.90) 0.42 (0.10,0.76 1.42) (0.42,0.76) 0.59 (0.43,0.84) 0.63 (3.24,4.92) 4.08 (1.19,1.82) 1.5 (0.64,1.00) 0.82 ------(12.86, 17.04) 14.95 (1.10,1.91) 1.51 (1.24,2.06) 1.65 (0.46,1.02) 0.74 (0.31,0.81) 0.56 (0.14,0.93) 0.54 http://bmjopen.bmj.com/ 3 year cumulative probability of ending work disability disability work ending of probability year3 cumulative BMJ Open (8.31, 13.21) 10.76 (41.14,47.73) 44.43 (40.18,44.13) 42.15 (67.52, 72.30) 69.91 (18.62,22.51) 20.56 (9.49, 11.09) 10.29 (31.25, 37.76) 34.50 (14.52, 21.35) 17.93 (33.35, 43.03) 38.19 (42.25, 49.60) 45.92 (35.40, 42.86) 39.13 (30.39, 32.46) 31.43 (23.34,25.57) 24.46 (28.60, 32.53) 30.56 (42.26,44.79) 43.52 (18.98, 20.57) 19.78 (16.91, 20.00) 18.46 (21.59, 23.90) 22.74 (33.10,35.33) 34.21 (41.28,47.10) 44.19 (27.52,30.55) 29.04 (27.20,30.09) 28.64 (27.29,30.23) 28.76 (26.03,29.04) 27.54 (29.01,34.17) 31.59 Death Death 4468 (29%) (29%) 4468 (1%) 167 (32%) 4943 (38%) 5964

on September 29, 2021 by guest. Protected copyright. [3.24-6.72] 4.98 [1.41-1.93] 1.69 [1.18-1.41] 1.31 [0.69-0.83] 0.75 [1.66-2.08] 1.83 [1.95-2.30] 2.10 [1.37-1.82] 1.56 [1.52-2.24] 1.73 [1.32-2.16] 1.67 [1.18-1.89] 1.46 [1.03-1.52] 1.24 [1.72-1.89] 1.79 [1.23-1.37] 1.30 [1.63-1.97] 1.79 [1.12-1.26] 1.18 [1.84-2.05] 1.94 [1.415-1.69] 1.51 [1.637-1.88] 1.76 [1.580-1.79] 1.70 [0.797-1.06] 0.91 [1.780-2.048] 1.89 [1.717-1.960] 1.83 [1.621-1.889] 1.76 [1.410-1.670] 1.54 [0.413-0.487] 0.45 (years) (IC 95%) 95%) (years) (IC in work disability disability work in Median time spent spent time Median

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml (4%) 614 (5.6%) 877 (15.4%) 2400 (9.1%) 1417 (10.7%) 1660 (35.%) 5511 (5.3%) 821 (3.1%) 486 (2.5%) 388 (4.6%) 709 (4.2%) 660 (50%) 7715 (37%) 5703 (13%) 2125 (38%) 5874 (62%) 9669 2421(15.6%) (32.5%) 5052 (44.7%) 6946 (7.2%) 1121 (22.2%) 3454 (24.2%) 3760 (23.4%) 3630 (21.8%) 3388 (8.4 %) 1311 15543 15543 [1.52-1.66] 1.59 n (%)n individuals individuals Total Total For peer review only

DescriptionTable 3. of the characteristics of the individuals included in the cohort, mediantheir spenttime the in disability, rate of those one experiencing of the three competing events and the of rate remainingthose work disabled after three years of follow-up. TOTAL n (%) n TOTAL Cancer site site Cancer CIS/Ben Neck & Head tract Digestive tract Respiratory Haematological Breast organs genital Women organs genital Men tract Urinary (CNS) system nervous Central (sarc.)/Skin/Thyroid Bone&Con Occupational Class Occupational Collar Blue Collar White &Self-employed Spouse Assisting Gender Gender Men Women Age at entry entry Age at 17-39 40-49 50-59 >=60 Year of entry Year entry of 2007 2008 2009 2010 2011

n=15,543 259 260 261 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 13 of 41 BMJ Open

1 2 3 4 262 Prediction patterns of the end of work disability (model 1) 5 6 7 263 Results in Table 4 suggest that good prognostic factors for the ability to work for both men and women 8 9 264 are disability experienced after 2011 and white-collar OC. Regarding the 11 cancer sites, men with 10 11 265 haematological or genital organ cancers are the most likely to be able to work. Among women, the cancer 12 13 266 sites with the best chance for ability to work are haematological and breast. 14 15 267 Concerning deathsFor among men, peer disabled workers review with respiratory tract,only CNS, bone & connective tissue 16 17 18 268 cancers are most at risk. Among women, those with respiratory tract, women genital organs, digestive 19 20 269 tract and head and neck cancers are most at risk. 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 13

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 14 of 41

14 14

- - - - - 95%CI 0.45,1.24 1.13,3.02 0.03,2.44 0.16,2.05 0.03,0.93 0.21,3.34 0.21,2.42 0.01,0.51 (0.15,3.49) - (0.02,2.01) (0.14,1.17) (54.3,204.8) Version 07 July 2017 2017 07 July Version Retirement

- - - - - SHR 1 0.75 1.84 1 0.25 0.57 0.18 0.84 0.72 0.05 0.73 0.73 0 0.20 1 0.41 1 105.5

- - SHR Ability to workAbility 1 1.43 1.07 1 0.77 0.69 0.27 1.11 1.42 0.81 0.74 0.41 0.75 1 0.81 0.64 0.49 1 2.03

- -

95%CI (0.96,1.18) (0.80,1.12) (2.66,6.77) (3.26,7.19) (6.71,14.92) (1.40,3.28) (0.85,1.83) (3.33,7.32) (2.20,6.38) (4.26,9.88) (3.50,8.07) (0.93,1.30) (1.22,1.67) (1.70,2.71) (0.96,1.36) Death

- - SHR 1 1.06 0.94 1 4.24 4.84 10.0 2.15 1.24 4.93 3.75 6.48 5.32 1 1.10 1.42 2.15 1 1.14

------

http://bmjopen.bmj.com/ 95%CI (0.82,2.60) (2.02,5.21) (0.12,1.17) (0.19,1.31) (0.05,0.59) (0.19,1.46) (0.46,3.09) (0.21,1.87) (0.03,0.74) (0.12,1.81) (39.73,284.06) 0.07,0.69 BMJ Open Retirement

------SHR 1 106.20 1 1.46 3.24 1 0.37 0.50 0.18 0.52 1.20 0.62 0.14 0.46 1 0.21

Ability to workAbility SHR 1 0.57 0.41 0.45 1 1.49 0.85 1 0.59 0.85 0.34 1.14 0.13 - 1.38 0.70 0.47 0.76 1 1.55

- - For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

95%CI (1.29,1.91) (1.60,2.31) (1.96,3.02) (0.92,1.14) (0.73,0.93) (1.97,3.94) (1.94,3.82) (4.14,8.17) (1.03,2.09) (0.45,3.67) (0.58,1.27) (1.78,3.85) (2.71,5.54) (2.01,4.27) (1.08,1.47) Death

For peer review only - - SHR 1 1.57 1.92 2.43 1 1.02 0.83 1 2.79 2.73 5.81 1.47 1.29 0.86 2.62 3.87 2.93 1 1.26

TableModel 4. sub-distribution1: ratio hazard the for theatstatus endof follow-up.

of entry 271 270 >=60 >=60 50-59 50-59 40-49 40-49 Age group Age 19-39 Self-Employed or assisting spouse spouse or assisting Self-Employed Blue Collar Blue Collar Occupational Class White Collar 11: Bone and connective 11:connective and Bone 10 Central Nervous System System 10 Nervous Central 9 Urinary Track 9 Track Urinary 8 Men Genital Organs Genital8 Organs Men 7 Women Genital Organs Women Organs 7 Genital 6 Breast 6 Breast 5 Haematological 5 Haematological 4 Respiratory Track 4 Track Respiratory 3 Digestive track 3 Digestive track 2 Head and Neck and2 Head Neck Cancer site Benign 1 and CIS 2011 2011 2007-2010 2007-2010

1 2 3 4 272 Social inequalities in the work disability of cancer survivors (model 2) 5 6 7 273 In the second model, we stratify by age and gender and allow interactions between both these variables 8 9 274 and OC and survival categories (Table 5). The absence of individuals in certain age categories entering 10 11 275 retirement (17-49 years, Table 2) leads to a convergence issue when modelling the cause-specific hazard 12 13 276 for this type of event and this is therefore not reported. 14 15 277 Table 5 shows that,For among men, peer blue-collar andreview self-employed workers only aged 50-59 years were less likely 16 17 18 278 to be able to work compared to white-collar workers. Similar results were found for blue-collar women 19 20 279 aged 17-39, 40-49 and 50-59. These results translate into larger social inequalities in the 50-59 age group 21 22 280 for both men and women. 23 24 281 Men self-employed were less likely to be able to work than white-collar workers when aged 17-39 or 50- 25 26 282 59 and similarly for women aged 50-59. 27 28 283 29 30 31 284 32 http://bmjopen.bmj.com/ 33 285 34 35 286 36 37 287 38 39 288

40 on September 29, 2021 by guest. Protected copyright. 41 289 42 43 44 290 45 46 291 47 48 292 49 50 293 51 52 294 53 54 295 55 56 57 296 58 Version 07 July 2017 59 60 15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 16 of 41

1 2 3 Sub-distribution hazard ratio at the end of follow-up 4 5 MEN WOMEN 6 Death Ability to work Death Ability to work 7 17-39 8 Survival rate High 1 1 1 1 9 Medium 1.16 (0.48,2.82) 0.78 (0.47,1.29) 1.69 (1.08,2.65) 0.75 (0.57,0.99) 10 Low 2.63 (1.21,5.72) 0.33 (0.19,0.60) 2.91 (1.99,4.26) 0.38 (0.27,0.53) 11 Missing 2.17 (0.88,5.40) 0.50(0.25,0.98) 1.33 (0.77,2.29) 0.58 (0.42,0.81) 12 Occupational White Collar 1 1 1 1 13 class Blue Collar 0.48(0.20,1.15) 0.86(0.58,1.27) 1.41 (1.02,1.93) 0.58(0.48,0.69) Self-employed 1.43 (0.53,3.84) 0.48 (0.26,0.87) 1.13 (0.60,2.11) 1.05 (0.81,1.38) 14 40-49 15 For peer review only 16 Survival rate High 1 1 1 1 Medium 1.53 (0.74,3.15) 1.22 (0.66,2.25) 3.19 (2.39,4.27) 0.58(0.46,0.72) 17 Low 3.32 (1.67,6.61) 0.63 (0.33,1.20) 5.84 (4.32,7.90) 0.24(0.17,0.35) 18 Missing 0.89 (0.32,2.47) 1.69 (0.82,3.50) 2.96 (2.08,4.22) 0.60 (0.46,0.78) 19 Occupational White Collar 1 1 1 1 20 class Blue Collar 1.12(0.53,2.34) 0.79 (0.42,1.47) 1.42(1.13,1.80) 0.59 (0.53,0.67) 21 Self-employed 1.27(0.47,3.42) 1.74 (0.88,3.43) 0.96(0.62,1.49) 0.84 (0.70,1) 22 50-59 23 Survial rate High 1 1 1 1 24 Medium 1.39(0.93,2.10) 0.82 (0.53,1.24) 3.21 (2.53,4.07) 0.55 (0.44,0.70) 25 Low 3.21(2.15,4.77) 0.38(0.24,0.61) 6.01 (4.73,7.64) 0.21 (0.14,0.31) 26 Missing 1.47(0.86,2.52) 0.76 (0.42,1.36) 2.22 (1.62,3.04) 0.75 (0.57,0.98) 27 Occupational White Collar 1 1 1 1 28 class Blue Collar 0.88 (0.57,1.34) 0.60 (0.39,0.92) 1.07 (0.86,1.34) 0.75 (0.65,0.86) Self-employed 0.75 (0.45,1.24) 0.57 (0.33,0.96) 1.34 (0.98,1.83) 0.69 (0.56,0.86) 29 30 ≥60 31 Survival rate High 1 1 1 1 Medium 1.13 (0.51,2.52) 2.26 (0.90,5.68) 2.85 (1.52,5.38) 0.91 (0.47,1.76) http://bmjopen.bmj.com/ 32 Low 3.16 (1.52,6.55) 0.72 (0.25,2.06) 3.92 (1.91,8.03) 0.29 (0.10,0.86) 33 Missing 2.14 (0.69,6.66 1.21 (0.29,4.99) 2.68(0.95,7.54) 0.41 (0.09,1.89) 34 White Collar 1 1 1 1 35 Occupational class Blue Collar 1.39 (0.61,3.13) 1.57 (0.60,4.08) 1.06 (0.54,2.09) 0.80 (0.46,1.39) 36 Self-employed 0.55 (0.23,1.34) 1.19 (0.46,3.10) 1.09 (0.55,2.18) 0.60 (0.32,1.12) 37 297 Table 5. Sub-distribution hazard ratio based on the Fine and Gray model stratified by age and gender, and 38 298 adjusted for year of entry. 39 299 1Low survival rates: "Esophagus", "Stomach", "Colon & rectum", "Pancreas", "Oth. Mal. Neop. of digestive organs and

40 300 peritoneum", "Mesothelioma", "Trachea and lung", "Myeloid and others", "CNS", "Oth. malignancies and undefined sites, on September 29, 2021 by guest. Protected copyright. 41 301 invasive" 42 302 2Medium survival rates: "Lip and oral cavity, nasal cavities, middle ear and accessory sinuses, pharynx, larynx","Non 43 303 Hodgkin's Disease", "Uterus", "Cervix","Ovary","Oth. mal. neop. Women genitals", "Cervix", "Ovary", "Oth. mal. neop. of 44 304 women genitals", "Bladder", "Urinary system other than bladder", "Bone and connective tissue" 305 3High survival rates: "Hodgkin Disease","Acute lymphoid leukemia and lymphoid leukemia, other", "Breast 45 306 women","Uterus","Kidney","Melanoma of the skin", "Thyroid and other endocrine glands" 46 307 4Missing survival rates:"Benign tumour", "Mal. neop. of skin other than melanoma", "Oth. malignancies and undefined sites, 47 308 CIS", "Tumours of uncertain and unspecified behavior" 48 309 49 50 310 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 16

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 17 of 41 BMJ Open

1 2 3 4 311 DISCUSSION 5 6 312 In this study, we aimed to identify the factors that influence the reason for exiting the long-term work 7 8 313 disability system and the length of work disability among cancer survivors. 9 10 314 To achieve this, we first measured the association between the duration of work disability and age, 11 12 13 315 gender, occupational class, the year of entry into the work disability system and 11 cancer sites. Secondly, 14 15 316 we estimated theFor distribution peer of three competing review reasons for exiting only the work disability system; and 16 17 317 thirdly, we investigated social inequalities in work disability among cancer survivors. 18 19 318 20 21 319 As not many of the population-based studies in this field include several cancer sites or use competing 22 23 320 risk analysis, making comparisons is not easy. Moreover, our follow-up starts one year after the first day 24 25 26 321 of sickness absence, i.e. that we only include long-term disabled workers. However, the impact of these 27 28 322 determining factors on labour market participation has been tested in previous studies [50]. Results 29 30 323 indicate that, overall, older age at entry into the work disability system and man gender are both factors 31 32 324 that decrease the chance of being economically active. Our results show that an older age (>60 years) http://bmjopen.bmj.com/ 33 34 325 increase the risk of dying or retiring, and that workers aged 40-49 were the most likely to remain disabled 35 36 326 for a long period (Table 3). Man gender did indeed reduce the likelihood of being able to work but 37 38 39 327 women experienced longer periods within the work disability system overall.

40 on September 29, 2021 by guest. Protected copyright. 41 328 42 43 329 Regarding the cancer sites, we found a strong association between respiratory tract, head and neck and 44 45 330 digestive tract cancers and death. The first two include smoking-related cancer sites [51], which represent 46 47 331 major sources of work disability and death in the working age population. 48 49 332 Other studies have compared different cancer sites to assess their association with employment status 50 51 52 333 after cancer diagnosis. In line with our results, workers with respiratory and women genital cancers 53 54 334 present smaller proportions of employment than workers with breast or haematological cancers, mainly 55 56 335 due to poor self-reported health status [26,27,52]. 57 58 Version 07 July 2017 59 60 17

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 18 of 41

1 2 3 336 4 5 337 6 In line with previous research, blue-collar and self-employed workers are less likely to be able to work 7 8 338 after cancer compared to white-collar workers, especially those aged 50-59 years. [26]. According to 9 10 339 other research, these social inequalities could be explained by more demanding working conditions [53], 11 12 340 later stage of cancer at diagnosis, differences in treatment [54] and lower participation in rehabilitation 13 14 341 services [55] among blue-collar workers. OC is also strongly associated with the income level, which may 15 For peer review only 16 342 represent an incentive to RTW when it is significantly different (higher) than sickness absence benefits 17 18 343 19 [36]. 20 21 344 A different impact of the OC on the risk of work disability according to age and gender has been shown in 22 23 345 a Norwegian county, where young workers with blue-collar jobs are more at risk than older men [56]. The 24 25 346 association between age and RTW has been reported with contradictory results in the literature, but the 26 27 347 majority found higher age to be associated with later RTW or reduced chance of employment [34]. Our 28 29 348 results show that, for Belgian cancer survivors, the opposite is found, with a larger impact of OC from the 30 31

349 http://bmjopen.bmj.com/ 32 age of 40 years onwards, compared to younger counterparts. 33 34 350 35 36 351 Demographic changes and the rising retirement age will increase the number of disabled workers and the 37 38 352 length of work disability; combined with the effects of the economic crisis (i.e. greater competition and 39

40 353 emphasis on maximum performance) this will worsen the situation if we do not implement measures, on September 29, 2021 by guest. Protected copyright. 41 42 354 interventions and rehabilitation programmes to better (re)integrate disabled workers in the labour market 43 44 355 45 [5]. 46 47 356 The measure introduced by the Belgian government by the end of 2010 seems to have had an impact 48 49 357 already, as the workers who entered the work disability system in 2011 showed better outcomes than the 50 51 358 others. In 2011, a new measure was implemented, allowing disabled workers to resume work without 52 53 359 prior agreement of the health insurer's medical advisor. 54 55 56 57 58 Version 07 July 2017 59 60 18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 19 of 41 BMJ Open

1 2 3 360 Further studies need to be carried out in future to confirm this trend. However, at the end of follow-up, 4 5 361 6 only 34.6% of the cancer survivors were able to work and 31% were administratively censored, remaining 7 8 362 disabled. 9 10 363 Based on our results, key features of (work) rehabilitation programmes can be drawn. The non-parametric 11 12 364 cause-specific cumulative incidence of time to ability to work (Figures 4) suggests that interventions 13 14 365 should be planned and implemented within the 2 years after the cancer diagnosis. Differences in age and 15 For peer review only 16 366 gender imply tailoring of and specific attention to the needs of young workers and women. The 17 18 367 19 association of the cancer site with the length of disability suggests that the awareness of oncologists who 20 21 368 treat breast cancer, digestive track cancers and head and neck cancers, should be raised on the RTW and 22 23 369 that they need to be involved in the assessment and management of symptoms. 24 25 370 The negative association of being blue-collar or self-employed calls for the revision of employment 26 27 371 policies for these high-risk groups, with e.g. the creation of incentives for employers to adjust the working 28 29 372 conditions of their sick-listed blue-collar workers. 30 31

373 http://bmjopen.bmj.com/ 32 33 34 374 Strengths, limitations and needs for further research 35 36 375 The main strength of this study is the representativeness of the data and the generalizability of our results. 37 38 376 We included in the analysis all Belgian workers disabled due to cancer between 2007- 2011, excluding 39 40 377 civil servants and individuals for whom we detected coding errors. on September 29, 2021 by guest. Protected copyright. 41 42 43 378 44 45 379 In most work disability studies, survival analyses are used to estimate the time to an event of interest. The 46 47 380 end of work disability is, however, more complex than this, and may be caused by multiple factors. 48 49 381 Therefore, the use of competing risks analysis becomes appropriate to avoid over- or under-estimating the 50 51 382 probability of experiencing each event [57]. This model is still rarely used in work disability studies and 52 53 383 its use should be encouraged. 54 55 56 384 57 58 Version 07 July 2017 59 60 19

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 20 of 41

1 2 3 385 Regarding the objective of predicting disability, the two models showed their capacity to and 4 5 386 6 effectiveness in predicting the length and the reasons for ending work disability among Belgian cancer 7 8 387 survivors. Our second model presents original findings, using the survival rates to identify social 9 10 388 inequalities. 11 12 389 Nevertheless, regarding the objective of providing insights on the content of work rehabilitation 13 14 390 programmes, crucial information is lacking: the stage at diagnosis, the type of treatment received and the 15 For peer review only 16 391 side and long-term effects of the treatment [58]; the specific occupation and the working environment 17 18 392 19 [59]. The inclusion of these variables in our models would allow a more complex but efficient model, 20 21 393 explaining the remaining differences that still exist among workers of the same age, gender, OC and 22 23 394 cancer site. This is feasible in the future, e.g. by linking data from Cancer Registries to data on 24 25 395 employment and socioeconomic status. Results could be used to develop rehabilitation programmes for 26 27 396 cancer survivors similar to those that already exist in other countries [60-63]. 28 29 397 30 31

398 http://bmjopen.bmj.com/ 32 While our paper focuses on work disability among cancer survivors in Belgium, it is important to realise 33 34 399 that the methods and principles used are generic, and applicable for addressing work disability as a whole. 35 36 400 Therefore, this report is also relevant for other conditions and SSS. This paper contributes towards closing 37 38 401 the knowledge gap on the transition among cancer survivors from long-term work disability to ability to 39

40 402 RTW. Linking these important results to predictions of cancer incidence should make it possible to plan on September 29, 2021 by guest. Protected copyright. 41 42 403 cancer rehabilitation needs and related sickness absence benefits. 43 44 45 404 46 47 405 48 49 406 50 51 407 52 53 408 54 55 56 57 58 Version 07 July 2017 59 60 20

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 21 of 41 BMJ Open

1 2 3 4 409 Contributorship statement 5 410 6 7 411 Régine Kiasuwa Mbengi, MPH, PhD Student 8 412 Designed the study 9 10 413 Requested/collected the data 11 414 Performed the analysis 12 415 Wrote the paper 13 416 14 417 Mioara Alina Nicolae, MSc,PhD 15 418 For peer review only 16 Provided statistical support to build the models 17 419 Provided advice on the writing of the sections on methods and results 18 420 19 421 Prof. Els Goetghebeur, MSc,PhD 20 422 Provided statistical support to build the models 21 423 Provided advice on the writing of the sections on methods and results 22 23 424 24 425 Renee Otter, MD, PhD 25 426 Provided clinical expertise support to build cancer categories 26 427 Substantially contributed to the writing 27 428 28 429 29 Katrien Mortelmans, MD, PhD 30 430 Substantially contributed to the writing 31 431 32 432 Sarah Missine, MSc, PhD http://bmjopen.bmj.com/ 33 433 Provided advice and support to integrate the inequalities perspective in the paper and in the 34 434 statistical model 35 36 435 Supported the writing of the paper 37 436 38 437 Marc Arbyn, MD 39 438 Substantially contributed to the preparation of the tables and figures and the abstract section 40 439 on September 29, 2021 by guest. Protected copyright. 41 440 42 Prof. Catherine Bouland, MsC, PhD 43 441 As PhD co-supervisor for Miss Kiasuwa, Prof. C. Bouland substantially contributed to the 44 442 preparation of the study, the design and the revision of the paper 45 443 46 444 Prof. Christophe de Brouwer, MD, PhD 47 445 As the PhD supervisor for Miss Kiasuwa, Prof. C. de Brouwer substantially contributed to the 48 49 446 preparation of the study, the design and the revision of the paper 50 447 51 448 52 53 54 55 56 57 58 Version 07 July 2017 59 60 21

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 22 of 41

1 2 3 4 449 Funding and competing interests 5 6 450 This study has been funded by the National Institute for Health and Disability Insurance. We 7 8 9 451 have read and understood the BMJ policy on declaration of interests and declare that we have no 10 11 452 competing interests. 12 13 453 This article does not contain any studies with human participants or animals performed by any of 14 15 For peer review only 16 454 the authors. 17 18 455 19 20 21 456 Data sharing statement 22 23 457 The complete and anonymised dataset used for this study can be made available by the 24 25 26 458 corresponding author for researchers interested in comparative studies. This request would be 27 28 459 subject of approval from the Belgian National Institute for Health and Disability Insurance, the 29 30 460 Scientific Institute of Public Health and the Université Libre de Bruxelles. 31 32 http://bmjopen.bmj.com/ 33 34 35 461 Figures legends 36 462 Figure 1. The Belgian social security scheme related to work disability 37 38 463 39 464 Figure 2. Flowchart of the number of workers disabled because of cancer between 2007 and 2011 in 40 on September 29, 2021 by guest. Protected copyright. 41 465 Belgium. 42 43 466 44 467 Figure 3. Kaplan-Meier estimator for the time in work disability, stratified by the year of entrance 45 46 468 into work disability. 47 48 469 49 470 Figures 4 -8. Non-parametric cumulative incidence of ability to work by the end of follow-up, 50 51 471 stratified per a) year of entry, b) age at entry, c) gender, d) occupational class and e) cancer sites. 52 53 472 54 473 Figures 9-12. Box plots of time to any of the three events (death, retirement, ability to work), by 55 56 474 gender, age group and occupational class. 57 58 Version 07 July 2017 59 60 22

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 23 of 41 BMJ Open

1 2 3 475 4 5 476 6 7 477 Reference List 8 9 478 10 11 479 1. Amir Z, Wilson K, Hennings J, Young A: The meaning of cancer: implications for family finances 12 480 and consequent impact on lifestyle, activities, roles and relationships. Psychooncology 2012, 21: 13 481 1167-1174. 14 15 482 2. Daniel C.Lustig:For Causal peerRelationships Betweenreview Poverty and Disability.only Rehabilitation Counseling 16 17 483 Bulletin 2007, 50: 194-202. 18 19 484 3. Schuring M, Robroek SJ, Otten FW, Arts CH, Burdorf A: The effect of ill health and socioeconomic 20 485 status on labor force exit and re-employment: a prospective study with ten years follow-up in 21 486 the Netherlands. Scand J Work Environ Health 2013, 39: 134-143. 22 23 487 4. Reichard A., Stransky M., Phillips K., Drum C., Mc Clain M.: Does Type of Disability Matter to 24 488 Public Health Policy Practice? Californian Journal of Health Promotion 2015, 13: 25-36. 25 26 27 489 5. OECD. Sickness, Disability and Work. Breaking the Barriers. 2010. 28 490 Ref Type: Report 29 30 491 6. Luengo-Fernandez R, Leal J, Gray A, Sullivan R: Economic burden of cancer across the European 31 492 Union: a population-based cost analysis. Lancet Oncol 2013, 14: 1165-1174. 32 http://bmjopen.bmj.com/ 33 493 7. Betsch N: Kankerpatiënten en werkhervatting: rechten en plichten. Instituut Voor Arbeidsrecht, KU 34 494 Leuven; 2013. 35 36 37 495 8. Desiron HA, Donceel P, de RA, Van HE: A conceptual-practice model for occupational therapy to 38 496 facilitate return to work in breast cancer patients. J Occup Rehabil 2013, 23: 516-526. 39

40 497 9. Desiron HA, Donceel P, Godderis L, Van HE, de RA: What is the value of occupational therapy in on September 29, 2021 by guest. Protected copyright. 41 498 return to work for breast cancer patients? A qualitative inquiry among experts. Eur J Cancer Care 42 499 (Engl ) 2015, 24: 267-280. 43 44 500 10. Desiron HA, Crutzen R, Godderis L, Van HE, de RA: Bridging Health Care and the Workplace: 45 46 501 Formulation of a Return-to-Work Intervention for Breast Cancer Patients Using an Intervention 47 502 Mapping Approach. J Occup Rehabil 2016. 48 49 503 11. Rommel W. Werken na kanker:welke problemen ervaren (ex-)patiënten die het werk hervatten? 50 504 2012. Vlaams Liga Tegen Kanker. 51 505 Ref Type: Report 52 53 506 12. Tiedtke C, Donceel P, Knops L, Desiron H, Dierckx de CB, de RA: Supporting return-to-work in the 54 507 face of legislation: stakeholders' experiences with return-to-work after breast cancer in Belgium. 55 56 508 J Occup Rehabil 2012, 22: 241-251. 57 58 Version 07 July 2017 59 60 23

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 24 of 41

1 2 3 509 13. Tiedtke C, Dierckx de CB, Donceel P, de RA: Workplace support after breast cancer treatment: 4 510 recognition of vulnerability. Disabil Rehabil 2014, 1-7. 5 6 7 511 14. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 8 512 Chapter 12. Future need for research. Scand J Public Health Suppl 2004, 63: 256-258. 9 10 513 15. Alexanderson K, Hensing G: More and better research needed on sickness absence. Scand J Public 11 514 Health 2004, 32: 321-323. 12 13 515 16. Alexanderson K, Norlund A: Swedish Council on Technology Assessment in Health Care (SBU). 14 516 Chapter 1. Aim, background, key concepts, regulations, and current statistics. Scand J Public 15 For peer review only 16 517 Health Suppl 2004, 63: 12-30. 17 18 518 17. INAMI. Rapport annuel. Satistiques des indemnités. 2014. 19 519 Ref Type: Report 20 21 520 18. Coleman MP, Gatta G, Verdecchia A, Esteve J, Sant M, Storm H et al.: EUROCARE-3 summary: 22 521 cancer survival in Europe at the end of the 20th century. Ann Oncol 2003, 14 Suppl 5: v128-v149. 23 24 522 19. Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, Rosso S, Coebergh JW, Comber H et al.: Cancer 25 26 523 incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 27 524 2013, 49: 1374-1403. 28 29 525 20. Pryce J, Munir F, Haslam C: Cancer survivorship and work: symptoms, supervisor response, co- 30 526 worker disclosure and work adjustment. J Occup Rehabil 2007, 17: 83-92. 31 32 527 21. Kennedy F HCMFPJ: Returning to work following cancer: a qualitative exploratory study into the http://bmjopen.bmj.com/ 33 528 experience of returning to work following cancer. Eur J Cancer Care 2007, 16: 17-25. 34 35 36 529 22. Mehnert A: Employment and work-related issues in cancer survivors. Crit Rev Oncol Hematol 37 530 2011, 77: 109-130. 38 39 531 23. Mehnert A, de BA, Feuerstein M: Employment challenges for cancer survivors. Cancer 2013, 119

40 532 Suppl 11: 2151-2159. on September 29, 2021 by guest. Protected copyright. 41 42 533 24. Verbeek J, Spelten E, Kammeijer M, Sprangers M: Return to work of cancer survivors: a 43 534 prospective cohort study into the quality of rehabilitation by occupational physicians. Occup 44 45 535 Environ Med 2003, 60: 352-357. 46 47 536 25. Dal ML, Guzzinati S, Buzzoni C, Capocaccia R, Serraino D, Caldarella A et al.: Long-term survival, 48 537 prevalence, and cure of cancer: a population-based estimation for 818 902 Italian patients and 49 538 26 cancer types. Ann Oncol 2014, 25: 2251-2260. 50 51 539 26. Kiasuwa MR, Otter R, Mortelmans K, Arbyn M, Van OH, Bouland C et al.: Barriers and 52 540 opportunities for return-to-work of cancer survivors: time for action-rapid review and expert 53 54 541 consultation. Syst Rev 2016, 5: 35. 55 56 542 27. Hewitt M, Rowland JH, Yancik R: Cancer survivors in the United States: age, health, and disability. 57 543 J Gerontol A Biol Sci Med Sci 2003, 58: 82-91. 58 Version 07 July 2017 59 60 24

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 25 of 41 BMJ Open

1 2 3 544 28. Amir Z, Brocky J: Cancer survivorship and employment: epidemiology. Occup Med (Lond) 2009, 4 545 59: 373-377. 5 6 7 546 29. Collins C.G, Ottati A., Feuerstein M: Cancer Survivorship. In Handbook of Work Disability. 8 547 Prevention and Management. Springer edition. Edited by Loisel P, Anema J.R. New York: 2013:289- 9 548 302. 10 11 549 30. Tiedtke C, de RA, Dierckx de CB, Christiaens MR, Donceel P: Experiences and concerns about 12 550 'returning to work' for women breast cancer survivors: a literature review. Psychooncology 2010, 13 551 19: 677-683. 14 15 For peer review only 16 552 31. Tiedtke C, Dierckx de CB, de RA, Christiaens MR, Donceel P: Breast cancer treatment and work 17 553 disability: patient perspectives. Breast 2011, 20: 534-538. 18 19 554 32. van Muijen P., Duijts SF, van der Beek AJ, Anema JR: Prognostic factors of work disability in sick- 20 555 listed cancer survivors. J Cancer Surviv 2013, 7: 582-591. 21 22 556 33. Grunfeld EA, Cooper AF: A longitudinal qualitative study of the experience of working following 23 557 treatment for gynaecological cancer. Psychooncology 2012, 21: 82-89. 24 25 26 558 34. van Muijen P., Weevers NL, Snels IA, Duijts SF, Bruinvels DJ, Schellart AJ et al.: Predictors of return 27 559 to work and employment in cancer survivors: a systematic review. Eur J Cancer Care (Engl ) 2013, 28 560 22: 144-160. 29 30 561 35. Sterud T, Johannessen HA: Do work-related mechanical and psychosocial factors contribute to 31 562 the social gradient in long-term sick leave: a prospective study of the general working http://bmjopen.bmj.com/ 32 563 population in Norway. Scand J Public Health 2014, 42: 329-334. 33 34 35 564 36. Piha K, Laaksonen M, Martikainen P, Rahkonen O, Lahelma E: Interrelationships between 36 565 education, occupational class, income and sickness absence. Eur J Public Health 2010, 20: 276- 37 566 280. 38 39 567 37. Melchior M., Krieger N, Kawachi I., et al.: Work Factors and Occupational Class Disaprities in

40 568 Sickness Absence: Findings From the GAZEL Cohort Study. American Journal of Public Health on September 29, 2021 by guest. Protected copyright. 41 569 2005, 95: 1206-1212. 42 43 44 570 38. Melchior M, Goldberg M, Krieger N, Kawachi I, Menvielle G, Zins M et al.: Occupational class, 45 571 occupational mobility and cancer incidence among middle-aged men and women: a prospective 46 572 study of the French GAZEL cohort*. Cancer Causes Control 2005, 16: 515-524. 47 48 573 39. Melchior M, Krieger N, Kawachi I, Berkman LF, Niedhammer I, Goldberg M: Work factors and 49 574 occupational class disparities in sickness absence: findings from the GAZEL cohort study. Am J 50 575 Public Health 2005, 95: 1206-1212. 51 52 53 576 40. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 54 577 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 55 56 578 41. Lindbohm ML, Kuosma E, Taskila T, Hietanen P, Carlsen K, Gudbergsson S et al.: Early retirement 57 579 and non-employment after breast cancer. Psychooncology 2014, 23: 634-641. 58 Version 07 July 2017 59 60 25

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 26 of 41

1 2 3 580 42. Carlsen K, Hoybye MT, Dalton SO, Tjonneland A: Social inequality and incidence of and survival 4 581 from breast cancer in a population-based study in Denmark, 1994-2003. Eur J Cancer 2008, 44: 5 6 582 1996-2002. 7 8 583 43. Carlsen K, Dalton SO, Diderichsen F, Johansen C: Risk for unemployment of cancer survivors: A 9 584 Danish cohort study. Eur J Cancer 2008, 44: 1866-1874. 10 11 585 44. Gerkens S, Merkur S: Belgium: Health system review. Health Syst Transit 2010, 12: 1-266, xxv. 12 13 586 45. Kiasuwa Mbengi R. La réinsertion socioprofessionnelle des patients atteints de cancer. Exposé des 14 587 positions. 2014. 15 For peer review only 16 588 Ref Type: Online Source 17 18 589 46. Onkelinx L. Toespraak van de Minister van Volksgezondheid en Sociale Zaken op het Symposium 19 590 Evaluatie van het Kankerplan. 2012. 20 591 Ref Type: Report 21 22 592 47. de Rijk A.: Work Disability Theories: A Taxonomy for Researchers. In Handbook of Work 23 593 Disability. Prevention and Managment. Edited by Loisel P., Anema J.R. New York: Springer; 24 25 594 2013:475-499. 26 27 595 48. Jason P.Fine and Robert J.Gray: A Proportional Hazards Model for the Subdistribution of a 28 596 Competing Risk. Journal of the American Statistical Association 1999, 94: 496-509. 29 30 597 49. Belgian Cancer Registry. Cancer Survival in Belgium. 2004-2008. 1-109. 2012. 31 598 Ref Type: Report 32 http://bmjopen.bmj.com/ 33 599 50. Rijken M, Spreeuwenberg P, Schippers J, Groenewegen PP: The importance of illness duration, 34 35 600 age at diagnosis and the year of diagnosis for labour participation chances of people with 36 601 chronic illness: results of a nationwide panel-study in The Netherlands. BMC Public Health 2013, 37 602 13: 803. 38 39 603 51. Regional variation in incidence for smoking and alcohol related cancers in Belgium. Edited by Kris

40 604 Henau, Elizabeth Van Eycken, Geert Silverman, Eero Pukkala. Cancer Epidemiology 39[1], 55-65. on September 29, 2021 by guest. Protected copyright. 41 605 2016. 42 43 606 Ref Type: Generic 44 45 607 52. Mehnert A, Koch U: Predictors of employment among cancer survivors after medical 46 608 rehabilitation--a prospective study. Scand J Work Environ Health 2013, 39: 76-87. 47 48 609 53. Ervasti J, Kivimaki M, Dray-Spira R, Head J, Goldberg M, Pentti J et al.: Socioeconomic gradient in 49 610 work disability in diabetes: evidence from three occupational cohorts. J Epidemiol Community 50 611 Health 2016, 70: 125-131. 51 52 53 612 54. Woods LM, Rachet B, Coleman MP: Origins of socio-economic inequalities in cancer survival: a 54 613 review. Ann Oncol 2006, 17: 5-19. 55 56 614 55. Holm LV, Hansen DG, Larsen PV, Johansen C, Vedsted P, Bergholdt SH et al.: Social inequality in 57 615 cancer rehabilitation: a population-based cohort study. Acta Oncol 2013, 52: 410-422. 58 Version 07 July 2017 59 60 26

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 27 of 41 BMJ Open

1 2 3 616 56. Krokstad S, Johnsen R, Westin S: Social determinants of disability pension: a 10-year follow-up of 4 617 62 000 people in a Norwegian county population. Int J Epidemiol 2002, 31: 1183-1191. 5 6 7 618 57. Noordzij M, Leffondre K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ: When do we need 8 619 competing risks methods for survival analysis in nephrology? Nephrol Dial Transplant 2013, 28: 9 620 2670-2677. 10 11 621 58. Glimelius I, Ekberg S, Linderoth J, Jerkeman M, Chang ET, Neovius M et al.: Sick leave and 12 622 disability pension in Hodgkin lymphoma survivors by stage, treatment, and follow-up time-a 13 623 population-based comparative study. J Cancer Surviv 2015. 14 15 For peer review only 16 624 59. Bradley CJ, Bednarek HL: Employment patterns of long-term cancer survivors. Psychooncology 17 625 2002, 11: 188-198. 18 19 626 60. Scott DA Mills M, Black A, Cantwell M, Campbell A, Cardwell CR, Porter S et al.. Multidimensional 20 627 rehabilitation programmes for adult cancer survivors (Review). Cochrane.Database.Syst.Rev. [3]. 21 628 2016. 22 629 Ref Type: Generic 23 24 25 630 61. Mewes JC, Steuten LM, Groeneveld IF, de Boer AG, Frings-Dresen MH, IJzerman MJ et al.: Return- 26 631 to-work intervention for cancer survivors: budget impact and allocation of costs and returns in 27 632 the Netherlands and six major EU-countries. BMC Cancer 2015, 15: 899. 28 29 633 62. de Boer AG, Taskila TK, Tamminga SJ, Feuerstein M, Frings-Dresen MH, Verbeek JH: Interventions 30 634 to enhance return-to-work for cancer patients. Cochrane Database Syst Rev 2015, 9: CD007569. 31 http://bmjopen.bmj.com/ 32 635 63. Zaman AC, Bruinvels DJ, de Boer AG, Frings-Dresen MH: Supporting cancer patients with work- 33 34 636 related problems through an oncological occupational physician: a feasibility study. Eur J Cancer 35 637 Care (Engl ) 2015. 36 638 37 639 38 39 640

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 Version 07 July 2017 59 60 27

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 28 of 41

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure. Overall survival stratified by the year of entrance into long-term disability on September 29, 2021 by guest. Protected copyright. 41 42 127x127mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 31 of 41 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 Figure 5. Cumulative incidence of ability to work stratified by the age at entry into long-term disability 43 44 60x65mm (300 x 300 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 33 of 41 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 Figure 6. Cumulative incidence of ability to work stratified by gender 44 45 59x65mm (300 x 300 DPI) 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 34 of 41

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 43 Figure 7. Cumulative incidence of ability to work stratified by occupational class 44 60x65mm (300 x 300 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 35 of 41 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 42 Figure 8. Cumulative incidence of ability to work stratified by cancer site 43 44 61x65mm (300 x 300 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 36 of 41

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 9. Box plot of time to any event (death, retirement or ability to work) stratified by the age at entry on September 29, 2021 by guest. Protected copyright. 41 into long-term disability 42 43 65x65mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 37 of 41 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 10. Box plot of time to any event (death, retirement or ability to work) stratified by gender on September 29, 2021 by guest. Protected copyright. 41 42 65x65mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 38 of 41

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 Figure 11. Box plot of time to any event (death, retirement or ability to work) stratified by occupational class on September 29, 2021 by guest. Protected copyright. 41 42 65x65mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from Page 39 of 41 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 29, 2021 by guest. Protected copyright. 41 Figure 12. Box plot of time to any event (death, retirement or ability to work) stratified by cancer site 42 64x65mm (300 x 300 DPI) 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2016-014094 on 17 February 2018. Downloaded from BMJ Open Page 40 of 41