Risk factors for return to work in colorectal cancer survivors

Abstract Background: The increasing incidence of colorectal cancer among individuals in the productive age‐group has adversely affected the labor force and increased healthcare expenses in recent years. Return to work (RTW) is an important issue for these patients. In this study, we explored the factors that influence RTW and investigated the influence of RTW on survival outcomes of patients with colorectal cancer. Methods: Data of individuals (N = 4408) in active employment who were diagnosed with colorectal cancer between 2004 and 2010 were derived from 2 nationwide databases. Subjects were categorized into 2 groups according to their employment status at 5‐year follow‐up. Logistic regression analysis was performed to identify the factors associated with RTW. Survivors were further followed up for another 8 years. Propensity score matching was applied to ensure comparability between the two groups, and survival analysis was performed using the Kaplan–Meier method. Results: In multivariable regression analysis for 5‐year RTW with different characteristics, older age (OR: 0.57 [95% CI, 0.48–0.69]; p < 0.001), treatment with radiotherapy (OR: 0.69 [95% CI, 0.57–0.83]; p < 0.001), higher income (OR: 0.39 [95% CI, 0.32–0.47]; p < 0.001), medium company size (OR: 0.78 [95% CI, 0.63–0.97]; p = 0.022), and advanced pathological staging (stage I, OR: 16.20 [95% CI, 12.48–21.03]; stage II, OR: 13.12 [95% CI, 10.43–16.50]; stage III, OR: 7.68 [95% CI, 6.17–9.56]; p < 0.001 for all) revealed negative correlations with RTW. In Cox proportional hazard regression for RTW and all‐cause mortality, HR was 1.11 (95% CI, 0.80–1.54; p = 0.543) in fully adjusted model. Conclusion: Older age, treatment with radiotherapy, higher income, medium company size, and advanced pathological stage showed negative correlations with RTW. However, we observed no significant association between employment and all‐cause mortality. Further studies should include participants from different countries, ethnic groups, and patients with other cancers.


| INTRODUCTION
Progressive population growth and aging have led to increased incidence of cancer and cancer-associated mortality in recent years. 1,2 Improved cancer screening and developments in therapeutic modalities have advanced the overall survival rate of cancer patients. This has also contributed to increased diagnosis of cancer in younger age-groups and an increasing number of cancer survivors in the productive agegroup. [2][3][4][5] The reduced working ability has an adverse effect on these patients as well as the society at large. Thus, there is an increasing interest in maintaining the employment of cancer survivors. 6 Colorectal cancer (CRC) is the third most common cancer in the world, accounting for 10.2% of all malignancies; an estimated 1.8 million cases of CRC are newly diagnosed every year. 1,7 The epidemiological patterns of CRC tend to vary in different parts of the world; however, some distinct trends are observed globally, that is, increases incidence and mortality, decreased mortality rate, and increasing younger age at diagnosis. 1,[7][8][9] Studies have shown that more than half of all cancer survivors avail a period of sick leave for receiving cancer therapy and to cope with the associated disability; in addition, most of these patients returned to work after treatment. 2,5,10 However, cancer patients were still found to have a higher risk of job loss, less probability of re-employment, and longer time for returning to work. 3,11,12 Furthermore, unemployment among cancer survivors was shown to adversely affect their quality of life (QoL); in addition, the reduced household income, declined physical ability and their psychosocial repercussions were shown to influence the prognosis of underlying diseases. 6,[13][14][15] Studies have also shown that being employed inculcates a sense of accomplishment, self-esteem, and normalcy. [16][17][18][19] From a societal perspective, the financial implication of resources spent on medical care, welfare, and reduction of the labor force due to absenteeism imposes an extra burden on the government. Therefore, there is increasing awareness of the importance of rehabilitation interventions for cancer survivors to facilitate their return to the work force. 13,20 However, to the best of our knowledge, no study has directly investigated the correlation between return to work (RTW) and survival outcomes.
Since maintaining the employment is a key concern for cancer patients, identification of factors that influence employment status is imperative. Several studies have explored the factors that influence the employment status among cancer survivors. 13,[21][22][23] Some of these studies have yielded inconsistent results depending on the cancer site or study area. Most studies that have investigated the correlates of change in employment status were based on European and American data. There is a paucity of studies conducted in Asia, which is home to 60% of the global population and accounts for approximately half of all cancer cases and cancer deaths. 1 In this study, we analyzed the data of employees who were diagnosed with CRC in Taiwan. The aim was to identify factors associated with RTW and to investigate the correlation between RTW and survival outcomes in CRC patients.

| Study design
This was a nationwide, retrospective cohort study. Data for this study were derived from two nationwide databases in Taiwan: National Health Insurance Research Database (NHIRD) and Labor Insurance Database (LID). Employees who were diagnosed with CRC between 2004 and 2010 were enrolled initially. Participants were followed up for 5 years after diagnosis of CRC. We analyzed the relationship of various variables with RTW in the 5th year after CRC diagnosis. Subsequently, the surviving patients were divided into RTW and non-RTW groups depending on their employment status and followed up for another 8 years. Lastly, we compared the survival outcomes in the two groups.

| Database
NHIRD is a nationwide database that contains sociodemographic (e.g., sex, age, residence) and health service-related information (e.g., health facility, clinical diagnosis, treatment details) of approximately 23 million residents in Taiwan. These data were obtained from National Health Insurance (NHI), an insurance system launched by the Taiwan government in 1995. The NHI had enrolled over 99% of Taiwan's population. In this study, we obtained health-related information from the NHIRD. Comorbidities and cancer diagnosis were derived according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes.
LID is another nationwide database, which was derived from the labor insurance system in Taiwan. The Taiwan government regulations require mandatory enrolment of all fulltime employees in labor insurance unless they quit their job. This database provides socio-demographic and labor-related (e.g., industry, company size, income) information. The industrial classification in LID is according to the industry distribution system, 9th revision of Executive Yuan, Taiwan, which is based on the International Standard Industrial Classification of All Economic Activities (ISIC), revision 4.

| Participants
From the NHIRD, we extracted data pertaining to all people aged ≥20 years who were newly diagnosed with CRC between 2004 and 2010. The dataset of CRC was identified according to the International Classification of Diseases for Oncology, third edition (ICD-O-3, code C18-C21). Among these patients, those with other primary malignancies were excluded. Subsequently, we linked the above dataset with LID and selected those individuals whose employment status was "under employment" or "self-employed" at the time of CRC diagnosis. A total of 4408 full-time employees were eligible for inclusion.

| Outcome measures
The primary outcome of this study was RTW 5 years after CRC diagnosis. Employment status was recorded and checked according to the data in LID. Each participant was followed up until death or the completion of a 5-year follow-up. These participants were divided into two groups, "RTW" and "non-RTW," based on the employment status at the 5th year after CRC diagnosis. RTW group included the participants who remained in the workforce with or without sick leave after a cancer diagnosis. Individuals who ceased working and did not RTW were classified as a non-RTW group. The correlates of RTW were analyzed in order to investigate the determinants of RTW in CRC patients.
The secondary outcome was long-term survival. Survival data were acquired through detecting the registration of participants in NHIRD. The surviving participants in the RTW and non-RTW groups at the 5th year were followed up for another 8 years. We applied propensity score matching in a 1:1 ratio before survival analysis. All-cause mortality was compared between the RTW and non-RTW groups to assess the correlation between RTW and survival. The study protocol is shown in Figure 1.

| Statistical analysis
The SAS 9.3 (SAS Institute) statistical package was used for data analysis. Continuous and categorical variables are presented as mean ± standard deviation and frequency (percentage), respectively. Between-group difference with respect to demographic characteristics and comorbid medical disorders Company size: small (less than 5 people), medium (less than 200 people in manufacturing, construction, mining, and quarrying; or less than 100 people in other industries), large (more than 200 people in manufacturing, construction, mining, and quarrying; or more than 100 people in other industries). were assessed using the independent sample t-test and Chisquared test. Univariate and multivariate logistic regression analyses were performed to assess the effect of each demographic characteristic on RTW. Variables that showed a significant association in the univariable model were included in the multivariate model.
In the analysis of all-cause mortality and RTW, propensity score matching was applied at baseline. Survival analysis was performed using the Kaplan-Meier method and differences between the RTW and non-RTW groups were assessed using the log-rank test. Univariate and multivariate Cox proportional hazard regressions were applied. Two-sided p values less than 0.05 were considered indicative of statistical significance.

| Characteristics of the study population
The study population comprised of 4408 employees who were diagnosed with CRC and underwent a 5-year follow-up of their employment status. The demographic characteristics of the study population are summarized in Table 1. A total of 2255 participants remained in the work force (1943 worked at the same company and 312 changed their jobs) while 2153 had quit their jobs without return to employment (802 unemployed and 1351 died) in the 5th year after diagnosis of CRC.

| Association of RTW with allcause mortality
To assess the influence of RTW on survival, we analyzed the correlation between RTW and all-cause mortality. After propensity score matching, there were 775 participants each in the RTW and non-RTW groups. Table 4 shows the demographic characteristics of the propensity score-matched cohort.
The result of Cox proportional hazard regression for RTW and all-cause mortality was presented in hazard ratios (HRs). HR was 0.94 (95% CI, 0.70-1.25; p = 0.652) in unadjusted model, and 1.11 (95% CI, 0.80-1.54; p = 0.543) in fully adjusted model. Figure 2 showed the result of survival analysis in Kaplan-Meier plot. No statistically significant difference was observed in all-cause mortalities among RTW and non-RTW groups.

| DISCUSSION
There were two main objectives of this study. The first objective was to assess the impact of demographic characteristics, health-related variables, and labor-related variables on RTW. The second objective was to assess the correlation between RTW and long-term survival of CRC survivors.
Among the characteristics that influenced employment status, age, gender, comorbidity (hypertension and cerebrovascular disease), treatment, living area, income, occupation, company size, and pathological stage showed a significant difference between RTW and non-RTW groups by 5 years after  13 To integrate these findings, we can identify some common factors that affect the RTW. First, financial issue was the primary concern that made patients RTW. 6,17,18,26 Irrespective of the cancer type and demographic characteristics, most cancer survivors indicated financial pressure as their primary consideration while deciding whether to continue and RTW. 18,27 Apart from income, the role of insurance has also been widely discussed. Adequate health insurance provides financial support, which increases the affordability of medical expenses and allows  Company size: small (less than 5 people), medium (less than 200 people in manufacturing, construction, mining and quarrying; or less than 100 people in other industries), large (more than 200 people in manufacturing, construction, mining and quarrying; or more than 100 people in other industries).

T A B L E 4 (Continued)
| 3947 patients to take time off for their cancer therapy without the apprehension of being unemployed. Furthermore, some studies revealed the correlation between marital status and change in employment status, which was also attributed to financial considerations. Married persons were shown less likely to RTW than singles as that they may have financial support from their partners. 28 On the contrary, people who were the only or the main source of income in their family are likely to experience greater financial pressure. 26 Second, RTW is also based on adequate physical condition and working ability. The poorer the physical status, the less is the probability of RTW. Although there were no quantified performance status variables such as Eastern Cooperative Oncology Group (ECOG) or Karnofsky performance score in this study, some previous studies have found that the impact of cancer type, staging, comorbidity, and treatment decision on change in employment may reflect the patients' physical status. 10,13,21,29 Decline in the ability to perform work and activities of daily living are a barrier for patients seeking a return to employment. Some patients chose to retire from their work after cancer diagnosis, while others RTW after perceiving the adequacy of their physical status. 18 Third, psychosocial factors also have an important influence on the decision to RTW. These factors include family, workplace environment, and the patients' mental status. We did not investigate these aspects in the present study. An exploratory study investigated the RTW experience of cancer patients, by performing patient interviews. The study elicited several considerations of patients. 18 Some patients went back to their work to acquire a sense of normality, while others returned to work due to their perceived sense of responsibility and feeling of loyalty toward their work. Studies have also indicated the importance of support from the employers and colleagues. 30 Table 5 highlights the facilitators and barriers for employment status identified in studies that included CRC patients. 10,12,21,22,24,25,28,[31][32][33][34][35] The present study had a distinctly large sample size (N = 4408). Lower income and undergoing surgery were identified as facilitators for employment and RTW, whereas older age, male sex, and advanced pathological stage were identified as barriers to employment and RTW. Income reflected a person's financial ability. Patients with higher income are likely to be more financially secure. In contrast, those with lower income might be forced to RTW as soon as possible due to their financial constraints. Advanced disease represents poorer physical activity, which imposed a burden on cancer survivors returning to their work. The impact of age on RTW is determined by both financial factors and physical ability. In general, aging is associated with the decline in physical condition. Furthermore, elderly tend to have better financial stability than middle-aged and young people. Both these aspects explain the negative correlations between age and RTW.
Of note, the observed influence of "income" on employment status in our study was not consistent with the result of previous studies. In our study, lower income was found to be a facilitator for RTW; however, other studies have yielded opposite findings. 21,32,34 This discrepancy is likely attributable to economic factors peculiar to Taiwan. Due to NHI coverage, health care and medical treatment in Taiwan is less expensive than that in most other countries. The financial stress in Taiwan is mainly reflected to the reduced productivity T A B L E 5 (Continued) due to sick leave or job loss, which increases the need for survivors with lower income to RTW. On the other hand, financial stress in other countries is mainly due to the medical expenses. Patients with higher income are more likely to receive better treatment, which explains the better outcomes and better preserved ability for working. However, there were no standard criteria to define income level in previous studies. Future studies with standardization of income level strata are required to identify correlation between income level and subsequent employment status. Apart from the factors that affect employment status, very few studies have investigated the influence of RTW on cancer survivors. In this study, we investigated the correlation between RTW and survival of CRC patients in Taiwan. We believe that the better survival of patients who RTW may be attributable to the following reasons. First, work ability is influenced by a combination of individuals' physical, psychological, and social resources. 2 Patient who RTW are likely to have better physical and mental status, which is liable to contribute to better survival outcomes. Second, RTW may have a positive influence on the physical and mental health of patients. Mahar et al. found that patients who continued working showed better physical and mental functioning, QoL, and lower psychosocial distress than patients who RTW with sick leave and patients who discontinued working after cancer diagnosis. 36 However, in this study, we observed no significant difference in all-cause mortality between RTW and non-RTW groups. This may be attributable to minimization of selection bias after the use of statistical techniques such as propensity score matching. The similar baseline characteristics in both groups may have annulled the influence of better physical and mental status on survival in the RTW group. Nevertheless, we did not evaluate other outcomes such as QoL, physical function, or psychosocial status between the RTW and non-RTW groups. The impact of RTW on outcomes among cancer survivors remains uncertain.
A key limitation of this study was that we grouped the participants according to their employment status at the time of follow-up, which means that randomization was unavailable in our study. Other limitations include the lack of quantified performance status data and the absence of tools to evaluate the quality of RTW. Moreover, the outcome measure was confined to survival and we did not measure other indices such as QoL. Lastly, the study population exclusively comprised of CRC patients in Taiwan. Future studies including participants from different countries and ethnic groups, and patients with other cancers are required to elucidate the impact of RTW on cancer survival.