Abstract
BACKGROUND:
Studies about work and cancer predominantly considered the return to work of cancer survivors. However, some studies highlighted that many patients work with cancer even immediately after the diagnosis. Little is known about the frequency, causes, and consequences of this behavior.
OBJECTIVE:
This study aimed to estimate how many cancer patients continue working in the month after the diagnosis in an Italian context and to determine which factors affect the decision to stop working in the same period.
METHODS:
One hundred seventy-six patients with breast, gastrointestinal, prostate, or female reproductive system cancer completed a survey with demographic, occupational, and psychosocial information. Clinical information was collected from medical records. We measured how many workers continued working in the month after cancer diagnosis without substantial interruptions and selected the best logistic regression model of this behavior’s predictors.
RESULTS:
Sixty-eight percent of the patients continued working in the month after the diagnosis. Patients were more likely to stop working with a higher level of perceived work-health incompatibility (OR = 2.64; 95%CI: 1.48–4.69), an open-ended contract (OR = 3.20; CI: 1.13–9.09), and a complex treatment (surgery+chemo-/radio-therapy, OR = 4.25; CI: 1.55–11.65) and less likely with breast cancer (OR = 0.20; CI: 0.07–0.56), and more children (OR = 0.59; CI: 0.37–0.96).
CONCLUSIONS:
To continue working with cancer is a common practice among the newly diagnosed. The decision to suspend work activity relates to evaluating how much work activities hamper one’s health care needs and the practical difficulties expected in handling cancer care and work.
Introduction
Several studies highlighted the importance of employment for cancer patients [1, 2], but there are substantial differences in personal decisions, feelings, and attitudes towards the possibility of returning to work. Different patterns of employment after cancer are recognized in the literature [3–6], but most of the studies have focused on the rate of return to work (RTW) and its predictive factors. This exclusive focus implies a unique standardized pattern of work trajectory: diagnosis, work suspension, and RTW or unemployment/retirement. However, this approach does not represent the complexity of cancer patients’ work experiences. Few studies have examined work behavior shortly after a cancer diagnosis or during treatment, as well as its determinants and implications for long-term employment [4, 7–14]. To stop working after diagnosis and then RTW is not the same experience as to continue working after diagnosis and during the disease. Petersson et al. [11] argued that we need more knowledge for all disease trajectory phases to prevent long-term adverse effects, and according to Moskowitz, occupational health practitioners should consider the diverse paths of individuals [3]. Nachreiner [13] suggested better understanding the typical work patterns shortly after diagnosis to help occupational health professionals and treating physicians recommend work plans for their patients.
The prevalence of cancer patients’ decisions to continue working without substantial interruptions during the first days after diagnosis is unclear. The proportion of patients working in the first month after diagnosis or during treatment varies in the different studies from 18%to 67%[4, 11–14].
The month after diagnosis is a relevant time frame for work-related decisions in cancer patients. A cohort study from Sweden observing 2738 cancer patients reported that the most substantial increase in the proportion of cancer patients with more than 15 days of sick leave occurred in the 30 days after diagnosis [9]. A US study reported that the month in which gynecological cancer survivors most frequently stopped working was the first one [13]. Finally, a study from Japan reported that 40%of job status changes after cancer diagnosis took place between the diagnosis and the beginning of the treatment [10]. Therefore, the present study’s first aim is to measure the proportion of patients that decided to continue working without substantial interruptions during the month after diagnosis among Italian workers diagnosed with breast, gastrointestinal, prostate and gynecological cancers. Survivors of these types of cancers are more likely to be unemployed than healthy workers [15].
Clinical and work-related psychosocial factors have been recently highlighted as important predictors of RTW after cancer [16–18]. Recent studies and reviews reported that RTW after cancer was affected by treatment type, cancer severity, comorbidity, job insecurity, job satisfaction, work autonomy, support from the workplace, and attitude and value towards work [17, 19–24]. However, no studies establish if the factors that influence RTW after cancer also affect patients’ work decisions shortly after a cancer diagnosis.
The first studies investigating factors associated with the different work behaviors after a cancer diagnosis analyzed only a few of the factors relevant for RTW [7, 12]. Only one study also considered psychosocial factors and revealed that working during treatment was associated with a higher level of work flexibility and disclosure to colleagues [7]. However, the same study discovered that working during treatment was associated with difficulties in managing fatigue and no paid time off to attend medical appointments [7, 25]. The other studies examined clinical and sociodemographic variables. They found that work during treatment [8] or post-diagnosis [12] was associated with cancer stage lower than IV, having prostate cancer compared to breast cancer, not having undergone surgery, being self-employed, and with a low household income. The second aim of the present research is to verify if factors that influence the RTW after cancer sick leave also affect the decision to stop working in the month following the cancer diagnosis. We will consider clinical (i.e., treatment type and cancer severity) and work-related psychosocial factors (i.e., job insecurity, job satisfaction, work autonomy, support from the workplace, and attitude towards work) and control for relevant sociodemographic variables.
Methods
Patients and procedures
This study provides cross-sectional results from the baseline data of a larger monocentric prospective cohort study conducted at one cancer hospital in northern Italy. Inclusion criteria were age between 18 and 65; a primary diagnosis of breast, gastrointestinal, prostate, or female reproductive system cancer with no metastasis; paid employment at the time of diagnosis; sufficient language skills to complete the questionnaires. Exclusion criteria were having a central nervous system disease or other disabling diseases and being under treatment with psychoactive drugs.
Two psychologists conducted the recruitment at the hospital between June 2014 and June 2015 using a similar in-person description and invitation. Patient information sheets were provided, and written consent was obtained before questionnaire completion during the cancer treatment (Mdn = 85 days after the first diagnosis).
The study was in accordance with the ethical standards of the institutional research committee (registration: R83/14-IEO 95, May 2014) and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.
Measures
Sociodemographic factors
We asked the participants about their education, financial status, affective relationship, civil state, and the number of children.
Clinical factors
Type and stage of cancer, treatment received, date of diagnosis, and comorbidity were collected from the patient’s medical records.
Work status
We collected information about job type, work contract, job position, physical job exertion, paid days off work, and the patient’s knowledge about the legal rights of workers affected by cancer.
Work-related psychosocial factors
We operationalized (positive) attitude and (negative) value towards work with work engagement and work disaffection. Work engagement was measured with the short version of the Utrecht Work Engagement Scale [26] (nine items, Cronbach’s α= 0.93). Work disaffection after cancer was measured with the cynicism scale of the Maslach Burnout Inventory - General Survey [27] (four items, Cronbach’s α= 0.82). We measured cognitive and affective job insecurity [28] (eight items, Cronbach’s α= 0.76).
The overall job satisfaction was measured with a single item. The single item measure’s reliability and validity have been extensively supported [29, 30].
To measure perceived work autonomy, we used the Autonomy scale of the Work design questionnaire [31] that measures work scheduling and decision–making autonomy (six items, Cronbach’s α= 0.94).
Support from the workplace has been measured with several scales that assess different facets of support. With a short version of the Psychosocial safety climate scale [32], we measured the extent to which the patient perceived that policies, practices, and procedures for the protection of workers’ psychological health and safety were present in the workplace (four items, Cronbach’s α= 0.88). We asked the patients if they had disclosed cancer to their supervisor and coworkers with two items. We measured the perceived support for health problems available in the workplace and the dissatisfaction in the management of personal health care due to work commitments with two scales of the Work-Health Balance questionnaire [33]: the External Support scale (six items, Cronbach’s α= 0.77) and the Work-Health Incompatibility scale (six items, Cronbach’s α= 0.81), respectively.
Work suspension
To measure the work suspension shortly after diagnosis, we asked the patients if they stopped working for more than two consecutive weeks or three fragmented weeks in the month after the cancer diagnosis. We considered the month after the diagnosis as our time frame because it has been identified as the period with the most substantial increase in cancer patients’ proportion on sick leave [9]. Moreover, it is sufficiently long to distinguish between stable and occasional work absences. We defined the limit of two consecutive weeks because we wanted to detect a stable job suspension that was not caused by contextual and temporary factors like days off work for medical visits, cancer treatment, or recovery days. We established the limit of three fragmented weeks of work absence to include workers who did not stop working for more than two consecutive weeks in a month but could not be genuinely involved in work activities working for seven or fewer days in the month.
Analysis
Data analysis was performed via R [34]. Our database presented missing values that we assumed were at least missing at random (MAR). We adopted the multiple imputation techniques (MIs), imputing a full dataset ten times with the R package “Amelia” [35]. To estimate the rate among cancer patients of work suspension shortly after diagnosis, we calculated the probability of stopping working for more than two consecutive weeks or three fragmented weeks in the month after the cancer diagnosis. To verify if the factors with an effect on RTW after cancer also affect the decision to stop working shortly after the cancer diagnosis, we searched the best model associated with work suspension. All the predictors were separately entered into a univariate logistic regression analysis against the outcome variable in each imputed dataset. Results for each predictor from the ten imputations were pooled into a final estimate following Rubin’s rules using the R package “Zelig” [36]. The predictors significantly associated with the outcome at a P-value lower than 0.10 were considered candidates for the final best model. To select the best model, we adopted the “majority method” [37]. We performed a backward selection for each of the ten imputed datasets with a probability to remove of 0.175, thus obtaining ten sets of predictors. The final model included those predictors that appeared in the final sets of predictors seven or more times (inclusion frequency level of 70%) [38].
The use of the variable selection procedure leads to biased, overestimated parameters if the same data are used for both selection and estimation. The shrinkage methods have been shown to be simple and effective approaches to address overestimation [39]. Given the presence of categorical variables in our model, we opted for the use of the “joint shrinkage factors” over the “parameterwise shrinkage factors” [40]. The closer the shrinkage factor to 0, the more likely an overestimate has occurred; values close to 1 indicate that an overestimation is highly unlikely. Less biased regression coefficients estimates are obtained by multiplying the joint shrinkage factors (JSF) with the commonly estimated predictors [39, 40]. We computed JSFs of the predictors in the final model for each of the ten datasets using the R package “shrink” [40]. Then, we calculated the average among datasets for each predictor [37].
In our analysis, odds ratios higher than one indicate increases in the probability of stopping working in the month after diagnosis. To allow a more straightforward interpretation, we calculated two representative variations in the probability of stopping working generated by the adjusted ORs in the final model 1 First, we provided the highest variation in the probability that the considered OR could generate. Second, we provided the variation in the probability generated by the considered OR when the reference group has the probability of stopping working equal to the one of the whole sample [41].
Results
We contacted 187 eligible patients, and 179 agreed to participate in the study. Among these, three did not return the filled questionnaire. Therefore, the first section of the analysis was conducted on a sample of 176 patients. Six percent of the entries in our database was missing. Table 1 shows the descriptive statistics of the sample. We found that 32%of patients stopped working in the month after the diagnosis. Hence, the work continuation rate was 68%.
Sample characteristics
Sample characteristics
Table 2 shows the univariate relations of all the variables measured in the study with the decision to stop working in the month after diagnosis. Two sociodemographic factors (i.e., have a partner and number of children), one work status indicator (i.e., type of contract), two clinical factors (i.e., cancer and treatment type), and five psychosocial factors (i.e., work engagement, job satisfaction, job autonomy, work-health incompatibility, and work disaffection) had a p-value lower than 0.10. They, therefore, were used in the subsequent best model selection procedure. All these factors were also statistically significant (p < 0.05)
Univariate odds ratio (OR) with 95%confidence interval (CI) for interrupting work activity in the month after diagnosis (n = 176)
Note. Parameters in bold had p-value lower than 0.10.
The best model selection procedure resulted in five factors associated with the interruption of work activity during the month after diagnosis. Table 3 lists the multivariate logistic regression analysis results of the final model with the respective JSFs and adjusted coefficients. Patients were more likely to work after diagnosis with more children and a breast cancer. Conversely, patients were more likely to stop working with an open-ended contract compared to self-employment, if they had to undergo surgery plus chemo-/radio- therapy compared to only surgery, and with higher levels of perceived work-health incompatibility.
The final model according to the “majority method”, the joint shrinkage factors (JSFs), the respective shrunken odd ratios, and corresponding possible effects on the probabilities of work suspension
Note. aNagelkerke pseudo-R2 of 0.32; bGenerated Δp = The possible differences in the probability of work suspension generated by one-unit increase of the predictor given the shrunken OR; cHighest = the highest difference in probabilities that the considered OR can generate; dBase rate = the difference in probability if the reference group has the 32%of probability to suspend the work activity in the month after cancer.
In this study, we aimed to estimate the proportion of working patients who continued to work with cancer in the month after the diagnosis and then evaluate the factors associated with this behavior.
We found that 68%of patients continued working in the month after diagnosis. Our work continuation rate is in line with other studies with similar [13, 14] or much larger samples size [4, 9]. However, there are other studies with considerably lower continuation rates [7, 12]. The absence of a standardized definition and measurement of work continuation after diagnosis may lead to very different results. Our finding, consistently with other researches [4, 13], suggests that most cancer patients do not stop working while research focuses primarily on cancer patients that did so and then seek to return to work. Future researches should investigate predictors and consequences of working with cancer and during treatment.
We, then, found that both sociodemographic (i.e., number of children and type of work contract), clinical (i.e., type of treatment and cancer), and psychosocial (i.e., work-health incompatibility) factors influence the decision to stop working during the month after cancer diagnosis.
Sociodemographic factors indicate that family and job conditions influence patients’ decision. For each additional child, the likelihood of work suspension decreases by about 9%. Future studies could explain this result, but one possible explanation is that parents with cancer want to keep their daily lives as normal as possible not to cause concern to their children. In line with previous studies [8, 12], self-employed workers are about 18%more likely than workers with an open-ended contract to continue working. This effect is consistent with the general trend observed in the literature showing that fewer self-employed cancer patients take time off work than salaried survivors [42]. Self-employed are likely forced to continue working because their income strictly depends on the days worked. On the contrary, salaried workers have better social security provisions with an income guaranteed by the social security system [12, 42].
The results about the clinical factors substantially replicate the significance of the same factors found in the Sharp and Timmons [12] study (i.e., type of treatment and cancer) but with different specific effects. In the present study having to undergo both surgery and chemo or radio –therapy increased the probability of work suspension by about 25%compared to surgery only. The effect of treatment type may relate to disease severity perception or the practical difficulties patients expect to face in managing work and treatment. More extended and diversified treatment may foster the perception of having more severe disease and complicate the organization of work during treatment. More studies should investigate this issue because in Sharp and Timmons [12], surgery determined higher chances of work suspension. The other clinical factors that significantly influenced work continuation was the cancer type. Specifically, breast cancer was associated with around 25%more chances to continue working compared to the other types of cancer considered as a single category.
In the univariate analyses, many psychosocial factors were associated with work continuation (i.e., work engagement, job satisfaction, job autonomy, work health incompatibility, and work disaffection). However, in the final model, only work-health incompatibility was retained. Work-health incompatibility is a recently developed measure that assesses the worker’s evaluation of the extent to which work activities currently hamper the management of personal health needs [43]. We found that the more the work is perceived as all-absorbing, with little space left for personal health needs, the more likely the patient will stop work activity in the month after the cancer diagnosis. One-unit increases in the five-step scale of work-health incompatibility lead to an estimated increase in the work suspension rate of approximately 20%.
It worth mentioning that the effect sizes reported derived from the adjusted OR. Thus, these values should be less affected by overestimation or overoptimistic estimates that usually occur when variable selection is a key part of data analysis.
Limitations
The present study has some limitations that need to be considered to evaluate the findings with due caution.
As already stated, there is no standardized definition and measurement of work continuation after diagnosis, limiting the comparability between results’ studies. In the present study, the work continuation rate concerns the period immediately after the diagnosis. This procedure is similar to what was done by Molina Villaverde et al. [8] and Sharp and Timmons [12]. However, this last study seems to consider work suspension even a day off work after diagnosis, whereas we established a more stringent criterion. Again, in other studies [4, 7], the work continuation rate concerned the period during treatment, usually after one month from the diagnosis. Sjövall et al. [9] reported a further increase in the proportion of cancer patients with sick leave in the second month after cancer diagnosis, even if smaller than the first month. Future studies should meticulously investigate the effect of the actual beginning of the treatment on the rate of cancer patients with sick leave. To improve comparability between results’ studies, we suggest future studies should clearly report how work continuation was measured. This study proposed a clear definition and a simple measurement of work continuation in the month after cancer diagnosis.
The study design is cross-sectional and limits us to reporting associations between variables instead of causal relationships. Moreover, the measurement of the psychosocial factors and the work interruption during the month after diagnosis was during clinical treatment, which may have induced some bias in the patients’ answers. However, some studies have shown that memory bias is unlikely if the retrospective measurement relates to a traumatic event like cancer diagnosis [44].
The study is monocentric, and most of the patients in our sample had breast cancer; thus, our results are not representative of the whole cancer population. Nevertheless, to our knowledge, these are the first results about the topic from an Italian context. More studies need to investigate cancer patients’ work behavior immediately after the diagnosis of other types of cancer.
Finally, among the factors in the final model, the type of contract presented the lowest level of the Joint shrinkage factor (JSF = 0.63). Therefore, this factor may have been selected due to specificities of the present sample instead of a genuine effect in the population. However, this possibility is less likely as the same factor was statistically significant in other studies [8, 12].
Conclusion
Because the majority (68%) of the study patients continued working in the month after cancer diagnosis, this study questions the almost exclusive attention given in cancer literature to the working path: diagnosis ⟶ work absence ⟶ RTW. Our results suggested that this is even truer for self-employed and breast cancer patients that, in our final model, were around 18%and 20%less likely than other patients to suspend the work activity, respectively. It is crucial to determine the consequences of such behavior on the disease’s evolution, on patient’s compliance, and the long-term work conditions.
This research sheds some light on the factors associated with the decision to continue working after diagnosis (i.e., clinical, socioeconomic, and psychosocial factors). The increase in the work suspension rate of approximately 20%for a one-unit increase in the scale of work-health incompatibility indicates that the practical problems that can arise from the attempt to conciliate cancer and work or from the need not to lose workdays are relevant in this period. Hence, interventions aimed at facilitating the management of work for newly diagnosed cancer patients may influence their work-related decisions, even during the month after the diagnosis and not only during the RTW after a long work absence. Moreover, as also sustained by Nachreiner [13], physicians and psycho-oncologists can discuss with the patient the real implications of the chosen treatment considering the specific patient’s job demands in order to promote a true perception of work-health incompatibility, based on accurate information and not false expectation about the implication of cancer and treatment.
Conflict of interest
The authors declare that they have no conflict of interest.
Informed consent
Informed consent was obtained from all individual participants included in the study.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Footnotes
To consider OR as a direct estimate of the change in probability, like a Risk Ratio, systematically overestimates its effect size (except when the probability of the event is very low). The same OR generates different variations in probability for different probabilities of the reference group. Hence, it is not possible to equal one OR to only one variation in probability.[
].