Return to work and cancer survivorship needs of breast cancer survivors: an observational prospective single-cohort study in Italy.

Introduction
Return to work is one of the important goals of the growing cohort of cancer survivors1, with 50% of new BC diagnoses occurring in working-age individuals in Italy2.
Work participation positively affects well-being and quality of life3; for cancer survivors, it facilitates their return to normalcy and social engagement, and it guarantees an income. Furthermore, work participation counteracts the decrease in productivity caused by cancer-related absenteeism, which carries costs for both companies and society4.
Although cancer survivors may reorganize their life priorities and place greater emphasis on their personal life in the first two years following a cancer diagnosis5, returning to work is still often perceived as an important step toward the restoration of normalcy6. Nevertheless, cancer survivors are 1.4 times more likely than healthy individuals to be unemployed7. Particularly for individuals with BC, sociodemographic and disease-related variables such as low education level, having no children, and having undergone a mastectomy and/or chemotherapy are associated with a higher likelihood of unemployment8–10. Notably, the duration of sick leave seems to be longer in the first year following diagnosis11, particularly for patients who have undergone chemotherapy or mastectomy plus radiotherapy12. Even though the side effects of treatments can persist for five years after BC diagnosis12, the likelihood of being employed is similar to that of the general population at the same time point13. Work-related variables affect the likelihood of work resumption. Psychologically and physically demanding jobs seem to hinder the return to work14. Among individuals with BC, difficulties may include limited arm mobility, fatigue, and psychological and cognitive issues3,15. Additionally, workplace relationships are crucial: supportive employers who support communication and accommodate work-related issues facilitate reintegration16, while negative attitudes hinder work participation14,17.
Thus, sociodemographic, disease-related, and work-related variables of an individual with BC may influence return to work at different time points throughout that individual’s cancer journey, increasing the risk of negative work outcomes.
In Italy, a few studies have investigated the return-to-work process: a population-based cross-sectional survey18, a longitudinal study on cancer survivors19, and a cross-sectional study on BC patients20. Although the latter found lower work ability among individuals living alone and those who had undergone a mastectomy, its cross-sectional design made it impossible to determine the causal relationship between potential determinants and employment outcomes or to identify any change that had occurred since diagnosis20. These gaps could be better filled through a prospective study.
Finally, treatment side effects impact the health status and quality of life of individuals with BC21,22, hindering their participation in other meaningful daily activities, not just employment15,23, which may not represent a priority at all stages of the disease trajectory. These non-work-related, unmet needs deserve greater attention and in-depth exploration. For example, patients may require support in physical, psychological, or social domains to improve health outcomes and regain normalcy. Addressing these broader needs also helps to contextualize the return-to-work process and underscores the complexity of cancer survivorship, thereby supporting a more patient-centered perspective. Nevertheless, cancer survivorship models do not always exhaustively address these needs, denoting a gap in the services provided24. In Italy, evidence on the needs of this population is still lacking.
Thus, the first aim of this study was to investigate return to work and the associated factors of unemployment, work difficulties, and long-term sick leave in employed individuals with BC; the second aim was to investigate the associated factors of cancer survivorship needs perceived by individuals with BC regardless of their employment status.

Methods

Study design and setting
This observational prospective single-cohort study, reported according to the STROBE guidelines (Online Resource 1, p. 1), was conducted at the local health authority Azienda USL–IRCCS di Reggio Emilia (Italy), which includes five provincial hospitals plus the main hospital and cancer research center, located in the city of Reggio Emilia; cancer treatment and research activities are concentrated at the main hospital. Part of the standard care of individuals with BC who have undergone sentinel or axillary lymph node removal, regardless of the hospital where breast surgery was performed, is an educational outpatient group session conducted by a physiotherapist, scheduled approximately two weeks after surgery. For this study, potential participants were approached at the end of the group session they attended; the study was explained, and the contact information of those who were interested was collected. The inclusion criteria were having had a diagnosis of BC, having attended the assigned outpatient group session, and being aged ≥ 18 years. Telephone contact took place approximately three days after the group session to answer any questions and to schedule the baseline appointment. Informed consent was obtained from all participants before data collection.
Recruitment was carried out between May 2022 and May 2023; follow-up lasted 12 months, until May 2024. During the 12-month follow-up period, participants were interviewed five times: T0 (baseline), T1 (one month after T0), T2 (three months after T0), T3 (six months after T0), T4 (12 months after T0).
The study was approved by the AVEN Ethics Committee (protocol number 2022/0035983) and was registered prospectively in ClinicalTrail.gov (ID NCT05309252, registration date 25/03/2022). All methods in the present study were performed in accordance with the relevant guidelines and regulations of the Declaration of Helsinki.

Outcomes
For the first aim, the outcomes collected for working-aged and for employed participants were:

return to work at baseline and at each follow-up time point. Return to work was defined as the occurrence of work resumption regardless of any accommodation in terms of work schedule or work tasks (i.e., individuals who returned to work with reduced work hours were classified as having returned to work). Accordingly, return to work was recorded as a dichotomous outcome (yes/no) at each follow-up time point.

work-related difficulties at baseline and at each follow-up time point (for participants who had returned to work), which were categorized using the International Classification of Functioning, Disability and Health Core Set for Vocational Rehabilitation-Onco (Core Set-VR-Onco)25, which was recently adapted to a population of cancer survivors26,27.

number of sick days at baseline and at each follow-up time point.

For the second aim, the outcome collected for participants regardless of age (working age or not) and employment status (employed or not) was:

cancer survivorship needs perceived at baseline and at each follow-up time point. Needs were collected through the Supportive Care Needs Survey–Short Form (SCNS-SF34)28,29.

Variables and assessment tools
At baseline, data were collected on sociodemographic (sex, age, education level, marital status, children, income range according to the tax bracket for total annual individual taxable income, residence, citizenship), work-related (type of worker, type of company, employment contract, working hours, night worker, shift worker, flexible work schedule, flexible work tasks, psychologically/physically demanding work, number of coworkers, work experience, disability certificate, Law 104 guaranteeing legal protection in the workplace according to Italian law in particular health conditions, i.e., paid monthly leave, work accommodations, return to work, work-related difficulties, and number of sick days), and disease-related information (type of tumor, staging, surgery, axillary lymph node dissection, chemotherapy, radiotherapy, hormone therapy, targeted therapy, rehabilitation), health-related quality of life, and treatment side effects (mood disturbances, sleep disorders, disability of the upper limb, social participation, and quality of life associated to specific symptoms and distress). At each follow-up time point, data was collected on work-related (return to work, work-related difficulties, number of sick days) and disease-related (type of treatment and rehabilitation) information, health-related quality of life, and treatment side effects.
For health-related quality of life, data were collected through the administration of the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C3030. For potential treatment side effects, data were collected through the administration of the Hospital Anxiety and Depression Scale (HADS)31, the Pittsburgh Sleep Quality Index (PSQI)32, the Disability of the Arm, Shoulder and Hand (DASH) questionnaire33, the Reintegration to Normal Living Index (RNLI)34,35, the Functional Assessment of Chronic Illness Therapy-Fatigue Scale (FACIT-Fatigue)36, the Functional Evaluation of Cancer Therapy-Cognitive Function (FACT-Cog)37, and the FACIT Comprehensive Score for Financial Toxicity (FACIT-COST)38.
A detailed description of the variables and the assessment tools used is reported in Online Resource 2 (p. 3).

Sample size
The convenience sample of approximately 110–120 individuals with BC was estimated considering the predefined eligibility criteria, local indicators of the Azienda USL–IRCCS di Reggio Emilia (229 patients who attended the outpatient group sessions in 2019), data available from the local Cancer Registry, and recruitment period. In a cross-sectional study conducted locally, the recruitment rate was 70%18. However, given the prospective nature of this study, we estimated a lower rate.

Statistical analysis
A descriptive analysis was performed to illustrate sociodemographic, disease-related, and work-related characteristics of all individuals with BC at baseline. Among the employed participants, the rates of those who returned to work, the absolute number of those who reported work-related difficulties, and the number of participants taking sick leave were calculated at each follow-up time point. All sick days were considered regardless of the perceived relationship to cancer. If sick leave was continuous, all seven days of the week were counted. The needs values for the domains of the SCNS-SF34 were calculated at each follow-up time point.
Outcomes were studied individually as dependent variables of generalized (multivariate) linear models for repeated measures, assigning the role of independent variables to potentially associated factors. Particularly, determinants of return to work, work difficulties, (log)days of non-work-attendance, and total needs at each time point were explored using penalized (LASSO) generalized linear mixed models (families were respectively binomial, binomial, gaussian, gaussian) with penalization parameter λ obtained by minimization of 5-fold cross-validation prediction error and a random intercept associated to the patient considered39. LASSO penalized GLMM models were adopted based on the number of patients involved (especially for primary outcome, n = 85), the type of primary outcome (dichotomous), and the number and types of covariates of interest to be included (several categorical variables implying a higher number of coefficients to be estimated) so as to limit the risk of overfitting and typical issues with classical (non-penalized) estimation. Covariates included in the estimate were age, education level, marital status, children, income range, type of tumor, surgery, axillary lymph node dissection, adjuvant chemotherapy, radiotherapy, hormone therapy, targeted therapy, rehabilitation, follow-up time point, type of worker, working hours, shift worker, night worker, working during weekend, flexible work schedule, psychologically demanding work, physically demanding work, number of coworkers, work experience, work accommodations, disability certificate, sick leave from diagnosis to surgery, sick leave after surgery, mood disturbances, sleep disorders, disability of the upper limb, social participation, and quality of life associated to specific symptoms and distress. All the results come from multivariate penalized analyses, where all the variables above were considered and collected at the same time points for all the patients. For better interpretability of estimates, only the coefficients not shrunk down to values suggesting no difference (e.g., 1 for OR) are reported so as to keep the tables readable. Finally, standard inference tools like confidence intervals and p-values are not available for this kind of analysis, and, as with penalized estimation, no difference/asymmetry in how independent variables were considered is present; all variables above were analyzed in each model, and no variable was considered fixed/to be necessarily present in final estimation.
Missingness was rather limited, and no ad hoc procedure was considered, for simplicity’s sake. Statistical analysis was conducted using R (version 4.3.0) with package glmmLasso version 1.6.339.

Results
Of the 442 individuals with BC invited to participate, 119 verbally agreed to participate, but only 113 of them signed the informed consent. Two participants did not show up at the baseline appointment, despite having signed the informed consent. Overall, the number of participants at each time point were: 111 at baseline, 103 at T1, 99 at T2, 94 at T3, and 90 at T4 (Online Resource 3, p. 5, Flow diagram of participants). A total of 83 participants (73.5%) attended all follow-up appointments.
As regards dropouts, there were five at T1, five at T2, and three at T3. Missing data also occurred: three participants missed the T1 appointment but came to the subsequent appointments; two missed T2 but came to the subsequent appointments; three missed T3 but came to T4, and seven participants missed this final follow-up appointment.
Table 1 reports the sociodemographic and disease-related characteristics of the study cohort and the work-related information of the 85 participants (76.6%) who were employed at baseline. The mean age of all the participants was 53.5 years (± 8.9), and most had a high school diploma or a university degree (74.8%). Conservative surgery was the most frequent surgery modality (61.3%); slightly more than 15% of the participants had undergone axillary lymph node dissection. Among the participants who were working at baseline, most were employees (89.4%), in the private sector (65.9%), with a permanent work contract (84.7%). More than 50% reported a psychologically demanding job.
Figure 1 describes the changes in the return to work of the 85 participants; these changes were recorded at baseline for all 85, for 79 at T1, 76 at T2, and 69 at T3 and at T4. The descriptive analysis suggested that, between baseline and T4, there was a progressive increase in the percentage of individuals with BC who had returned to work; at T4, 7.2% of participants had not yet returned to work. Online Resource 4 (p. 6) represents the changes in the return to work of the 85 participants categorized as employees or self-employed.

Online Resource 5 (p. 7) represents the work-related difficulties perceived during the return-to-work process. According to the descriptive analysis, the number of difficulties appeared to increase progressively from baseline to T4. The highest absolute numbers of problems in Body Functions (53 problems) and Environmental Factors (25 facilitators, 20 barriers) were reported at T3, while problems in Activity and Participation were more frequently reported at T4 (53 problems). At T3, the main problems concerned fatigue (category b455), sensation of pain (b280), lifting and carrying objects (d430), undertaking complex tasks (d220), and hand and arm use (d445). The main facilitators were receiving support from coworkers and employer (e325, e330); the main barrier was not receiving support from the employer (e330). At T4, fatigue (b455), sensation of pain (b280), and lifting and carrying objects (d430) were still reported as the main problems; complex relationships (category d720) were also reported among problems at this time point. Again, receiving support from coworkers and employer (e325, e330) were reported among facilitators at T4, while not receiving support from employer (e330) represented a barrier. A further barrier reported at T4 was prescribed drugs (e1101).
Online Resource 6 (p. 8) reports the number of participants employed at baseline and data regarding their sick leave calculated at baseline (from surgery) and for each subsequent time point. The number of participants who reported taking sick days decreased from baseline to the final follow-up appointment.
Figure 2 shows the extent of needs for each domain of the SCNS-SF34 for all individuals with BC (111 participants at baseline). Overall, needs in all domains seemed to decrease between T0 and T1. At T2, needs in all domains increased, except for those in the psychological domain. At T3, needs in all domains decreased, except for those in the sexuality domain.

Table 2 reports the odds ratio (OR) for determinants of return to work, having difficulty, sick days, and total needs.
Being unmarried (OR 1.18) or divorced/widowed (OR 1.24) seemed to be associated with a higher likelihood of returning to work, as was a having higher income (OR 1.39). Having one child or more seemed to predict a lower rate of return to work (OR 0.81). Lobular carcinoma (OR 0.86), undergoing lymph node dissection (OR 0.66), and having targeted therapy (OR 0.78) seemed to be associated with less likelihood of returning to work. Self-employment (OR 1.42), having a flexible work schedule (OR 1.18), and having a disability certificate (OR 1.08) seemed to have higher return-to-work rates. Having taken sick days between diagnosis and surgery (OR 0.74), being a shift worker (OR 0.85), having a higher number of coworkers (OR 0.96 for 50–249 workers, OR 0.98 for ≥ 250 workers), and having a physically demanding job (OR 0.64) seemed to be associated with a lower return-to-work rate. Greater disability of the upper limbs (DASH OR 0.98) seemed negatively correlated to return to work. Follow-up time points seemed positively associated with return to work (OR 1.61–6.54).
A better quality of life (FACIT-Fatigue OR 0.93; FACT CogQoL OR 0.79; FACIT-COST 0.98) seemed positively correlated to a diminished likelihood of having work difficulty.
Having taken sick days between diagnosis and surgery appeared positively associated with taking sick leave after surgery (OR 13.78).
More severe symptoms (HADS-Anxiety OR 0.66 and HADS-Depression OR 0.78), sleep quality (PSQI OR 0.40), and disability of the upper limbs (DASH OR 0.13) seemed associated with higher total needs for both employed individuals with BC and all individuals with BC.

Discussion
This study highlights associations between sociodemographic, work-related, and disease-related information, health-related quality of life, and treatment side effects with return to work, work difficulties, and sick leave among employed individuals with BC. Furthermore, it reports on the associations between potential predictors and the unmet cancer survivorship needs of adult individuals with BC.
Almost all employed participants in this study returned to work 12 months after surgery (91.3%), a rate higher than that reported in China (68.2%)40 and in France (42.2%)41. Among employed patients, factors such as having children, having undergone lymph node dissection, a diagnosis of lobular carcinoma, having received targeted therapy, having taken sick leave between diagnosis and surgery, being a shift worker, working in a large company, performing a physically demanding job, and experiencing greater upper limb disability seemed negatively associated with return to work. Having children was associated with a lower likelihood of returning to work, probably because family responsibilities and health concerns were prioritized over paid work42. Nevertheless, other studies did not confirm this association8,9. Lymph node dissection was associated with a reduced likelihood of returning to work, though the evidence remains inconsistent9,43. The influence of BC subtype remains largely unexplored, limiting comparability. Targeted therapy seemed to predict a lower likelihood of returning to work, although evidence from France suggests that this treatment modality does not affect work participation among individuals with localized BC44, although it does seem associated with sick leave three years after diagnosis13. Furthermore, three out of the four individuals with BC in our cohort who did not return to work underwent targeted therapy.
Shift work seemed negatively associated with the likelihood of returning to work. This finding is still poorly investigated in the literature17 despite its relevance to planning accommodations of work schedule. Employees in larger companies also seemed to be less prone to going back to work, possibly reflecting the extended sick leave policies in the public sector. Conversely, smaller companies may provide greater social supportive environments45 but fewer opportunities for workplace accommodations. The association between having a physically demanding job and a lower return-to-work rate was in line with previous findings from both cancer survivors in the same context17 and BC survivors46. This kind of work is generally more difficult to accommodate as it consists of predetermined work schedules, standardized methods of execution, a high degree of coordination with coworkers, and reduced decision latitude. Moreover, individuals with BC experience treatment-related side effects of the upper limbs, which further limits their ability to perform strenuous tasks. We found that greater upper limb disability was negatively associated with return to work, in line with previous findings retrieved one year after surgery3 and among patients who had undergone axillary lymph node dissection47. Follow-up time points were positively associated with return to work, with greater time since diagnosis increasing the likelihood of work resumption. This result, consistent with previous evidence48, should be interpreted alongside the descriptive finding that most participants reported work difficulties at T3 and at T4, when 73.9% and 91.3% of participants, respectively, had returned to work. While difficulties did not prevent work participation, they probably made the whole process more complicated as problems may arise afterward, possibly due to declining workplace social support or long-term treatment side effects. Hormone therapy seemed associated with work impairment but not with unemployment49. From a rehabilitation perspective, this finding could inform the timing of intervention50 to ease reintegration and prevent long-term adverse outcomes51.
Moreover, better quality of life associated with fatigue, cognition, and financial toxicity seemed to predict fewer work difficulties, consistently with the evidence showing that BC survivors who returned to work two years post-diagnosis reported less fatigue and cognitive impairment52.
Sick leave decreased between T0 and T4. In Finland, the highest percentage of sick leave days (68% during treatment) occurred among patients undergoing surgery, chemotherapy, and radiotherapy53. Similarly, chemotherapy and/or radiotherapy predicted delayed return to work in China40, whereas high income, not doing a physical job, and better physical functioning were associated with faster reintegration40. Our finding illustrates that taking sick days between diagnosis and surgery appeared negatively associated with return to work and positively associated with taking sick leave after surgery. Sick leave prior to surgery, rarely investigated in the literature, may reflect neoadjuvant treatment side effects or higher psychological distress at diagnosis54.
Cancer survivorship needs decreased between T0 and T1 but increased in most domains between T1 and T2; compared to T3, needs in psychological and physical and daily living domains increased. Moreover, psychological needs had the highest priority at each follow-up time point. Our recent systematic review showed that at one year or more after diagnosis, an increase in the trend of needs in all domains except for patient care and support needs was registered55. Greater anxiety, depression, sleep disturbances, and upper limb disability were linked to increased perceived needs. The odds ratios for employed individuals with BC and the total population were superimposable. Mental and physical health status and comorbidities seemed to predict unmet needs, consistently with a prior systematic review55.
One strength of this study is the low risk of recall bias due to the prospective nature of the study. However, it may have affected some factors (e.g., number of sick days), especially during the 6-month period between T3 and T4. Another possible strength is that there was only one rater. However, this could have led participants to providing answers that would make themselves be viewed more favorably in the eyes of the interviewer, introducing the social desirability bias56. Finally, the number of participants reached the amount hypothesized.
One limitation is that during the 12 months of follow-up, a dropout rate of 19% was recorded. Unfortunately, this may have been due to the demanding nature of the commitment, both in terms of the number of follow-up appointments and the number of administered questionnaires. To reduce the risk of drop out, participants were given the option of scheduling an appointment in person at the hospital closest to their home or else remotely. Another limitation concerns the eligibility criteria, as only individuals with BC who had undergone breast surgery with sentinel lymph node removal were included. In addition, the recruitment procedure may have introduced a selection bias. The study was presented following an outpatient group session all individuals with BC had been invited to attend after surgery, but participation was voluntary. This fact limits the generalizability of the results to individuals with BC with different clinical characteristics or those who did not attend the group session (e.g., those who had transport issues, more fragile individuals). From a research perspective, future studies should address this potential selection bias by adding alternative recruitment settings (e.g., first oncology consultation) in order to include a broader population. Moreover, although return to work is dynamic, in this study it was operationalized as a binary outcome assessed at multiple time points over 12 months post-surgery. Therefore, future studies should consider more nuanced outcome measures such as partial return to work or return to work with workplace accommodations and extend the follow-up period to better capture the evolution of return to work. Finally, future research should focus on formally testing the hypotheses generated by this exploratory study.
From clinical and rehabilitation perspectives, this prospective study allows early recognition of vulnerable individuals with BC, thereby facilitating the monitoring of their condition over time, supporting their return-to-work process, and addressing their needs through tailored solutions in line with the best evidence. The most effective interventions aimed at facilitating the reintegration of cancer survivors into the workplace appear to be multidisciplinary in nature57, relying on the coordinated involvement of different disciplines to address a broad range of work-related difficulties, for example, by including vocational counselling, patient education, and physical exercise. These interventions may be specifically targeted to individuals with BC58,59, whereas others are designed for cancer survivors regardless of their diagnosis50,60; the results of this prospective study could inform both types of intervention. These findings suggest that employment is shaped by key determinants, whose early identification in inpatient settings may help mitigate adverse outcomes. Particularly, our findings support the early identification of individuals with BC at higher risk of adverse work outcomes and those who could benefit more from a tailored monitoring of the work condition, which could start immediately after diagnosis or surgery. This underscores that healthcare professionals who interact with patients at risk must be adequately trained to identify and to proceed with a timely referral to the intervention. However, the timing of providing an intervention may vary; based on the descriptive data collected, interventions may be more appropriately concentrated several months following surgery or even after the actual return to work. Furthermore, the results of this study provide valuable information that can contribute to furthering the support offered to patients in the province of Reggio Emilia through the project UNAMANO50. As the intervention is intended for all cancer survivors, future research should verify the feasibility of the proposal in its current form and whether it adequately addresses the needs of individuals with BC.
Finally, survivorship needs extend beyond work-related aspects to encompass daily life. Because these needs evolve over the first year after surgery, they should be assessed according to the patient’s survivorship phase. This suggests that survivorship care interventions for individuals with BC should account for the variability in needs over time, which has been previously reported55, and patient-centered approaches to support the implementation of tailored survivorship care models should be adopted.

Supplementary Information
Below is the link to the electronic supplementary material.