Much work to do about measuring work

Abstract

Work ability is a critical economic and well-being indicator in cancer care. Yet, work ability is understudied in clinical trials and observational research and is often undocumented in medical records. Despite agreement on the importance of work from well-being, health insurance, and financial perspectives, standardized approaches for collecting, measuring, and analyzing work outcomes are lacking in the health-care setting. The necessary components for closing the gap in patient and caregiver employment research in health-care settings involve a common set of measures, including those that replace or translate generic measures of mental and physical functioning into work outcomes in observational and clinical trial research, standardized approaches to data collection and documentation, and the use of longitudinal data to understand the consequences of reduced work ability over time. We present a conceptual framework for the inclusion of work ability in outcomes research. We cover constructs for employment and work ability measurement that can be adopted in research, recorded as patient-level data, and used to guide treatment decisions. The inclusion of return to work and hours worked, productivity, and ability to perform in a similar job can support conversations that guide treatment decisions and minimize economic consequences. Our hope is that by considering impact on work ability, improved treatments will be developed, health inequities reduced, and resources directed toward aiding patients and their caregivers in balancing work and health demands.

In 2022, there were more than 18 million cancer survivors living in the United States, and there is expected to be 1.9 million new cancer diagnoses in 2023 (1). As the survival rates of cancer increase (2), many people will navigate treatment and side effects while trying to preserve aspects of their normal life. Some cancer survivors will remain employed throughout their diagnosis and treatment, whereas others may adjust their hours, take a leave of absence, or leave their job entirely. Informal caregivers also often incur work loss and financial strain when a loved one is diagnosed with cancer (3). An understanding of how treatment will affect work ability is especially crucial for the 40% of cancer patients diagnosed while working age (4). Of note, screening guidelines for several cancers (eg, breast, cervical, colorectal, lung) that start prior to the traditional retirement age of 65 years have led to cancer detection that may have otherwise gone undetected until after retirement (5).

Certain treatment options may be more detrimental to employment depending on treatment side effects and job requirements. Among women receiving breast cancer treatment, axillary lymph node dissection had a statistically significant reduction in work ability in the early months following diagnosis, and chemotherapy and locoregional radiation therapy were associated with long-term reductions in work ability (6). Another study reported that breast cancer survivors who received neoadjuvant chemotherapy were 1.4 times more likely to be unemployed 4 years after diagnosis relative to cancer survivors who did not receive chemotherapy (7). Common side effects from treatments that can impact work ability are fatigue, nausea, neuropathy, insomnia, and pain (8). Knowing the short- and long-term differences in treatment options can help guide patient decision making when choosing a treatment plan.

With the continued rise in new cancer drug prices (9), employment is a critical means of financial security and health-care access. This is especially true in the United States where there is no national health insurance program for adults younger than age 65 years or federal paid sick leave mandates. The median annual price for new oncology drugs has grown substantially in the past decade, up from almost $65 000 in 2012 to $260 000 in 2022 (9). The loss of employment and employment-related health insurance can contribute to or cause financial toxicity, which has been widely documented (10). Financial toxicity is associated with higher mortality, depression, fatigue, and a reduced quality of life (11,12). Therefore, understanding, measuring, and documenting cancer and its treatment impact on employment are all the more important to patients and their caregivers. Those who are employed are more likely to have health insurance for themselves and their dependents as well as the financial capacity to pay for health-care services relative to similarly aged patients and caregivers who are not part of the workforce (13). Furthermore, adults who can work while receiving medical care are more likely to retain their employment in the future (14).

Work and work ability are also an important part of self-worth, independent functioning, and financial well-being. The benefits of working include increased social connections, more fulfillment and purpose in life, fewer adverse mental health events such as anxiety and depression, and maintaining cognitive abilities (15-18). For cancer survivors and their caregivers, continued employment or returning to work can promote a sense of control and normalcy during an arduous time (19).

The failure to consider the consequences of reduced work, such as insurance loss, wage reduction, and adverse mental health outcomes, leaves many cancer survivors and their caregivers without complete information to make treatment decisions that are concordant with their preferences and goals. Some are unable to adequately plan for time away from work, request accommodations, or prepare themselves for altered work roles. Collecting data on work ability will help facilitate a conversation between patients, caregivers, and clinicians to better set expectations for treatment outcomes and promote shared decision making. We offer guidance on how researchers interested in the economic consequences of cancer might develop and execute studies that examine work-related outcomes in the health-care context.

Conceptual framework

Figure 1 presents a conceptual framework for employment outcomes in health-care research. Current measures are focused primarily on generic or disease-specific measures of quality of life or domains for physical and mental functioning. In addition, there are disease-specific symptoms that are often measured. The MD Anderson Symptom Inventory is an example of a symptom measurement scale (20) that focuses on common symptoms experienced by patients who are undergoing treatment for cancer. Although these instruments may be correlated with work ability, they do not directly measure work or work outcomes. In addition, these instruments are deployed sporadically in observational and clinical trial research and usually given at the start and end of treatment, but patients are not routinely followed long-term, and caregivers are not assessed at all.

Figure 1.

Conceptual framework of patient-centered employment outcomes.

In Figure 1, we propose the inclusion of measures that are directly related to work ability and outcomes in addition to measures of functioning and symptom severity. We suggest wide and consistent deployment of these measures in health outcomes studies with gender stratification and healthy participants as appropriate, followed by documentation in electronic health records (EHRs). Assessment of workflows and processes to determine the best fit in the clinical setting and optimization of EHR designs will be needed. We also advocate for immediate, intermediate, and long-term follow-up before, during, and after treatment. These measures may also be extended to caregivers.

Work ability measurement

Despite their importance, work outcomes are understudied in the health-care context and specifically in outcomes research. The lack of standardized instruments to measure work and the consequences of work loss hinder patient-centered economic outcomes research. Instead, researchers tend to assess related and overlapping constructs such as quality-of-life measures (eg, RAND-36), disability status, and symptoms (eg, fatigue, depression) as well as disease severity and treatment intensity (21-26). Physical and mental functioning and symptom severity currently used in clinical research settings are not directly translatable into whether patients and their caregivers can perform their jobs. Although these measures may be predictive of work ability, we argue that work ability is a separate outcome that is not readily informed by general measures of physical and mental functioning or symptoms. Research is needed to understand if diminished functioning in some domains is predictive of work ability for patients and their caregivers. Nevertheless, we argue that a common and widely used measure of work ability is essential to understand employment and economic consequences for patients and their caregivers.

A more thorough measure of diminished work outcomes is necessary when assessing employment. As an example, many researchers may be tempted to measure employment as a binary indicator (yes, no) without considering the critical differences between full- and part-time workers (27, 28). Measuring weekly hours worked is a better measure of work ability and whether access to health insurance, that is often offered only to full-time workers, will be affected. In addition to these objective measures of work (eg, employed yes or no, weekly hours worked), more subjective productivity measurement is also an important component. We list 3 instruments as examples that address these dimensions in Supplementary Table 1 (available online). Job type, perceived employer accommodations, and a supportive work environment are other important factors associated with returning to work that can mediate health impacts on work ability (29). These areas are ripe for research into work-related outcomes in the clinical setting.

Patients, caregivers, and care team input is vital to formulate research questions, survey development, and outcome measurement that are most relevant to patients and their caregivers for maintaining their work status. Employers may also have a role. They may have experiences that can inform measures that are helpful for evaluating employees’ ability to work and type of accommodations needed (30). Collectively, these constituents can provide multiple perspectives on survey content, item wording, and survey length.

Study design considerations

Longitudinal and primary data collection are often needed to study work outcomes. Few secondary datasets have sufficient working-aged people with cancer to allow for detailed analyses. Even surveys such as the Medical Expenditure Panel Survey that have oversampled people with specific diseases (eg, Medical Expenditure Panel Survey Experiences with Cancer supplement) (31) have insufficient large samples of working-aged people to allow for stratification by cancer site and treatments. Treatments, when available, are generally described in broad terms (ie, chemotherapy) without stage information and specific treatment regimens.

Unlike survey data, most cancer registries in the United States have information on incidence, treatment, and survival but no information on work outcomes. Data in these registries may be 2 or more years old and can also be incomplete (32). Therefore, if patients are drawn from these sources for further primary data collection, they may suffer recall bias when recounting their treatment-related experiences that led to changes in work outcomes (33). Another limitation to registry data is that the sample is constructed of only those diagnosed with a specific disease, prohibiting comparisons to healthy individuals. Linking registry data to additional datasets allows for a more rigorous study design to incorporate healthy individuals in a control group.

For research questions that investigate the widespread impact of cancer on work ability, a population-based sample with healthyparticipants is ideal. Without healthy controls, the effects of illness cannot be disentangled from the effects of factors such as aging or changes in the labor market from which the sample was drawn (34). However, healthy controls can be expensive and difficult to recruit for study. Some studies have resorted to patients’ peers as controls (35), but these peers may be influenced by their knowledge of the patients’ experiences. Other studies have used healthy controls from population-based employment surveys drawn from the same geographic region as patients (23). Examples of population-based employment surveys include the Current Population Survey (36) and Panel Survey of Income Dynamics (37), and for older adults, surveys such as the Health and Retirement Study are good options (38). Employment and hours worked questions from these surveys are validated, have been used on thousands of individuals, and avoid pitfalls that investigators new to studying employment outcomes may encounter. These questions can be useful as core set questions on which to build disease- or treatment-specific questions. See the Supplementary Materials (available online) for how the Current Population Survey questions on employment and hours worked questions can be tailored to study a health condition.

A study design that incorporates primary and secondary data is efficient and can be enhanced by using the same questions in the same order in the primary data collection as used in the employment survey for healthy controls. Survey timing can also be aligned to account for temporal changes in the labor market. Because health information may be unreported in employment surveys, a few controls will be misclassified as healthy when in fact they have the condition of interest. It is important to conduct assessments to determine how closely the treatment and control samples match (39).

A challenging aspect of studying work outcomes is the need to stratify the sample by sex gender (40-42). Women in general, relative to men, work fewer hours and often have jobs that do not offer benefits such as retiree health insurance. In addition, men and women are unevenly distributed and compensated across occupations and industries. As a result, men and women may experience distinct responses to a cancer diagnosis and treatment. Failure to stratify the sample may result in attenuated responses for one group who may be more adversely affected. Gender stratification will require separate samples of men and women that can increase the needed sample size, depending on how the effect size varies by gender.

Data infrastructure

The fragmented health-care system makes it difficult to align treatment, whether through medical records, surveys, or surveillance registries, with work outcomes. Opportunities to improve data collection include consistent methods of collecting, reporting, and standardizing work ability across EHRs along with the incorporation of consequences that may result from changes in work ability. Workflows and processes for incorporating data collection and outcomes in EPIC, Cerner, or other EHR systems need to be better understood. With standardized data collection and centralized EHRs, structured data could inform how to study and, subsequently, improve work outcomes. Key data elements include job type, role and responsibilities, typical hours worked, and critical tasks where physical and mental function must be preserved. As an intermediate step, work ability and employment outcomes data could be collected as part of clinical and pragmatic trials. As many treatments are taken for the remaining lifespan, the absence of these data at the time of drug development and during long-term follow-up is a critical missed opportunity that needs remedied.

Challenges

Potential challenges to collecting and analyzing work ability and outcomes data exist. The need to standardize data collection instruments, validate patient reports, and longitudinally follow affected individuals takes considerable effort and resources. The collection of additional data may also be a burden to busy clinical staff. EHR documentation duties are commonly associated with provider burnout and frustration (43). Nonetheless, promising solutions such as automation and user-centered EHR module designs (44), and hiring medical scribes (45), can be leveraged to reduce reporting burden. In addition, there may be privacy concerns that data could be shared with employers. Widespread knowledge that a cancer diagnosis or other health conditions and treatments interfere with work performance may raise fears of discrimination or other legal challenges in the workforce. However, without this knowledge, adverse outcomes cannot be mitigated, and personalized treatment decisions cannot be prospectively made.

A better understanding of cancer and its treatment impact on employment allows for clinicians and patients to reduce work-related burdens. Possible solutions include choosing treatments with fewer side effects that may interfere with work; managing side effects more proactively; providing referrals to occupational medicine, physical therapy, and occupational therapy; and scheduling treatments at a time that has less interference with work. Studies have also shown that planning and requesting work accommodations can reduce work impact (46). Widespread engagement is critical to ensure that the research is informed by patient and caregiver needs and the steps are taken to educate, inform, and ensure privacy and protections for patients and caregivers. Research is needed iteratively to update outcomes as new treatment approaches are developed and the workforce environment evolves to improve work-related outcomes.

Conclusions

Our recommendations for studying work and work ability are summarized in Table 1 where we offer an approach for collecting this important aspect of patient-centered economic outcomes. We start by including constituents who have valuable input into the measurement and application of work outcomes. We then highlight the need for adoption of validated and reliable instruments and suggest several settings for deployment. We summarize study design attributes including the use of healthy controlparticipants, sex stratification, and the need for longitudinal outcomes.

Table 1.

Recommended processes for studying employment outcomes among patients and caregivers

Constituent engagement	Convene shareholders including patients, caregivers, employers, and care team to seek input for information most valued and how to measure.
Data collection	Deploy validated and reliable methods of collecting, reporting, and standardizing employment data across electronic health record systems along with the incorporation of patient- and caregiver-reported outcomes. Collect employment outcome data as part of clinical and pragmatic trials. Data collection efforts should include job type, role and responsibilities, typical hours worked and any subsequent changes, and critical tasks where physical and mental function must be preserved.
Study design	When analyzing the impact of a disease on work ability using secondary datasets, compare the cohort diagnosed with cancer to a set of healthy participants. Stratify by sex when assessing work outcomes. Assess both short- and long-term work outcomes.

As advances in cancer drugs and therapies transform this life-threatening condition into a more chronic condition, patients who are employed when diagnosed may continue working, including unpaid work (eg, childcare, volunteer responsibilities), during treatment or return to work following treatment. Likewise, informal caregivers experience similar strains on work ability and economic consequences (47) as they bear much of the responsibility in managing the patient’s care and should be considered in patient-centered outcomes research. Reasons for work continuation include psychological well-being, financial stability, and employment-based health insurance coverage, without which treatment and ongoing care may be unaffordable. Therefore, the methods used to estimate work ability, productivity loss, and other economic consequences attributable to this illness require careful consideration so that reliable findings can be used to shape health-care decisions and evaluate treatment.

Data availability

No new data were generated or analyzed in this commentary.

Author contributions

Cathy Bradley, MPA, PhD (Conceptualization; Writing—original draft; Writing—review & editing), Sara Kitchen, MPH (Writing—original draft; Writing—review & editing), and Kelsey M. Owsley, MPH, PhD (Conceptualization; Writing—original draft; Writing—review & editing)

Funding

Bradley’s work was supported by National Cancer Institute grant to the University of Colorado Cancer Center Core Grant (P30CA046934). The authors acknowledge funding from the Office of the Assistant Secretary for Planning and Evaluation to support travel to the Symposium on Building Data Capacity to Study Economic Outcomes for Patient-Centered Outcomes Research that was held on December 5, 2022, and provided an opportunity for the authors to present their work and receive feedback from attendees.

Conflicts of interest

CB, who is a JNCI Associate Editor and co-author on this paper, was not involved in the editorial review or decision to publish the manuscript. The authors have no conflicts of interest.