[Proposal] Addressing Financial Hardship in Cancer through Policy and the Clinic

Addressing Financial Hardship in Cancer through Policy and the Clinic

Background: One half of American adults will experience a cancer diagnosis in their lifetime (American Cancer Society 2020) and nearly 80% of those patients will face serious financial burdens such as filing for bankruptcy or draining savings to pay for care (Azzani, Roslani et al. 2015). Financial burden is associated with reduced survival (Ramsey, Bansal et al. 2016), decreased quality of life (QOL; (Jones, Nguyen et al. 2020)) and worse medication adherence (Neugut, Subar et al. 2011).

Critical barriers to addressing financial hardship include a lack of rigorous systems to quantify and assess a patient’s financial burden (Zafar and Abernethy 2013). To address these barriers, we have developed a theoretical model of financial burden after cancer, displayed in Figure 1 (Jones, Henrikson et al. 2020). The model posits that costs of care and employment changes lead to greater financial burden and adversely affect outcomes such as QOL, morbidity and mortality (Park and Look 2018, Goulart, Unger et al. 2021). Even with health insurance or nationalized healthcare, patients still experience financial burden (Buttner, Konig et al. 2019). Employment changes are also a frequent cause of financial burden (Nekhlyudov, Walker et al. 2016).

To improve screening patients for financial burden, my conceptual model divides financial burden into: 1) material, tangible financial burden (objective); and 2) psychological financial burden (PFB; subjective). Material financial burden includes financial coping, actions patients take to afford care and basic needs, or financial consequences, events that patients experience because they are unable to cope with the costs of care and employment changes. PFB represents the thoughts and emotions people experience related to financial burden, including worry about affording future care and financial depression and rumination triggered by previous financial consequences and coping (Azzani, Roslani et al. 2015, Nekhlyudov, Walker et al. 2016). Financial worry has been largely ignored in both cancer care and research.

While financial hardship measures have been developed, these assessment tools have several drawbacks. Some have not been validated in cancer (Archuleta, Dale et al. 2013, Consumer Financial Protection Bureau 2015) while others conflate psychological and material burden and ignore important aspects of financial burden (de Souza, Yap et al. 2014). To address barriers to reducing financial hardship specifically in cancer, this proposed project will develop new survey item banks to assess financial burden after cancer. We will create item banks for worry about affording healthcare, financial coping, financial consequences, and financial rumination/depression using item response theory statistical models. The measures will be used to screen patients for financial burden and identify the specific types of financial burden that need to be addressed for each individual patient.

Methodology: My lab has created preliminary item banks to assess four aspects of financial burden as outlined in Figure 1 based on previous studies (Altice, Banegas et al. 2017, Gordon, Merollini et al. 2017): worry about affording healthcare; financial coping; financial consequences; and financial rumination/depression. We solicited expert feedback about the item banks. We will then survey people with cancer on Prolific using the item banks to create item response theory parameters for scoring the measures in research and clinical practice.

Analysis: Item Response Theory. Item response theory is a family of statistical models for analyzing and scoring survey and questionnaire data that accounts for differences in item severity (Samejima 1969, 2015). For example, worrying about paying for a one-time, $20 prescription may indicate a more severe burden than worry about paying for costly major surgery. Item response theory uses a logistic model to estimate the severity (also called the threshold parameter) of each item. This differs from traditional analyses that simply weight all items equally and instead, items are weighted by severity. Item response theory also accounts for the accuracy (also called the slope) for each item, weighting items with better accuracy higher than items with lower accuracy. For example, fatigue may reflect depression but is not as accurate an indicator of depression as depressed mood. Item response theory analyses use the accuracy and severity parameters to construct item characteristic curves (ICCs). In Figure 2, each curve represents a response category for one item. The x-axis is the level of the construct (worry in this case) and the y-axis is the probability of choosing the category. Using the severity and accuracy parameters, item response theory then models the error (or conversely, the reliability) and calculates the participant’s likely level of the construct. Item response theory overcomes the drawbacks of traditional measure development by using the severity and accuracy parameters to weight items in creating a continuous, interval-scale score of the construct. The use of item response theory also means future researchers and clinicians can use the item banks to create their own measures of financial burden, but scores will still be comparable even if different items are used.


To test the item response theory assumption of unidimensionality, we will run confirmatory factor analysis (CFA) models. Each item bank will be tested for whether a one factor model fits using the following indices: root mean square error of approximation (Browne and Ceduk 1993); comparative fit index; and root mean square residual.

Sample Size: Our sample of 400 will provide good power to detect poorly fitting CFA models (Chen 2007, Wolf, Harrington et al. 2013) and to properly estimate IRT parameters, based on published guidelines and simulation studies (PROMIS 2012, Şahin and Anıl 2017).

Study Costs: We are requesting US$2,600. Each participant will receive US$5 for completing the survey for a total of US$2,000 for participant payments and US$600 for service charges.

Open Reproducible Science: We plan to publish our findings in open access journals and include the item banks as appendices so that anyone who wants to use the questionnaires can do so freely and easily.

Preregistration: https://aspredicted.org/blind.php?x=rp3tv9